Efficiency and Distributive Goals in the EU Tech Regulatory Strategy

On 20 March 2025, The New York Times reported the story of Joseph Coates, who owes his life to artificial intelligence (AI). Mr Coates was affected by a rare blood disease and was too sick to receive a stem cell transplant. An AI-powered model suggested an untested formula combining immunotherapy, chemotherapy and steroids, and Mr Coates is now in remission (Morgan, 2025). This story epitomises the incommensurable potential of frontier technologies to revolutionise people’s life expectations.

Advanced technologies may be a lifeline to the economy, too. A growing body of literature supports the existence of large productivity gains resulting from AI adoption.¹ For example, Brynjolfsson et al. (2025) find that AI assistance increases worker productivity by 15% on average.²

This explains Europe’s current fear of missing out. In the face of apparent significant gains that advanced technologies bring about, several indicators suggest that the European Union is lagging in technological development and uptake. For example, in 2024, European AI startups raised $12.5 billion compared to $81.4 billion raised by US-based AI startups (Saharova et al., 2025). In the same year, only 13.5% of European companies with ten or more employees used AI-powered technologies to conduct their business (Eurostat, 2025).

However, during the European Commission President Ursula von der Leyen’s first mandate (2019-2024), regulation, rather than direct industrial support, has come to embody the EU priorities for the digital economy. In recent years, the EU has produced a considerable amount of new regulations directly applicable to digital markets. To mention a few: the Digital Markets Act (DMA, Regulation (EU) 2022/1925), the Digital Services Act (DSA, Regulation (EU) 2022/2065), the Artificial Intelligence Act (AIA, Regulation (EU) 2024/1689). And the list goes on and on.³

In the EU digital policy debate, two questions are thus gaining increasing traction: does the EU technology regulation bear at least some responsibility for the technological gap between the EU and other comparable economies? If so, should the EU embark on a deregulatory agenda to attempt to fill that gap?

The Commission’s present approach, featuring a structural recalibration from regulation to competitiveness, seems to be based on an affirmative answer to both questions.⁴ For example, after proposing simplifying rules on sustainability reporting and investment in the EU with a first and a second Omnibus package,⁵ the Commission has proposed to ease its privacy regulation for small and medium-sized enterprises (European Commission, 2025b).

Little has been done, however, to lay out a sound economic framework in which those questions can be meaningfully assessed. In this paper, I aim to address that deficiency, suggesting a way to coherently evaluate digital regulation and its effects on EU digital markets.

The methodology I propose is based on the definition of two distinct categories of goals that can be pursued through digital regulation: efficiency-oriented goals and distributive goals. With the former, regulation contributes to expanding the amount of value that the economy produces. With the latter, instead, regulation contributes to defining how that value is distributed within society.⁶ A regulatory norm may be motivated by efficiency and distributive goals, so explicitly separating them is often not obvious. For example, a regulation intended to open markets to competition can be motivated by efficiency goals (competition increases productivity) and distributive goals (competition transfers surplus from sellers to buyers).

Both categories of goals are very relevant in the context of digital markets. Digital services, for example, tend to benefit from scale and network externalities, and they thus have a natural tendency to concentrate (Mariniello, 2022). Concentrated markets allocate resources less efficiently than competitive markets and beg for efficiency-oriented regulatory intervention. Similarly, distributive goals also matter significantly in digital markets, given technology’s deep societal implications. For example, even seemingly harmless AI features may ultimately entail all-encompassing cultural effects, and these can be mitigated through distributive regulation.

In this paper, I anchor these two categories of goals to examples from the EU digital regulatory playbook. The main policy suggestion is to deploy a regulatory strategy hinging on a clear layout and analysis of efficiency and distributive goals. Whenever possible, efficiency can be improved with simplifying actions (e.g. refining or repealing current laws) or with complementary actions (e.g. deploying new regulations to make the current ones more effective). When, instead, regulation implements political choices that entail unavoidable distributive trade-offs, this should be stated explicitly.

This is important because the predominant policy narrative around European competitiveness does not explicitly recognise the contrast between efficiency and value distribution. The distributive dimension rarely emerges as a relevant element explaining the differences in the performance of countries’ digital economies. For example, in his report on The Future of European Competitiveness, Mario Draghi (2024) articulates several factors that explain why the EU lags behind the US and China on key digital economy indicators. However, none of those factors refer to the EU’s distributive choices, despite inevitably affecting digital business performance. For example, the EU has high privacy protection standards. If European data companies face restrictions on data use, they would naturally be disadvantaged compared to Chinese companies that are not entangled in the same constraints.⁷

Recognising the distinct existence of efficiency and distributive goals leads to two important insights.

First, the EU can improve its performance on efficiency grounds and reduce the perceived gap with international competitors. However, to the extent that their digital economies are conditioned by different underlying structural social preferences, the European, American and Chinese digital economies are not comparable.

Second, a corollary of the first point is that observing that the European economy does not match the performance of American and Chinese digital economies should not, as such, be concerning. It is certainly worrisome when that happens by accident due to an inefficient design or enforcement of the EU regulatory framework. It may not be the case, when the alleged “underperformance” is underpinned by a deliberate political choice based on citizens’ democratically expressed preferences (voters may be willing to sacrifice competitiveness to support other conflicting goals, for example).

The primary advantage of spelling out the underlying goals of regulatory choices is to identify clearly where deregulation may be helpful and where, instead, it conceals backtracking on political commitments.

Two caveats are due. First, the focus of this paper is EU regulation. However, it should not be forgotten that a significant part of European companies’ regulatory burden originates in their countries’ national legislation.⁸ Second, I venture only marginally into speculating on the relative significance of EU digital regulation as compared to other factors potentially hampering growth. There are indications that excessive EU regulation is not the primary reason why the EU is lagging in the tech space.⁹ However, the scope of this paper is limited to considering the relative effect of EU regulation, taking all other potential factors that may limit growth as given.

The next section introduces the concept of efficiency-oriented goals and analyses their application in the EU digital market context. The subsequent section discusses distributive goals and distributive choices made by the EU legislator. The paper concludes with policy recommendations.

Efficiency-oriented goals

Efficiency-oriented goals guide regulation towards maximising the absolute value that an economy can produce. Value is an all-encompassing concept that includes measurable quantities (such as companies’ profits or total output), but also hardly objectively quantifiable benefits, such as product quality and variety.

Theoretically, “frictionless” markets maximise value autonomously: producers’ self-interested decisions and consumers’ preferences steer the economy to produce and consume according to its possibilities.¹⁰ In practice, however, no market is frictionless. To an extent, all existing markets fail to deliver their full potential. That happens, for example, because they are not competitive enough¹¹ or because there is a mismatch between what companies consider valuable and what is beneficial for the economy as a whole.¹²

Regulation can maximise efficiency by removing the sources of market failures that prevent digital markets from generating their potential value.

By far, the primary way in which EU regulation of the digital space pursues efficiency goals is through an expansion of digital markets from national to European. It does so by homogenising regulation over the European territory: it fixes common supranational rules that supersede national approaches. In other words, it contributes to creating a Digital Single Market (DSM).

To see this, note that the production and provision of digital products significantly benefit from scale. Thus, the value generated in a hypothetically completed DSM would be greater than the value obtained by summing up the value generated within each EU member state, if considered in isolation. A completed DSM minimises uncertainty for business (when facing 27 national frameworks, future regulatory risk is less predictable)¹³ and increases market competition.

Generally speaking, provided that EU regulation fills gaps that could be filled by national regulators, it likely produces efficiency effects. For example, the Platform Work Directive (PWD, Directive 2024/2831) introduces a methodology to classify the employment status of individuals providing services through online platforms, such as Uber or Deliveroo. Its main advantage consists in reducing uncertainty: by 2023, more than 100 court judgments in EU countries had already dealt with platform workers’ employment status, reaching different conclusions (European Parliament, 2023).

As a second example, consider the General Data Protection Regulation (GDPR, Regulation 2016/679), aimed at protecting personal data within the EU, and the Regulation for the Free Flow of Non-Personal Data (Regulation 2018/1807). The two regulations, taken together, should in principle allow for the seamless movement of any data across the EU territory, overcoming national barriers, expanding the data market from the national to the EU level, with significant pro-efficiency effects.¹⁴

Finally, consider the DMA, a regulatory framework designed to make digital markets fair and contestable. The DMA mandates online platforms enjoying highly entrenched market power (aka “gatekeepers”) to, for example, make some of their services interoperable. Without that regulatory provision, competition in those markets would remain low because the gatekeepers benefit from large network effects, giving them an enormous advantage vis-à-vis potential challengers. By opening markets up to competition, the DMA aims to tap into their potential and drive them to generate more value for the economy than they currently do.

The DSM is, however, far from being completed, and often regulations that are good on paper fail to deliver when implemented.¹⁵ That is probably the better explainer of the competitive disadvantage suffered by the EU when compared with much less fragmented economies, such as the US or China (Letta, 2024; Draghi, 2024). Europe currently lacks large companies (aka European champions) that can rival global digital giants. Rather than excessive EU regulation, the reason may be traced to the fragmented EU market, which is not a good basis for businesses to get started, financed and scaled up.

Contrary to the general sentiment, EU regulation has the potential, if well-crafted, to support, rather than hinder, the emergence of European champions.

Regulation, however, does not come for free. Three categories of efficiency costs should be associated with it:

• administrative and enforcement costs for public authorities
• compliance costs for business
• market distortion risks.

As an example of administrative and enforcement costs for public authorities, consider the DSA. The DSA attempts to limit the spread of illegal and/or harmful content on online platforms. In 2024, the European Commission spent €50 million on DSA enforcement (70% of which covered operation and administrative costs; European Commission, 2024).

As an example for compliance costs for business, consider the GDPR. Its average compliance costs have been estimated to amount to €500,000 for small and medium enterprises (SMEs) and up to €10 million for large organisations (Draghi, 2024).

Regarding market distortion risks, note that whenever regulators introduce constraints to companies’ behaviour, they interfere with natural market dynamics; this may result in inefficient outcomes. For example, regulators may reduce competition by (intentionally or inadvertently) erecting barriers to market entry. Or they may favour companies that do not deserve to be advantaged because they are less efficient than their competitors. As an illustration, consider the EU Copyright Directive (Directive 2019/790). The directive introduced a right for publishers to be remunerated for the use of small news excerpts by online news aggregators. Since online news aggregators and publishers supply largely complementary (not substitute) products, the introduction of the new right has created inefficient frictions, potentially leading to more concentration in both markets.¹⁶

Another example of a regulation introducing significant distortions is the GDPR.¹⁷ One of the most consequential of those distortions derives from the introduction of mandatory data management processes for companies that meet certain requirements (such as dealing with large or highly sensitive datasets). The adoption of the mandatory data management processes entails high fixed costs. This implies that, among those companies that meet the requirements, the larger ones have a comparative advantage (the cost of complying with the GDPR is for them relatively small compared to the cost for medium companies, due to the resulting economies of scale). Peukert et al. (2020) report evidence that web technology service markets became more concentrated after the introduction of the GDPR.

The risk of distortion can never be fully eliminated when regulation is adopted. However, it can be minimised by restricting regulatory intervention to what is strictly needed to enhance market efficiency. That is, by making sure that existing regulations are tightly bound to the market failure that they intend to correct. In other words, regulation should not tackle issues that the market can sort out by itself.

This is often not the case with EU regulation. Consider, for example, the AIA, a comprehensive framework regulating the development, placing on the market, putting into service, and use of AI systems in the EU. Article 15 of the AIA mandates that AI systems considered high-risk (i.e. with a higher potential to harm) are accurate, robust and cyber-secured. Competition in the design of AI systems is, however, high, and accuracy is one parameter in which developers compete most strenuously.¹⁸ Therefore, one should expect the market to incentivise and reward more accurate AI systems. The need for a regulatory requirement for accuracy is thus all but evident.

Distributive goals

Since efficiency entails an increase in the aggregate value produced by the economy, when pursuing efficiency goals, regulatory or deregulatory strategies have the potential to make everybody happy (when the pie gets bigger, subsequent arrangements can theoretically be made so that no one is worse off).¹⁹ However, regulation also directly determines how that value is redistributed. Regulation affects the value share that different market players, users, platforms, traditional industries, developers, citizens, etc., receive. For example, a regulatory framework may be friendly to data business models, with few restrictions on personal data use, favouring AI system developers. Conversely, a strict privacy regime may grant a higher service quality for users, for example, because they are exposed to a lower risk of misuse of their personal information when using social network services. However, this may reduce the amount of data available to companies and thus reduce their competitiveness.²⁰ In other words, regulation also entails making political, distributive choices.

For several reasons, there is a limit to how much markets can achieve. Efficiency goals alone cannot ensure that value distribution would be considered fair by members of society, even if those members operate and manifest their preferences in the economy’s markets.²¹

Policies with distributive effects also affect productivity. And a more symmetric distribution of resources does not necessarily entail worse economic performance. The most obvious example is policies to foster market competition: they induce a redistribution of value from producers to consumers, but they also expand production, making the economy richer.

However, in this paper, I specifically focus on distributive goals as regulatory goals that always involve a trade-off. When considering those goals, the value in the economy is assumed to be given and fixed (recall that value is not only how much output is produced in the economy, but also the quality and variety of production). Thus, with distributive goals, the regulator chooses between different combinations of the factors (such as competitiveness, privacy, safety, etc.) that generate the same amount of value. Each combination involves different winners and losers.

First, even the most efficient regulation cannot eliminate all possible sources of market failure. The most straightforward illustration is individual behaviour. If users behave irrationally (i.e. they make suboptimal choices that do not maximise their interest),²² the market cannot reach fully efficient outcomes (this is obvious because there is a mismatch between how much individuals are willing to pay for a product and how much they really value it). Regulation cannot correct that source of market failure directly. It can only mandate the outcome regulators believe should emerge if individuals were rational.

For example, consider the privacy paradox:²³ when accessing social networks, users may relinquish their personal information as they perceive it as having little value. At the same time, when surveyed, they may express strong preferences for high privacy standards. This may motivate more prescriptive privacy regulation, such as forbidding specific personal data uses, and go beyond the efficiency goal of transparent and competitive data markets. Bonnefon et al. (2016) identify another illustrative paradox. They find that people favour programming automated vehicles to sacrifice their passengers to avoid killing a greater number of people in a road accident. However, the same respondents would refrain from buying such cars. Thus, markets would be unlikely to yield socially optimal equilibria, begging for prescriptive regulatory intervention.

Second, markets can be fully efficient (i.e. produce all the value they can) and still lead to outcomes deemed unfair by common standards. Consider, for example, discrimination. The data economy is founded on the idea that users’ information can be employed to produce value. The more producers know about their customers, the more they can tailor their offer to their users’ preferences. Online e-commerce platforms can, for example, set higher prices for buyers who have a higher willingness to pay.²⁴ At an aggregate level, this may be efficient because the supplier (the e-commerce platform) sells to everyone and sells as much as it can. However, most of that aggregate value is absorbed by the discriminating platform. In the extreme case of first-price perfect discrimination, users are left only with the consumption value (Mariniello, 2022).²⁵ They enjoy no “surplus”, i.e. the possible difference between how much they are willing to pay and how much they actually pay. Since they all pay different prices, they may feel unfairly treated.

Discrimination is opposed by the EU digital regulator, and this is reflected in several explicit regulatory provisions. For example, Article 6(12) DMA prohibits gatekeepers from imposing “discriminatory general conditions of access” for business users. Article 10(2) AIA mandates data governance requirements to mitigate discrimination by high-risk AI systems. Article 9 GDPR prohibits the processing of personal data that would lead to discriminatory outcomes (such as ethnic origin or religious beliefs). All these provisions express a distributive choice by the regulator because, on purely efficiency grounds, discrimination could theoretically be justified.²⁶

Finally, markets may be efficient within their boundaries. But they usually have social spillover effects. For example, consider the market for algorithmic management (AM), or the use of AI by employers willing to monitor and manage their employees. This market may be very efficient, but the ones who are mostly affected by it are outside its boundaries. They are the AM customers’ employees. Without regulatory intervention overseeing the use of AI in job places, the AM markets would likely harm workers.²⁷

Figure 1 describes an illustrative trade-off by plotting the value generated in the economy as a concave line using different combinations of two commodities: innovation/competitiveness (on the y-axis) and privacy (on the x-axis). The further the value line moves away from the origin, the higher the value (thus V’ > V).

By adjusting its regulatory action according to its efficiency goals, the EU can aim to hop on combinations where the total value is higher. For example, eliminating the distortions introduced by the GDPR could increase privacy (because of a more efficient implementation) and innovation/competitiveness (because companies would face fewer constraints). An efficient revision of the GDPR would allow the EU economy to move from V to V’.

However, there is a limit to how much efficiency can be achieved. That limit is represented by the purple line, the production-possibility frontier (PPF).²⁸ Once the economy is on that line, the value pie cannot get bigger. Then, political choices determine whether the economy should yield relatively more privacy compared to competitiveness (thus choosing to be at the combination point A’) or vice versa.

Disentangling efficiency from fairness in the EU deregulatory strategy

Regulatory laws should not just be counted: quantification brings little information about their impact on the economy. Regulation should rather be weighed and assessed in the context in which it is implemented. In principle, the European Commission has the tools to make such a qualitative assessment: its “better regulation toolbox” includes an impact assessment process through which the Commission is supposed to assess the implications of EU laws for all relevant stakeholders (users, suppliers, small companies, citizens, etc.).²⁹ In practice, however, impact assessments are often used to justify ex post policy decisions taken before the start of the evaluation process.³⁰

There is a high risk that this could be the case when considering the Commission’s strategy for the deregulation of the EU digital economy. Facing increasing public pressure to ease EU regulation,³¹ the Commission has grown a prejudice against the regulatory burden that companies developing or adopting digital services must bear. In its Competitiveness Compass, the European Commission (2025a) sets an overarching target of reducing the cost of all administrative burdens for companies by 25% (35% for SMEs).

However, the size of the quantitative cut cannot be predetermined. It should rather be the end result of a realignment of regulation with its efficiency and distributive goals (the qualitative assessment may well indicate that a quantitative cut of the overall regulatory burden is necessary; as I have explained, this is most certainly the case for the GDPR, for example). In other words, the Commission is inverting the logic, assuming what should come at the conclusion of a coherent evaluative process. It has internalised (without justifying it) the mantra put forward by business lobbies, that regulation as such is a necessary evil, and that if the EU intends to be successful in the digital space, it must reduce the regulatory burden to the maximum possible extent.³²

In this paper, I propose a way to escape that form of regulatory capture. This consists of laying out explicitly what parts of EU regulations primarily stem from efficiency-oriented considerations and which ones are instead primarily distributive-oriented choices.

Explicitly disentangling the two categories of goals brings a significant advantage. It forces EU policymakers to take responsibility for the political choices they make. At the same time, it narrows the discussion on efficiency improvements at the technical level: economics, not politics, should determine how to reduce small companies’ administrative burdens, for example.

When making its political choices, the Commission should openly admit the costs that those choices entail. Assuming that political preferences may be more homogeneous within than between the EU, the US or China (for example, for cultural and historical reasons), those choices necessarily affect those economies’ comparative economic performance. China may have a structural advantage in the data economy if, for example, EU citizens attach a stronger value to privacy than Chinese citizens, on average.

Some may feel disturbed by the gloomy corollary of this analysis: that, because of its social model, on several metrics, the European digital economy may never be able to catch up with the US or China. However, that conclusion is disturbing only to the extent that the relevant metrics used for international comparison focus on the quantitative rather than the qualitative aspects of the value that the different economies produce. The qualitative aspects of digital goods and services (such as their safety, their social impact, or the risk they pose to privacy) are very relevant too. However, there can be no objective indicator capturing quantitative and qualitative elements together for international comparison, simply because citizens in different economies express different preferences. Therefore, their appreciation of the combination of quantitative and qualitative elements differs. International comparisons can thus achieve very little, besides providing some rough indications of the areas where economies can become more efficient, mirroring what others are doing.

This suggests that it would make little sense if the EU were to attempt to bridge its technological gap by compromising on its distributive choices, rather than just focusing on efficiency gains and acknowledging that some of its structural differences with other economies limit what it can achieve.

In his recipe for the EU economy, Mario Draghi had one simple, powerful selling point for investing in competitiveness: refrain from investing, and by 2050, the EU will not be able to “preserve our [European] values” (Draghi, 2024). Given how rapidly the digital economy is transforming society, the EU will not have the luxury to wait until 2050 to pursue its distributive choices.

• 1 For an overview, see Filippucci et al. (2024).
• 2 For a general review of the relation between data analytics and productivity, see Bogdan and Borza (2019).
• 3 For a full overview of EU digital legislation, see: https://www.bruegel.org/dataset/dataset-eu-legislation-digital-world.
• 4 At the beginning of her second mandate, President von der Leyen anticipated a slowing down on potential new regulatory actions and initiatives to reduce regulatory burden to unleash companies’ growth potential. Commission Work Programme 2025 – Moving forward together: A Bolder, Simpler, Faster Union.
• 5 Omnibus package contains different independent proposals aimed at the simplifications of EU directives or regulations.
• 6 This classification is broadly consistent with the mainstream economic literature on regulation. See, for example, Joskow and Rose (1989) and Baldwin et al. (2011).
• 7 Researchers have attempted to explain that there is no regulation vs competitiveness dichotomy (see, for example, Bradford (2024)). I see merit in this argument, to the extent that many factors influence competitiveness, and regulation cannot be considered the primary culprit for low economic performance. However, assuming everything else is equal, regulatory constraints, by definition, affect competitiveness. To be binding, a regulatory constraint must prevent rational players from implementing actions they would otherwise undertake to maximise their profits. Thus, for example, a rule forcing companies not to use personal data without consent, if binding, necessarily implies a reduction of companies’ potential profits.
• 8 See Draghi (2024), for example.
• 9 According to a 2024 European Investment Bank survey, for example, European companies believe that the major obstacle to investment is their inability to find employees with the right skills, rather than regulation. See https://www.eib.org/en/press/all/2024-386-eib-investment-survey-2024-more-than-60-of-european-companies-have-invested-in-climate-mitigation-and-adaptation-and-more-than-70-in-their-digital-transformation.
• 10 This is a cornerstone of liberal economic thinking. See Smith (2000).
• 11 Competition increases the value generated in an economy by forcing companies to expand their supply, become more efficient and innovate to stay ahead of their rivals. Competition selects market players based on their efficiency performance (inefficient companies that cannot withstand competition must exit the market).
• 12 That is: companies do not “internalise” the (positive or negative) “externalities” that their products have on the economy. Thus, they make suboptimal decisions that do not maximise the total value that the market can produce. For example, they may decide not to share their data with other companies, even if data sharing would not be costly for them and would enable others to produce additional value.
• 13 According to the EIB survey mentioned in footnote 9, uncertainty is considered a major obstacle to investment by 44% of European companies, while regulation by 32%.
• 14 For an estimation of the efficiency effects of larger EU data markets, see, for example: https://ec.europa.eu/newsroom/repository/document/2025-13/EDM_2024__2026__First_Report_on_Facts_and_Figures_9cU0iIjcuSZhjmwI9TchwqenidQ_114043.pdf.
• 15 The enforcement of the GDPR, for example, is not uniform throughout the EU, as it should be. See Gentile and Lynskey (2022).
• 16 Online news aggregators use small news excerpts to promote the publishers’ content. Thus, they increase feeding traffic to publishers. The introduction of frictions in the use of exerts thus tend to penalise those publishers that need feeding traffic the most, i.e. smaller ones. See Quintais (2019) and Mariniello (2022).
• 17 For an overview of competitive effects of the GDPR, see Gal and Aviv (2020).
• 18 According to the Stanford 2025 AI Index Report, “The [AI] frontier is increasingly competitive—and increasingly crowded.” (Maslej et al., 2025).
• 19 Technically, there is scope for efficiency gains if a Pareto improvement is possible. That is, resources can be allocated in such a way that the value for at least one market player is increased while value is not reduced for any of the remaining market players. Thus, if an efficiency-oriented regulation initially causes a loss for a market player, that loss can always be compensated through transferring value from one player to the other, for example, through efficient taxation (this is a version of the Coase’s theorem (Coase, 2013)).
• 20 Note that the trade-off between privacy and competitiveness is not always the case. Sometimes, privacy protection can enhance competitiveness. See Acquisti et al. (2016) and OECD (2024).
• 21 What “fairness” amounts to is a matter for political philosophy. For a discussion, see, for example, Rawls (2001) and Rawls (2005).
• 22 The extensive work by Daniel Kahneman and Amos Tversky has proved that this is indeed the case in the greatest number of cases, and that humans’ choices are constrained by their bounded rationality. See Kahneman (2003).
• 23 See Acquisti (2004) and Kokolakis (2017).
• 24 Chen et al. (2016) shows, for example, that sellers on Amazon Market Place would adjust their pricing even hundreds of times per day through algorithmic optimisation.
• 25 First-price perfect discrimination occurs when a supplier has full information about its customers’ willingness to pay for its products.
• 26 There are many reasons why discrimination can decrease or increase efficiency. It depends on the context. For an analysis, see Papandropoulos (2007).
• 27 For an overview of AM’s harmful effects, see Wood (2021). The European Commission is currently considering whether to complement the AIA with an AM dedicated regulation (Kroet, 2025).
• 28 The use of the production-possibility frontier is a standard economic methodology for the graphical appreciation of trade-offs. See Lovell (1993).
• 29 See https://commission.europa.eu/law/law-making-process/better-regulation/better-regulation-guidelines-and-toolbox/better-regulation-toolbox_en.
• 30 For an analysis of the reasons why impact assessments tend to be ineffective, see Carroll (2010).
• 31 Most notably, Draghi (2024).
• 32 See Nielsen (2024).

References

Acquisti, A., Taylor, C., & Wagman, L. (2016). The Economics of Privacy. Journal of Economic Literature, 54(2), 442–92.

Acquisti, A. (2004). Privacy in Electronic Commerce and the Economics of Immediate Gratification. Proceedings of the 5th ACM Conference on Electronic Commerce, 21–29.

Baldwin, R., Cave, M., & Lodge, M. (2011). Understanding Regulation: Theory, Strategy, and Practice. Oxford University Press.

Bogdan, M., & Borza, A. (2019). Big Data Analytics and Organizational Performance: A Meta-Analysis Study. Management and Economics Review, 4(2), 1–13.

Bonnefon, J.-F., Shariff, A., & Rahwan, I. (2016). The Social Dilemma of Autonomous Vehicles. Science, 352(6293), 1573–76.

Bradford, A. (2024). The False Choice Between Digital Regulation and Innovation. Northwestern University Law Review, 118(2).

Brynjolfsson, E., Li, D., & Raymond, L. (2025). Generative AI at Work. The Quarterly Journal of Economics.

Carroll, P. (2010). Does Regulatory Impact Assessment Lead to Better Policy? Policy and Society, 29(2), 113–22.

Chen, L., Mislove, A., & Wilson, C. (2016). An Empirical Analysis of Algorithmic Pricing on Amazon Marketplace. Proceedings of the 25th International Conference on World Wide Web, 1339–49.

Coase, R. H. (2013). The Problem of Social Cost. The Journal of Law and Economics, 56(4), 837–77.

Draghi, M. (2024). The Draghi Report: In-Depth Analysis and Recommendations (Part B).

European Commission. (2024). Annual report on costs incurred for the fulfilment of the Commission’s tasks under the Digital Services Act (DSA) for 2024.

European Commission. (2025a). Competitiveness compass.

European Commission. (2025b). Extension of certain mitigating measures available for small and medium sized enterprises to small mid-cap enterprises and further simplifying measures [Proposal for a directive and a regulation].

European Parliament. (2023, December 13). Platform workers: deal on new rules on employment status.

Eurostat. (2025, January 23). Usage of AI technologies increasing in EU enterprises.

Filippucci, F., Gal, P., Jona-Lasinio, C., & Nicoletti, A. L. G. (2024). The Impact of Artificial Intelligence on Productivity, Distribution and Growth: Key Mechanisms, Initial Evidence and Policy Challenges. OECD Artificial Intelligence Papers, 15.

Gal, M. S., & Aviv, O. (2020). The Competitive Effects of the GDPR. Journal of Competition Law & Economics, 16(3), 349–91.

Gentile, G., & Lynskey, O. (2022). Deficient by Design? The Transnational Enforcement of the GDPR. International & Comparative Law Quarterly, 71(4), 799–830.

Joskow, P. L., & Rose, N. L. (1989). The Effects of Economic Regulation. Handbook of Industrial Organization, Vol. 2, 1449–1506.

Kahneman, D. (2003). Maps of Bounded Rationality: Psychology for Behavioral Economics. American Economic Review, 93(5), 1449–75.

Kokolakis, S. (2017). Privacy Attitudes and Privacy Behaviour: A Review of Current Research on the Privacy Paradox Phenomenon. Computers & Security, 64, 122–34.

Kroet, C. (2025, February 14). EU Parliament Plans AI and Workplace Initiative for Next Year. Euronews.

Letta, E. (2024). Much More than a Market: Speed, Security, Solidarity. Report to the European Commission.

Lovell, C. A. K. (1993). Production Frontiers and Productive Efficiency. In H. O. Fried, C. A. K. Lovell, & S. S. Schmidt (Eds.), The Measurement of Productive Efficiency: Techniques and Applications (pp. 3–67).

Mariniello, M. (2022). Digital Economic Policy: The Economics of Digital Markets from a European Union Perspective. Oxford University Press.

Maslej, N., Fattorini, L., Perrault, R., Gil, Y., Parli, V., Kariuki, N., Capstick, E., Reuel, A., Brynjolfsson, E., Etchemendy, J., Ligett, K., Lyons, T., Manyika, J., Niebles, J. C., Shoham, Y., Wald, R., Walsh, T., Hamrah, A., Santarlasci, L., Lotufo, J. B., Rome, A., Shi, A., & Oak, S. (2025). The AI Index 2025 Annual Report. AI Index Steering Committee, Institute for Human-Centered AI, Stanford University.

Morgan, K. (2025, March 20). AI and Drug Repurposing. The New York Times.

Nielsen, N. (2024). Businesses, Unions Clash on Draghi’s Call to Cut EU Regulations. Euractiv.

OECD. (2024). The Intersection between Competition and Data Privacy – Background Note.

Papandropoulos, P. (2007). How Should Price Discrimination Be Dealt with by Competition Authorities? Concurrences, 3, 34–38.

Peukert, C., Bechtold, S., Batikas, M., & Kretschmer, T. (2020). European Privacy Law and Global Markets for Data. Center for Law & Economics Working Paper Series, No. 1.

Quintais, J. (2019, June 17). The New Copyright Directive: A Tour d’Horizon (Part II) – Of Press Publishers, Upload Filters and the Real Value Gap. Kluwer Copyright Blog.

Rawls, J. (2001). Justice as Fairness: A Restatement. Harvard University Press.

Rawls, J. (2005). Political Liberalism (2nd ed.). Columbia University Press.

Saharova, D., Bohl, M., Eder, A., & Chiavarini, L. (2025). Green Computing in the AI Era. White Paper, World Fund VC.

Smith, A. (2000). The Wealth of Nations. Modern Library. (Original work published 1776).

Wood, A. J. (2021). Algorithmic Management: Consequences for Work Organisation and Working Conditions. JRC Working Papers Series on Labour, Education and Technology, No. 2021/07.