When Mario Draghi published his report on the Future of European Competitiveness in September 2024, one observation stood out above all others: Europe’s competitiveness problem is a technology problem.
Here’s Draghi in his own words: “the EU has less activity in sectors in which much of the productivity growth has originated in recent years, notably the ICT sector and the exploitation of large-scale digital services. Due to slow technology diffusion within industries, the EU’s productivity growth gap compared to the US was particularly pronounced in these industries with very high productivity growth.”
This perspective is central for Europe as it works to regain competitiveness, but, unfortunately, one that now needs to be observed with greater clarity. Next week, European leaders will meet at the March EU summit to discuss the competitiveness agenda – and they will do so at a time when Europe’s reform efforts are losing steam and direction.
In the past months, Brussels has yet again returned to its habitual routine of quarrelling about basic industries and state aid to big industrials – indeed, various versions of a “European preference” and “Made in Europe” or “buy European” policies are floated. There is now even a proposal for raising the share of the manufacturing sector in Europe’s value added, which can only happen if the growth of other sectors is artificially kept very low. It is Groundhog Day – the same type of ideas, initiatives, discussions, and rebuttals that have taken up so much energy since the 1990s, yet failed to deliver anything of value.
Increasingly, we have two parallel worlds in the competitiveness debate. One is the real world, where Europe increasingly struggles to keep up with frontier innovation and offer a hospitable climate for experimenters, innovators, and investors. The other is a narrative world where Europe’s main competitiveness problem is said to be found in the shrinking industrial sector. No doubt some industrial sectors are having a hard time with high energy costs and growing competition, but Europe can only regain competitiveness and connect with modern streams of productivity growth if it focuses on challenges of the real world. Building on the Digital Omnibus, the AI Act revision, the Digital Markets Act review, and new ideas for a single market for services, it should redouble current reform efforts with the view of boosting technology performance.
Defining the Technology Performance Problem: Upstream and Downstream
Mario Draghi’s diagnosis of Europe’s technology-performance problem was correct, but not new. The evidence has been accumulating for decades. In the late 1990s, the average EU worker produced 95 per cent of what their American counterpart generated per hour. That ratio has since declined, and the timing is not coincidental: the divergence maps almost exactly onto the rise of the digital economy.
Figure 1 shows that the productivity gap is highest within the services sector, where the US achieved a growth index of 159 compared to a stagnant 110 in the EU-12. While European gains have been more concentrated in traditional industry. The data in the figure confirms Draghi’s diagnosis: the EU has not been able to harness the positive productivity effect of digital technologies.
Figure 1: Hourly labour productivity growth by main sectors for the EU-12 and the US, 2021 (2001=100)
Source: ECIPE calculations based on EUKLEMS & INTANProd data.
There are different factors explaining why Europe are behind in technology and, especially, technology-service performance. Let’s start upstream and gradually move to downstream factors.
Long-term patterns in upstream Research & Development (R&D) funding are one factor. Europe under-invests in research and development relative to countries at the technological frontier, such as the United States. In 1991, US R&D expenditure was already approximately 40 per cent larger than the EU’s. By 2021, it was 77 per cent larger. Between 2014 and 2021, the average year-on-year growth rate in R&D expenditure in the US was 5.6 per cent, compared to 2.7 per cent in the EU. This means that the gap is not only growing in absolute terms; it is accelerating.
Only 7 out of 126 first-level EU regions have R&D spending as a share of GDP that exceeds the US average. European economic powerhouses such as Bavaria, Île-de-France, and Lombardy all devote fewer resources to R&D in relation to their economic output than the US average. Baden-Württemberg, the EU’s top R&D-spending region, allocated USD 33 billion, roughly half of the USD 67 billion invested by the private sector in Silicon Valley.
Let’s stay on this point a bit longer. A closer examination of R&D data reveals that the EU’s overall shortfall closely reflects the structural problem identified earlier. While the EU industrial R&D, measured as a share of GDP, is broadly comparable to the US, the divergence is almost entirely in services. In 2003, the share of services in total corporate R&D was 9.2 per cent in the EU versus 18.2 per cent in the US. By 2023, services R&D accounted for just 15.3 per cent of total corporate R&D in the EU, compared to 43.5 per cent in the US. Manufacturing still dominates EU business R&D at 84.3 per cent.
The sectoral breakdown is more striking still. In the US, computer programming accounts for 18.6 per cent of total business R&D and software for a further 15.8 per cent. In the EU, computer programming accounts for only 6.2 per cent, and software is virtually absent at 0.01 per cent. This is a fundamental divergence in the structure of the innovation economy. The sectors driving the largest productivity gains are precisely those in which European firms are most absent.
Figure 2 below shows EU and US industry R&D tracking closely throughout the period. Again, the divergence is entirely in services: US services R&D accelerates sharply from the early 2010s onward, while the EU line remains flat.
Figure 2: Corporate R&D spending in the industry and services sectors as a share of GDP for the EU-27 and the US, 2003–2023 (percentage of GDP)
Source: ECIPE calculations based on EU Industrial R&D Investment Scoreboard panel data.
If the preceding data describe a structural lag that has been building for almost two decades, AI investments add another wrinkle – yes, concern! – to Europe’s position. In 2024, the US private sector invested over USD 110 billion in AI, bringing its cumulative total between 2013 and 2024 to USD 471 billion. The EU’s top-performing member states attracted USD 13 billion in 2024, with a cumulative total of USD 64 billion. In generative AI specifically, the US attracted USD 29 billion in 2024. The EU managed USD 1.5 billion. China invested USD 2 billion.
The consequences are visible in model development and compute capacity. In 2024, the US was responsible for 40 new AI models; China for 15. Europe produced three. On compute, the US had 173 supercomputers in 2024, compared to 64 across the whole of Europe.
It does not have to be this way. Moving to some other – and more downstream – factors explaining Europe’s technology-performance problem, it is clear that the digitalised services companies that have emerged in Europe are performing well. This is a point worth repeating: where it is allowed to grow, Europe’s digital economy is in pretty good health. Just look at some key characteristics of these firms (based on data from Eurostat):
- A sector like information and communication services contributes more to productivity growth in Europe than the manufacturing sector.
- The same sector has higher earnings per employee than manufacturing.
- The same sector has a higher share of high-growth firms than any other sector.
This leads us to the final data chart – taking stock of how EU-12 actually compares to the United States in the contribution of ICT services to Gross Domestic Product. Figure 3 shows the comparison over 20 years and, undoubtedly, summarises Europe’s “Draghi problem”. Europe is falling behind because modern and high-productive services do not give enough of a contribution to general levels of economic growth.
Figure 3: Contribution of ICT services to value-added growth (percentage points, 2000-2020)
Source: ECIPE calculations based on EUKLEMS & INTANProd data.
Regulation and Technology Performance
Two other factors add to Europe’s problem with technology performance: industry structure and the development of regulation. Industry structures (sectors that represent a big part of the economy) take time to change, and the best attitude from policymakers is to allow change to happen – and avoid working against the natural process of change. The second factor, regulation, can be addressed more directly. In fact, a better regulatory climate for the development of technology and services would unleash much faster growth in Europe.
Right now, high regulatory restrictions affect the extent to which firms can adopt digital technologies. For instance, restrictive regulation can limit access to digital technologies and services; decrease a firm’s ability to use ICT-related capital; and disincentivise firm growth, which hinders efforts to adopt digital technologies. Much research shows that there is a negative association between digital adoption rates and the restrictiveness of digital regulation across the EU. The more restrictive the regulation, the less use of new digital products and services.
In the last years, the amount of EU regulation in the digital sector has continued to grow – indeed, at a remarkably fast clip. Regulation is necessarily not wrong, but the design and quality of it matter fundamentally. Regulations such as the General Data Protection Regulation (GDPR), the Digital Markets Act (DMA), and the Artificial Intelligence Act (AI Act) are complex. They combine product regulation, market access rules, and competition-style behavioural obligations in several overlapping instruments. They are frequently ambiguous, generate significant legal uncertainty, and do not easily map onto traditional regulatory categories, making their economic costs difficult to measure and easy to underestimate.
Moreover, they don’t always align with each other and other parts of the EU rulebook. For instance, one EU regulation mandates as much interoperability as possible, while another mandates the strictest form of data privacy controls. Firms may soon be in the position of having to violate one EU regulation to comply with another.
These regulations also impact what economists call endowments – especially how endowments can be used in the economy. Historically, economists studied three key endowments: land, labour, and capital. In the digital economy, the key endowments are data and digital competencies. When regulation restricts access to these endowments, it does not merely raise compliance costs; it reshapes what firms can produce, which markets they can serve, and how they compete. In the most damaging cases, it causes new services and innovations not to be introduced in Europe at all.
It is the behavioural effects that Europe should be mostly worried about. Figure 4 sketches a simple framework for thinking about how restrictions on modern endowments ripple through the economy and impact firm and user behaviour. By limiting the use of the endowments, regulations shape the way firms create different competitive advantages and how they specialise, which ultimately leads to specific economic flows such as sales and investment. Hence, restrictions on data and other modern endowments lead to downstream effects for firms. They can be likened to a regulation that would restrict the human capital of labour – another endowment – and limit their use.
Figure 4: Model of the behavioural effects of regulation
The behavioural effects have been identified empirically in the case of the GDPR (other regulations are too recent for empirical analysis). Take, for instance, Article 5 of the GDPR, which limits firms’ ability to combine data for purposes other than those originally intended. These limitations on endowments affect Europe’s comparative advantage and the specialisation of firms. For example, EU firms had to destroy substantial amounts of data upon the entry of the GDPR. Forward data endowment creation was also damaged: EU firms stored 26 per cent less data on average than US firms two years after the GDPR, and reduced computation relative to US firms by 15 per cent. Ultimately, GDPR also contributed to changes in flows such as innovation, with new app entries falling by half, and firm demographics in favour of old apps and companies.
The DMA are causing other but related impacts on economic behaviour. The regulation builds on the assumption that the combination of endowments should be limited when pursued by gatekeeping platforms. A core aspect of the DMA, for instance, has the explicit intention of making it more difficult for firms to combine different sets of data. Another is that the DMA imposes access and interoperability provisions on an unprecedented scale, regardless of the impact on competition and competitiveness, making it more difficult to use endowments for launching new innovations in the EU.
The obvious result is that, in Europe, gatekeeping firms will have to retire or reduce the use of their endowments and, consequently, the competitiveness of the services they provide or could potentially offer in the future. This impacts sales and adoption in the EU by slowing or preventing the rollout of platform technology – to the detriment of both European consumers and business users. This is one of the reasons why some firms have decided to pause the introduction of new data and AI services in the EU. Moreover, it risks impacting where companies locate investments and place innovation. After all, other jurisdictions have made different calls on how they want to regulate gatekeepers and their endowments. For instance, Japan and the UK have already learned from the DMA and adopted more proportionate and pragmatic approaches.
Going Forward: Redoubling Efforts to Improve Europe’s Policy Environment
Mario Draghi was correct in his analysis: Europe’s chief problem with competitiveness and productivity growth is the slow development of technological change and modern digital services. Europe can improve its performance radically if it establishes better conditions for investment and innovation in digital products and services. This includes capital markets reforms that will spur more financing of new and growing firms. Single market reforms would boost market conditions for technology-driven companies.
However, there is also a regulatory problem – and Draghi also pointed to it. The EU regulates new technologies and innovation a lot more than its peers. Many of the new regulations that have been implemented, proposed, or, in the case of the AI Act, withdrawn have added not just costs but a lot of uncertainty for companies that bring new innovation to the market. As a result, companies have postponed investment in Europe and decided not to introduce new products and services. This is a strong indictment of the policy climate in Europe.
Europe’s laggard technology performance is not an accident. It is the consequence of a regulatory environment that restricts access to the endowments on which digital competitiveness depends. When regulation limits firms’ ability to accumulate, combine, and deploy data, it does not merely raise their costs but reshapes their comparative advantages and, ultimately, the structure of economic activity itself. The result is precisely what the data shows: an EU economy in which high-productivity digital services contribute far less to growth than in the United States.
Europe now needs to catch up with frontier countries. It needs to offer better opportunities than other regions in the world to invest, experiment, innovate, and market new products and services. If it does not, Europe will have a very difficult time achieving the ambitions that Mario Draghi set out.