AI Boosts EU Productivity by 4% Without Cutting Jobs, Major Study Finds

A comprehensive study of over 12,000 European companies reveals for the first time the concrete impacts of artificial intelligence on productivity and employment. The study by the non-profit organization Centre for Economic Policy Research (CEPR), based in Paris, provides robust causal relationships and paints a differentiated picture of AI adoption in Europe.

Key Findings of the Study

The researchers identified three essential insights. First, AI use increases labor productivity in the EU average by 4 percent. This effect is statistically robust and economically significant, but falls short of optimistic forecasts of a comprehensive productivity boom.

Second, the scientists found no evidence of employment declines due to AI adoption. While simple comparisons initially suggested more employees in AI-using companies, this relationship disappears when selection effects are taken into account. The combination of productivity gains without job losses points to a specific mechanism: capital deepening. AI complements the work of employees without replacing them.

Third, productivity gains are unevenly distributed. Medium and large companies benefit significantly more than smaller firms.

Methodological Approach

The study uses data from the European Investment Bank Investment Survey (EIBIS) combined with balance sheet data from Moody’s Orbis. To isolate causal effects, the researchers developed a novel instrumental variables strategy. For each EU company, they identified comparable US firms by sector, size, investment intensity, innovation activity, financing structure, and management practices. The AI adoption rate of these US counterparts serves as a proxy for the exogenous AI exposure of the EU company.

AI Adoption in Europe: A Heterogeneous Picture

The study shows that average AI use in the EU and USA is similarly high. Beneath the surface, however, considerable heterogeneity emerges. In financially developed EU countries such as Sweden and the Netherlands, around 36 percent of companies use big data analytics and AI (as of 2024), comparable to the USA. In less financially developed EU economies such as Romania and Bulgaria, the adoption rate is only about 28 percent.

Company size also plays a decisive role. While 45 percent of large companies (over 250 employees) use AI, this applies to only 24 percent of small firms (10 to 49 employees). AI-using companies systematically invest more, are more innovative, and face greater difficulties finding qualified workers.

The Crucial Role of Complementary Investments

A central finding of the study: AI adoption alone is not sufficient. Companies must make complementary investments to unlock full potential. The analysis reveals remarkable differences in the impact of various types of investment:

• An additional percentage point of investment in software and data infrastructure increases the productivity effect of AI by 2.4 percentage points
• An additional percentage point for employee training strengthens AI-driven productivity gains by 5.9 percentage points

These findings underscore that the productivity dividend from AI depends not only on acquiring the technology, but on companies’ ability to integrate it through investments in intangible assets and human capital.

Wage Development in AI-Using Companies

Employees in AI-using companies benefit through higher wages, both in total and per employee. Whether these wage gains persist in the long term and whether they are distributed evenly across different skill levels remains an open question according to the researchers.

Implications for Economic Policy

The study authors draw several policy conclusions. First, medium and large companies benefit especially from AI. Policy measures should therefore help smaller firms reach the critical size needed to benefit from AI. This requires well-functioning financial markets that can direct capital to innovative, growing companies.

Second, public support must go beyond subsidizing AI hardware or software. Effective support requires incentives for investments in integration, workflow redesign, and continuous learning. Training programs should prioritize so-called “fusion skills”: capabilities such as prompt engineering, data responsibility, and human oversight in AI-driven decision-making processes.

Third, the researchers warn against complacency. While the study shows no immediate job losses, the observed capital deepening effects could be temporary. As AI systems become more capable and integration experience grows, job-displacing effects could occur. Moreover, wage gains could disproportionately benefit highly qualified workers and increase income inequality.

Data Foundation and Limitations

The analysis is based on survey data from the European Investment Bank Investment Survey, combined with balance sheet data from Moody’s Orbis. The study captures short-term effects on productivity levels, not long-term impacts on total factor productivity growth. The authors emphasize that their results do not necessarily reflect the views of the participating institutions (Bank for International Settlements and European Investment Bank).

A comprehensive study of over 12,000 European companies reveals for the first time the concrete impacts of artificial intelligence on productivity and employment. The study by Aldasoro et al. (2026) provides robust causal relationships and paints a differentiated picture of AI adoption in Europe.

Key Findings of the Study

The researchers identified three essential insights. First, AI use increases labor productivity in the EU average by 4 percent. This effect is statistically robust and economically significant, but falls short of optimistic forecasts of a comprehensive productivity boom.

Second, the scientists found no evidence of employment declines due to AI adoption. While simple comparisons initially suggested more employees in AI-using companies, this relationship disappears when selection effects are taken into account. The combination of productivity gains without job losses points to a specific mechanism: capital deepening. AI complements the work of employees without replacing them.

Third, productivity gains are unevenly distributed. Medium and large companies benefit significantly more than smaller firms.

Methodological Approach

The study uses data from the European Investment Bank Investment Survey (EIBIS) combined with balance sheet data from Moody’s Orbis. To isolate causal effects, the researchers developed a novel instrumental variables strategy. For each EU company, they identified comparable US firms by sector, size, investment intensity, innovation activity, financing structure, and management practices. The AI adoption rate of these US counterparts serves as a proxy for the exogenous AI exposure of the EU company.

AI Adoption in Europe: A Heterogeneous Picture

The study shows that average AI use in the EU and USA is similarly high. Beneath the surface, however, considerable heterogeneity emerges. In financially developed EU countries such as Sweden and the Netherlands, around 36 percent of companies use big data analytics and AI (as of 2024), comparable to the USA. In less financially developed EU economies such as Romania and Bulgaria, the adoption rate is only about 28 percent.

Company size also plays a decisive role. While 45 percent of large companies (over 250 employees) use AI, this applies to only 24 percent of small firms (10 to 49 employees). AI-using companies systematically invest more, are more innovative, and face greater difficulties finding qualified workers.

The Crucial Role of Complementary Investments

A central finding of the study: AI adoption alone is not sufficient. Companies must make complementary investments to unlock full potential. The analysis reveals remarkable differences in the impact of various types of investment:

• An additional percentage point of investment in software and data infrastructure increases the productivity effect of AI by 2.4 percentage points
• An additional percentage point for employee training strengthens AI-driven productivity gains by 5.9 percentage points

These findings underscore that the productivity dividend from AI depends not only on acquiring the technology, but on companies’ ability to integrate it through investments in intangible assets and human capital.

Wage Development in AI-Using Companies

Employees in AI-using companies benefit through higher wages, both in total and per employee. Whether these wage gains persist in the long term and whether they are distributed evenly across different skill levels remains an open question according to the researchers.

Implications for Economic Policy

The study authors draw several policy conclusions. First, medium and large companies benefit especially from AI. Policy measures should therefore help smaller firms reach the critical size needed to benefit from AI. This requires well-functioning financial markets that can direct capital to innovative, growing companies.

Second, public support must go beyond subsidizing AI hardware or software. Effective support requires incentives for investments in integration, workflow redesign, and continuous learning. Training programs should prioritize so-called “fusion skills”: capabilities such as prompt engineering, data responsibility, and human oversight in AI-driven decision-making processes.

Third, the researchers warn against complacency. While the study shows no immediate job losses, the observed capital deepening effects could be temporary. As AI systems become more capable and integration experience grows, job-displacing effects could occur. Moreover, wage gains could disproportionately benefit highly qualified workers and increase income inequality.

Data Foundation and Limitations

The analysis is based on survey data from the European Investment Bank Investment Survey, combined with balance sheet data from Moody’s Orbis. The study captures short-term effects on productivity levels, not long-term impacts on total factor productivity growth. The authors emphasize that their results do not necessarily reflect the views of the participating institutions (Bank for International Settlements and European Investment Bank).