AI adoption raises firms' labor productivity by about 4% in Europe, driven by capital deepening rather than immediate job losses; productivity and wage gains concentrate in medium and large firms and depend on complementary investments in software, data and training.

AI adoption, productivity and employment: Evidence from European firms

Aldasoro, Iñaki, Gambacorta, Leonardo, Pal, Rozalia, Revoltella, Debora, Weiss, Christoph, Wolski, Marcin · January 01, 2026 · Econstor (Econstor)

openalex quasi_experimental medium evidence 8/10 relevance Summary only summary available; pdf_status=paywall DOI Source PDF

Structured author observations

Linked only from stored provider relations; the raw author line above is never matched by name.

OpenAlex

Latest observation: August 25, 2026

Aldasoro, Iñaki provider ID
Gambacorta, Leonardo provider ID
Pal, Rozalia provider ID
Revoltella, Debora provider ID
Weiss, Christoph provider ID
Wolski, Marcin provider ID

Using an IV that assigns US peers' AI adoption rates to EU firms, the paper finds AI adoption raises firm labor productivity by ~4% driven by capital deepening, with no short-run employment decline, concentrated gains in medium/large firms, and stronger effects where firms invest in software, data, or workforce training.

Citation observations

Cumulative provider counts captured on specific dates; providers are never combined.

0 cumulative citations

OpenAlex · Observed August 25, 2026

View corpus context

This paper provides new evidence on how the adoption of artificial intelligence (AI) affects productivity and employment in Europe. Using matched EIBIS-ORBIS data on more than 12,000 non-financial firms in the European Union (EU) and United States (US), we instrument the adoption of AI by EU firms by assigning the adoption rates of US peers to isolate exogenous technological exposure. Our results show that AI adoption increases the level of labor productivity by 4%. Productivity gains are due to capital deepening, as we find no adverse effects on firm-level employment. This suggests that AI increases worker output rather than replacing labor in the short run, though longer-term effects remain uncertain. However, productivity benefits of AI adoption are unevenly distributed and concentrate in medium and large firms. Moreover, AI-adopting firms are more innovative and their workers earn higher wages. Our analysis also highlights the critical role of complementary investments in software and data or workforce training to fully unlock the productivity gains of AI adoption.

Summary

Main Finding

AI adoption raises firm-level labor productivity by about 4% in the short run. These productivity gains reflect capital deepening rather than reductions in firm-level employment; AI adopters are also more innovative and pay higher wages. Gains are concentrated in medium and large firms and depend importantly on complementary investments (software, data, training).

Key Points

Estimated effect: AI adoption → ~4% higher labor productivity.
Mechanism: capital deepening (higher capital intensity) drives productivity gains; no evidence of adverse firm-level employment effects in the short run.
Distributional pattern: productivity and other benefits are concentrated in medium and large firms (less so in small firms).
Complementarities: firms that invest in complementary software, data, or workforce training capture larger productivity gains from AI.
Other outcomes: AI-adopting firms show higher innovation activity and pay higher wages to workers.
Uncertainty: longer-run employment effects remain unclear (short-run findings do not rule out future labor reallocation or displacement).

Data & Methods

Data: matched EIBIS (European Investment Bank Investment Survey) and ORBIS data covering >12,000 non-financial firms in the EU and the US.
Identification strategy: to isolate exogenous variation in AI exposure for EU firms, the authors instrument EU firms' AI adoption using the adoption rates of comparable US peer firms (i.e., assigning US peers’ adoption rates to EU firms). This approach aims to capture foreign-driven technological exposure rather than endogenous domestic adoption decisions.
Outcomes analyzed: firm-level labor productivity, employment, wages, and measures of innovation; complementary investments (software, data, training) are examined as moderators.
Empirical approach: instrumental-variables framework to address endogeneity of AI adoption (details such as controls, fixed effects, and robustness checks are reported in the paper).

Implications for AI Economics

Short-run productivity gains: AI can raise output per worker principally via capital deepening rather than immediate labor substitution; models of AI impact should allow for capital complementarity and higher capital/labor ratios.
Labor-market nuance: absence of short-run firm-level job losses suggests displacement may be limited initially, but longer-run equilibrium effects (reallocation, task changes, sectoral shifts) are still an open question for research and policy.
Heterogeneous adoption effects: larger firms capture most gains, implying scale and organizational capacity matter for turning AI into productivity. This has implications for inequality across firms and potentially across workers.
Role of complementarities: software, data infrastructure, and workforce training are critical for realizing AI’s productivity potential—policy that supports these investments (especially among smaller firms) could increase diffusion and widen benefits.
Policy priorities: encourage diffusion of complementary capital and skills, monitor labor reallocation over time, and design interventions to help small firms and displaced workers capture gains from AI.

Assessment

Paper Typequasi_experimental Evidence Strengthmedium — The paper uses a large matched dataset and a plausible IV strategy that helps address endogeneity, and it reports effects on multiple outcomes (productivity, employment, wages, innovation). However, the instrument's exclusion restriction may be vulnerable to violation (US peer adoption could proxy for global demand, sectoral shocks, or other cross-border spillovers), measurement of AI adoption may be survey-based and noisy, and results are short-run and concentrated in larger firms, limiting causal certainty and external validity. Methods Rigormedium — Appropriate use of matched administrative and survey data and IV estimation demonstrates reasonable methodological care, including heterogeneity and complementarity analyses; nevertheless, potential issues remain around instrument validity, omitted confounders, measurement error in AI adoption, and limited discussion (in the summary) of robustness checks, placebo tests, or sensitivity analyses that would elevate rigor to 'high'. SampleMore than 12,000 non-financial firms in the European Union and the United States, constructed by matching EIBIS survey data (firm-level AI/digital adoption and investment questions) to ORBIS administrative accounts; analysis instruments EU firm adoption with US peer adoption rates; time period not specified in the summary. Themesproductivity labor_markets skills_training innovation adoption IdentificationInstrumental variables: EU firms' AI adoption is instrumented with the AI adoption rates of US peer firms (a peer-exposure instrument) to isolate exogenous technological exposure from endogenous firm-level adoption decisions. GeneralizabilityShort-run effects only — longer-term labor displacement or reallocation may differ, Non-financial firms only — results may not apply to financial sector or other service niches, EU and US context — may not generalize to low- and middle-income countries, Effects concentrated in medium and large firms — limited applicability to small firms, Instrument-based identification relies on cross-border peer exposure, which may not capture local adoption dynamics or industry-specific confounders

Claims (9)

Claim	Direction	Outcome	Confidence & Evidence	Details
AI adoption increases the level of labor productivity by 4%. Firm Productivity	positive	labor productivity (level)	Reading fidelity high Study strength medium	n=12000 4% 0.48
Productivity gains from AI adoption are due to capital deepening. Firm Productivity	positive	capital deepening / capital intensity (mechanism for productivity gains)	Reading fidelity high Study strength medium	n=12000 0.48
AI adoption has no adverse effects on firm-level employment. Employment	null_result	firm-level employment (headcount/employment levels)	Reading fidelity high Study strength medium	n=12000 0.48
In the short run, AI increases worker output rather than replacing labor, though longer-term effects remain uncertain. Firm Productivity	positive	worker output / productivity per worker (short run)	Reading fidelity high Study strength medium	n=12000 0.48
Productivity benefits of AI adoption are unevenly distributed and concentrate in medium and large firms. Firm Productivity	positive	distribution of productivity gains by firm size	Reading fidelity high Study strength medium	n=12000 0.48
AI-adopting firms are more innovative. Innovation Output	positive	innovative activity / innovation outcomes	Reading fidelity medium Study strength medium	n=12000 0.29
Workers in AI-adopting firms earn higher wages. Wages	positive	worker wages / earnings	Reading fidelity high Study strength medium	n=12000 0.48
Complementary investments in software and data or workforce training are critical to fully unlock the productivity gains of AI adoption. Firm Productivity	positive	productivity gains conditional on complementary investments	Reading fidelity medium Study strength medium	n=12000 0.29
The study instruments EU firms' AI adoption by assigning the adoption rates of US peers to isolate exogenous technological exposure. Adoption Rate	null_result	AI adoption (instrumental variable construction / identification strategy)	Reading fidelity high Study strength medium	n=12000 0.48