人工智能的采用、生产力与就业：来自欧洲公司的证据

by Iñaki Aldasoro, Leonardo Gambacorta, Rozalia Pal,Debora Revoltella, Christoph Weiss and Marcin Wolfski Monetary and Economic Department January 2026 Keywords:artificial intelligence,firm productivity,Europe, digital transformation BISWorking Papers are written by members of the Monetary and EconomicDepartment of the Bank for International Settlements, and from time to time by othereconomists, and are published by the Bank. The papers are on subjects of topical This publication is available on the BIS website (www.bis.org). ©Bank for International Settlements 2026. All rights reserved. Brief excerpts may be ISSN 1020-0959 (print)ISSN 1682-7678 (online) AI adoption, productivity and employment:Evidence from European firms I˜naki AldasoroBISRozalia PalEIBDebora RevoltellaEIBMarcin WolskiEIBLeonardoGambacortaBIS&CEPRChristophWeissEIB Abstract This paper provides new evidence on how the adoption of artificial intelligence (AI)affects productivity and employment in Europe. Using matched EIBIS-ORBIS dataon more than 12,000 non-financial firms in the European Union (EU) and UnitedStates (US), we instrument the adoption of AI by EU firms by assigning the adop-tion rates of US peers to isolate exogenous technological exposure.Our resultsshow that AI adoption increases the level of labor productivity by 4%. Productiv-ity gains are due to capital deepening, as we find no adverse effects on firm-levelemployment.This suggests that AI increases worker output rather than replac- 1Introduction Artificial intelligence (AI) is poised to transform the global economy by simultaneouslyaffecting aggregate demand and productivity on the supply side.Yet robust evidenceon its firm-level effects outside the United States (US) remains scarce. While optimisticprojections predict that AI could drive productivity booms (Baily, Brynjolfsson, and Korinek, 2023), others caution that adoption barriers and productivity constraints fromunbalanced growth may significantly limit the gains from AI adoption (Acemoglu, 2024;Filippucci, Gal, and Schief, 2024). Europe’s position is both central and paradoxical: it This paper provides the first causal evidence on how AI adoption affects productivityand employment in European firms, leveraging a unique survey on EU and US non-financial firms matched to administrative financial data. We address two core questions.First, we explore the firm characteristics that predict AI adoption in Europe, which is The analysis reveals four key insights. First, AI adoption increases labor productivityby 4% on average after addressing endogeneity.Second, the gains stem from capitaldeepening rather than job displacement.Third, benefits are concentrated in mediumand large firms – posing risks of widening inequality in the benefits from AI adoption. We leverage data of the European Investment Bank Investment Survey (EIBIS), anannual survey covering approximately 12,000 non-financial firms across the EU and 800US non-financial firms, matched with ORBIS data on corporate financial statements. Our country (Brutscher et al., 2020), while its detailed questions identify AI adoption based onwhether firms use big data analytics and AI technologies in their operations. Crucially,EIBIS also captures granular firm characteristics on investment behavior, investment Three stylized facts emerge from the data, informing our approach for causal iden-tification of the effects of AI. First, adoption is highly stratified:45% of large firms (more than 250 employees) deploy AI, compared to only 24% of small firms (10 to 49employees).1 This echoes technology diffusion theories (Comin and Hobijn, 2010). Sec-ond, emerging EU economies (e.g., Romania, 22%) exhibit adoption rates substantially To establish causality, we develop a novel instrumental variable (IV) strategy inspiredby the work on financial dependence and growth following Rajan and Zingales (1998). Wematch each EU firm with comparable US firms (same sector and size) that exhibit similarlevels of innovation intensity, investment, managerial practices and external finance. Wethen assign the US firms’ AI adoption rate as an exogenous proxy for EU firms’ adoption Our results show a robust causal link between AI adoption and productivity.Con-trolling for a wide set of country, sector and year fixed effects and observable financialvariables, we find that firms adopting AI have 4% higher labor productivity. This effectis economically significant and aligns with mid-range macroeconomic projections recentlyreported in the literature (Acemoglu, 2024; Bergeaud, 2024), rather than with optimistic AI-adopting firms have so far benefited from higher wages.The productivity benefitsare unevenly distributed, with medium and larger firms experiencing significantly higherproductivity gains than their smaller peers.The stratification highlights concerns that AI could exacerbate income gaps (Cazzaniga et al., 2024; Cornelli, Frost, and Mishra, The findings carry significant imp