Bank for International Settlements working paper finds AI adoption lifts European firms’ labour productivity by 4% without short-run job losses

The Bank for International Settlements has published a working paper on artificial intelligence adoption in European non-financial firms, concluding that adopting AI raises labour productivity by around 4% and does not reduce firm-level employment in the short run. The results attribute the productivity gain to capital deepening rather than labour displacement, with benefits concentrated in medium and large firms and accompanied by higher wages at AI-adopting firms. The analysis uses matched European Investment Bank Investment Survey (EIBIS) and Moody’s ORBIS data for more than 12,000 firms in the European Union and 800 in the United States from 2019 to 2024, defining AI adopters as firms that use big data analytics and AI in at least parts of the business. To address endogeneity, EU firms’ AI adoption is instrumented by assigning the adoption rates of comparable US peers via stratified propensity score matching. The paper reports that AI adoption is more common among large firms (45% with more than 250 employees) than small firms (24% with 10 to 49 employees), and finds no statistically significant employment effect once adoption is instrumented and firm controls are included; average wages per employee and total wage bills are higher at AI adopters. Productivity effects are stronger for medium and large firms, and larger where firms make complementary investments, particularly in software and data or employee training.