The Bank for International Settlements published a working paper analysing whether generative artificial intelligence (gen AI) has different short-run effects on real value-added growth across countries. Using data covering 56 economies and 16 industries, the paper concludes that near-term gains are likely to be larger in advanced economies than in emerging market and developing economies (EMDEs), reflecting differences in sectoral exposure to gen AI, production structures and countries’ readiness to adopt AI. The analysis applies a Rajan-and-Zingales-style approach that interacts a US-benchmarked industry exposure-to-AI measure with an IMF AI Preparedness Index (AIPI) spanning digital infrastructure, human capital, innovation and legal frameworks. Results for 2022–23 suggest that, for the same one standard deviation increase in AI preparedness, highly exposed sectors grow about two percentage points faster than low-exposure sectors; the relationship remains robust when controlling for industrial robot intensity. A country-level simulation based on sector weights estimates average predicted value-added growth increases of about 0.6 percentage points in advanced economies versus about 0.4 percentage points in EMDEs relative to the lowest-impact country (Cambodia), with large within-group dispersion; Luxembourg and Hong Kong SAR are among the highest estimated beneficiaries, partly linked to the role of finance. The paper notes that translating these value-added effects into total factor productivity measures and longer-run cross-country projections would require a structural model incorporating aggregate demand and intersectoral spillovers.

Original source View in tracker