亚洲的人工智能与经济分化

AI and EconomicDivergence in Asia Natasha Che, Weining Xin, and Taichi Yoshida WP/26/166 IMF Working Papersdescribe research inprogress by the author(s) and are published toelicit comments and to encourage debate.The views expressed in IMF Working Papers arethose of the author(s) and do not necessarilyrepresent the views of the IMF, its Executive Board,or IMF management. 2026AUG IMF Working Paper Asia and Pacific Department AI and Economic Divergence in AsiaPrepared byNatasha Che(FAD),Weining Xin(APD), andTaichi Yoshida(University of Essex)* Authorized for distribution by Corinne Deléchat (APD)August2026 IMF Working Papersdescribe research in progress by the author(s) and are published to elicitcomments and to encourage debate.The views expressed in IMF Working Papers are those of theauthor(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management. ABSTRACT:Using a small open economy overlapping generations model, this paper examines how AI candrive economic divergence across and within Asian economies. While AI adoption may promise sizableproductivity gains, it could create temporary but potentially longlasting divergence across countries.Structurally-prepared advanced economies tend to adopt AI earlier and see immediate growth gains whileemerging markets and developing economies (EMDEs) face delayed adoption and initial growth headwindsfrom rising costs of capital. Structural reforms that boost productivity and strengthen human capital not onlyaccelerate adoption in EMDEs but also amplify the growth gains. Within countries, AI adoption could wideninequality along multiple dimensions: across skill groups, as high-skilled workers benefit disproportionately fromcomplementarity with the more capital-intensive technology, and across generation, as the shift of nationalincome toward capital favors asset-rich older households relative to younger workers who rely primarily onlabor income. Redistributive policies can help mitigate these distributional pressures, though they entail equity-efficiency trade-offs that vary with country-specific fiscal and demographic conditions. AI and Economic Divergence inAsia Prepared byNatasha Che,Weining Xin, andTaichi Yoshida1 1Introduction Asia’s remarkable economic success over the past few decades now faces fundamentalchallenges. The region’s potential growth is projected to decline over the medium term asdemographic dividends disappear and productivity growth stagnates (Che et al., 2025;IMF, 2024). Working-age populations across Northeast Asia are projected to shrink byover 200 million by 2050 (United Nations, 2024), while global trade tensions and reshoringpressures threaten the export-led strategies that transformed the region into the world’smanufacturing hub. Artificial Intelligence (AI) holds the promise to counter these headwinds and supportgrowth in Asia. The deep integration into global value chains of many Asian economiescreates scope for AI-enabled gains in logistics, process optimization, and smart manufac-turing. The technology also offers pathways to accelerate the transition toward highervalue-added services, potentially enabling countries to leapfrog traditional developmentstages. Most critically, AI is a potential catalyst for augmenting human capabilities andreinvigorate productivity growth, especially when demographic trends are constraininglabor supply growth across much of the region. For the purpose of the analysis, we define AI as the engine of modern automation,encompassing technologies that perform complex cognitive tasks, either standalone or askey components of modern mechanization of both manufacturing and service production.This definition includes not only generative AI such as large language models (LLMs),but also robotics and more generally physical intelligence. We focus on AI adoptionatscale—economy-wide deployment of AI that raises the capital share in the aggregateproduction function, as firms and governments invest not only in AI applications but alsoin the enabling physical infrastructure (data centers, specialized chips, cloud platforms)needed to deploy the technology broadly.1This is distinct from marginal adoption—suchas firm- or sector-specific subscriptions to language model services—which may boostfirms’ and sectors’ productivity but does not, by itself, shift factor shares or show up asmacroeconomic capital deepening. Our modeling of adoption at scale as capital deepeningwith a declining labor share is in the same spirit as Cazzaniga et al. (2024), building onDrozd et al. (2022) and Moll et al. (2022). We model AI through this capital-deepeningchannel because it is the channel through which AI bears on the questions we study:cross-country divergence, global capital demand, and the distribution of income acrosscapital, skills, and generations all operate through forces that shift aggregate factor shares. AI is a general-purpose technology that is itself embodied in capital—comp