The 'picking the fittest' approach and spatial dynamics in China's artificial intelligence regional development

Datasets for the paper: Barabuffi, S., Cricchio, J., & Di Minin, A. (2025). The 'Picking the Fittest' Approach and Spatial Dynamics in China’s Artificial Intelligence Regional Development. Papers in Regional Science, 100096. Abstract: This paper investigates the role of regional specialization in ICT in fostering AI patenting performance and inter-regional spatial spillovers across China’s provincial-level regions. Using panel fixed effects estimators, a Spatial Autoregressive Regression model and by adapting the technological frontier IV strategy on a comprehensive database covering 2006–2021, we find that positively selecting areas where regional ICT specialization is leveraged – the “picking the fittest” approach – can increase AI patenting performance while exacerbating regional disparities. Furthermore, we find that geographical proximity to developed AI regions impedes AI patenting progress in neighboring areas. The findings highlight the need for collaborative regional strategies and urge policy-makers to achieve a balance between strengthening regional specialization and promoting cooperation. Dataset variables: Description Abbreviation Sources Log of total AI-related patents per million population AIpat/Pop Orbis Intellectual Property Database, China Statistical Yearbook 3 years MA Revelead Technological Advantage in ICT RTA ICT Orbis Intellectual Property Database, Annual Report of Patent Statistics Log of High-tech industry R&D expenditures on GDP HTRDexp/GDP China Statistical Yearbook on Science and Technology, China Statistical Yearbook Log of High-tech industry R&D personnel on total employment HTRDpers/Emp China Statistical Yearbook on Science and Technology, China Statistical Yearbook Log of GDP per capita GDP/Pop China Statistical Yearbook Log of FDI on GDP FDI/GDP China Statistical Yearbook spatialinvdist: Quadratic inverse distance matrix calculated from the latitude and longitude of each region’s capital where the general element indicates that region i is spatially connected to region j.