Spatiotemporal Evolution and Nonlinear Associations of Digital Enterprise Agglomeration in China: An XGBoost‐SHAP Approach

Abstract

Based on micro‐level data for nearly five million firms, this study measures the spatial agglomeration of digital firms in 297 Chinese cities from 2000 to 2023 using the DO. GIS‐based spatial analysis and an XGBoost‐SHAP machine learning model are then used to examine its spatiotemporal evolution and associated factors. The results show that digital firm agglomeration in China increased over the study period. The agglomeration scale is 15–35 km. The spatial pattern shifted from point‐based clustering in a few southeastern coastal cities to a multi‐core agglomeration structure. Digital inclusive finance, university supply, and research and development human capital are identified as the core factors associated with digital firm agglomeration. When university supply exceeds 13.51, the interaction effect of digital inclusive finance becomes positive. When the number of research and development personnel exceeds 128,136, its positive effect increases substantially. The heterogeneity results show clear regional differences. Eastern China depends on digital inclusive finance and university supply, whereas central China depends mainly on digital inclusive finance. Western China is more strongly affected by basic urban conditions, such as the level of economic development. Higher‐tier cities depend on innovation resources, whereas lower‐tier cities rely more on digital inclusive finance. Different types of digital firms are influenced by knowledge and talent resources, user bases, and digital inclusive finance, respectively. This study provides empirical evidence for optimizing the spatial layout of digital industries and formulating differentiated policies.