The multifaceted impact of proximity on New Quality Productive Forces

This paper selects a sample of panel data for 285 Chinese cities spanning from 2009 to 2022, combined with annual report data from 4,183 key listed companies. It employs the Word2Vec and TF-IDF machine learning models to generate the five core explanatory variables of "proximity": cognitive, organizational, social, institutional, and geographical. Building on this, a fixed effects model is constructed to identify the mechanism by which New Quality Productive Forces (NQPF) is formed under different proximity dimensions, as well as the resulting regional heterogeneity. Furthermore, the formation mechanism of NQPF is further analyzed and revealed based on the threshold effect model and the spillover effects among the different proximity dimensions, thus providing practical pathways for the region-specific development of NQPF. The results show that geographical, social, and cognitive proximity generally promote NQPF, whereas institutional proximity exerts a persistent inhibitory effect and organizational proximity is more contingent. The threshold analysis indicates clear non-linearities: geographical proximity becomes beneficial only beyond a higher intensity, social proximity weakens its negative effect as it increases, and institutional proximity imposes stronger constraints at higher levels. Interaction results further show that spillovers are heterogeneous, with geographical proximity substituting for social and organizational proximity, while social–institutional, institutional–organizational, institutional–cognitive, and organizational–cognitive pairings generate complementary effects. Overall, NQPF upgrading depends less on selective support for leading sectors than on potential excavation, factor coordination, and policy synergy tailored to regional conditions.