Can National AI Pilot Zones Improve Corporate Carbon Performance? Evidence from China

Abstract This study utilizes the National AI Innovation Pilot Zone policy as a quasi-natural experiment to investigate how AI policies influence corporate carbon performance. By analyzing panel data from Chinese listed companies spanning 2007 to 2023 and applying difference-in-differences (DID) as well as triple-difference (DDD) models, we present comprehensive evidence regarding the effects of the policy and the mechanisms that underlie them. The findings indicate that the policy regarding AI pilots greatly enhances the carbon performance of corporations, and these results hold steady across various identification approaches. Analysis of the mechanism indicates that the policy functions via two primary channels: proactive innovation, which is demonstrated by a rise in green invention patents, and optimization of efficiency, indicated by enhancements in green total factor energy efficiency. Further analysis indicates that the policy effect is strengthened by supply chain coordination, industry digitalization, human capital, and government environmental attention, while it is weakened by higher capacity utilization. Moreover, we document substantial ownership heterogeneity: the policy effect is significantly stronger for state-owned enterprises (SOEs), with innovation channel being primarily driven by SOEs, whereas private firms rely more on government environmental attention and supply chain concentration. Overall, this study uncovers a dual-channel mechanism through which AI policies facilitate green transformation and highlights the critical role of institutional context in shaping policy effectiveness. The findings provide new micro-level evidence on the environmental implications of digital transformation and offer important policy insights for promoting low-carbon development.