Artificial intelligence, global value chain position and manufacturing firm emissions

Utilizing firm-level data from 2000 to 2012, sourced from the Annual Survey of Industrial Firms, China’s Environmental Statistics Database, the International Federation of Robotics, and China Customs Trade Statistics, we estimate the effects and mechanisms of artificial intelligence (AI) on firms’ sulfur dioxide (SO2) emissions. Our analysis reveals that AI significantly reduces firms’ SO2 emissions, and this result remains robust to extensive checks and an instrumental variable approach addressing endogeneity. Furthermore, AI decreases firms’ SO2 emissions through three channels: enhancing energy efficiency, optimizing supply chain management, and strengthening pollution abatement capacity. Additionally, heterogeneity analysis indicates a more pronounced reduction in SO2 emissions for capital-intensive and emission-intensive industries, as well as for firms that are more productive, larger, older, and located in eastern regions. Finally, our analysis yields the valuable insight that firms located more upstream in the global value chain accrue more substantial environmental benefits from AI adoption, thereby helping to mitigate the global issue arising from environmental bias in trade policies. Overall, the study underscores AI’s potential to reduce firms’ SO2 emissions and contributes to the literature on the environmental impacts of digital technology.