Abstract: As a vital driving force of the digital economy, e-commerce is highly aligned with China’s strategic needs for achieving the “Dual Carbon” goal and high-quality economic development. Based on the panel data from 267 prefecture level cities in China from 2006 to 2022, and taking the construction of national e-commerce demonstration cities as a quasi natural experiment, this paper systematically explores the impact of the construction of the demonstration cities on carbon emission intensity by adopting the dual machine learning method and taking the environmental regulation as an external key variable, then it reveals its nonlinear characteristics through a panel threshold model. The findings show that the construction of national e-commerce demonstration cities has significantly reduced carbon emission intensity, and its policy effects show consistency under various robustness tests. The mechanism analysis shows that the construction of the demonstration cities achieves a reduction in carbon emission intensity through structural upgrading effects, green technology innovation effects, and environmental regulation effects. Further analysis shows that the impact of the construction of the demonstration cities on carbon emission intensity exhibits a non-linear characteristic of an inverted U-shape at different stages of development and resource endowment conditions. Therefore, it is recommended to strengthen the synergy of top-level design, implement differentiated policy paths, deepen the linkage mechanism between environmental regulation and technological innovation, so as to fully leverage the engine role of e-commerce in low-carbon transformation.

Key words: national E-commerce demonstration cities, carbon emission intensity, double machine learning, impact mechanism, threshold effect