Digital Infrastructure Construction and Energy Efficiency: A Quasi-Natural Experiment with Double Machine Learning

As China advances toward its “dual carbon” objectives, enhancing energy efficiency (EE) has become a strategic imperative. In this context, digital infrastructure construction (DIC) acts as a key driver of EE improvements. The effect of DIC on EE is assessed using China’s “Broadband China” policy (BCP) as a quasi-natural experimental setting. A double machine learning approach is applied, drawing on data from 281 cities spanning 2011 to 2021. The results demonstrate that DIC significantly enhances EE by fostering industrial agglomeration, mitigating resource misallocation, and promoting green technological innovation. Heterogeneity analysis indicates that such effects are more pronounced in the eastern and central regions, particularly in resource-based cities and those with an old industrial base. Furthermore, DIC is shown to enhance corporate-level EE. This study highlights the contribution of DIC to EE, elucidates its transmission mechanisms, extends the environmental perspective of the BCP, and provides valuable implications for policy-making.