Can central bank digital currency pilots enhance urban energy efficiency? Evidence from machine learning

Amidst surging global energy demand and intensifying carbon mitigation pressures, the digitalization wave presents new opportunities for energy revolution realization. However, limited research has focused on whether central bank digital currency (CBDC), a disruptive digital financial innovation, can enhance urban energy efficiency through its technical characteristics and spillover effects. Utilizing data from 286 Chinese cities spanning 2016–2023, and employing a three-stage super-efficiency slacks-based measure model combined with double machine learning and causal forest methods, this study finds that: CBDC pilots significantly improve urban energy efficiency, with indirect effects primarily transmitted through three pathways — digitalization level, fiscal support effectiveness, and openness degree — while policy effects are influenced by financial development depth and industrial structure, exhibiting a ”double threshold effect”. These findings enrich the theoretical foundation at the intersection of digital finance and energy economics, advance machine learning applications in policy evaluation, and provide empirical evidence for formulating differentiated CBDC implementation and energy governance strategies.