Digital infrastructure and industrial decarbonization: evidence from the dual perspectives of green innovation and digital coordination in China

Tackling climate change requires profound industrial decarbonization, where digital infrastructure is envisioned as a key enabler. However, the micro-foundations and meso‑level channels through which it operates remain underexplored. Utilizing a unique dataset of 985 Chinese listed industrial firms (2014–2022) matched with city-level data, this study investigates how digital infrastructure affects corporate carbon emission intensity. We propose and test a dual-channel mechanism: enhancing firms' green innovation capacity and fostering industry-wide digital coordination efficiency. Our extended timeframe captures firm behavior before and after the announcement of China's 'Dual Carbon' goals in 2020, offering a unique quasi-experimental setting to examine the policy's reinforcing effect. Our findings, robust to instrumental variable and quasi-natural experiment approaches, confirm that digital infrastructure significantly reduces emission intensity. The mediation analysis validates that both corporate green innovation and industry digital coordination are significant transmission channels. Furthermore, we identify a negative moderating effect of city size and notable heterogeneity across firm types and industries—most notably, a potential 'rebound effect' in technology-intensive sectors. This study makes three primary contributions. First, by shifting the analytical focus from the provincial to the city level and establishing a direct link between urban digital infrastructure and firm-level conduct, we provide more precise and granular micro-evidence that mitigates potential aggregation biases. Second, we theoretically elaborate and empirically test a dual-channel mechanism, incorporating both the micro-foundation of corporate green innovation and the meso‑level channel of industry digital coordination, thereby systematically unpacking the 'black box' of digital decarbonization. Third, methodologically, we pioneer the measurement of industry digital coordination efficiency using input-output tables, offering a novel and replicable approach for future multi-level research on digital economy externalities.