Selective Corporate Social Responsibility Disclosure Under Low-Carbon Policy Pressure: A Double Machine Learning Analysis Based on New Institutional Theory and Resource Dependence Theory

Using 40,409 firm-year observations for Chinese A-share firms from 2007 to 2022, this study examines the association between city-level low-carbon policy pressure and CSR disclosure through double machine learning (DML). Departing from studies that focus primarily on discrete low-carbon policy shocks, this study examines whether increasing policy intensity reshapes firms’ broader social responsibility disclosure. New Institutional Theory (NIT) and Resource Dependence Theory (RDT) provide complementary explanations for firms’ responses to policy pressure, highlighting the trade-off between legitimacy demands and resource constraints. The results indicate that stronger low-carbon policy pressure is associated with lower CSR disclosure, and this finding remains stable across alternative specifications. Mechanism analysis identifies greater government environmental attention, lower government subsidies, and tighter financing constraints as key channels. The negative association is particularly pronounced for firms in eastern China, younger firms, and those with lower cash holdings. This study highlights a potential unintended consequence of low-carbon governance and offers an interpretation of firms’ selective responses by jointly considering institutional pressure and resource constraints.