Information disclosure, polycentric governance, and green technology innovation: A machine learning-based evaluation of environmental information disclosure pilot policy

This study examines how information-based environmental governance institutions influence green technological innovation by leveraging China’s Environmental Information Disclosure (EID) pilot as a quasi-natural experiment. We conceptualize EID not merely as a transparency policy, but as a digital information management mechanism that reconfigures strategic interactions among governments, firms, and the public. Drawing on a panel dataset covering 284 prefecture-level cities (2004–2018) and A-share listed firms (2008–2021), we develop a dynamic evolutionary game model to capture how disclosure-driven governance incentives evolve across actors. Empirically, we apply Double Machine Learning (DML) to estimate the causal effect of EID on green innovation, finding that pilot cities exhibit significantly greater improvements in green patenting. We further employ Doubly Debiased LASSO and causal path analysis to uncover a triadic mechanism: increased governmental attention → administrative enforcement and media pressure → public petitions and legislative initiatives → strengthened corporate environmental responsibility. Among these, government actions exhibit the strongest marginal effect, followed by public oversight and firm-level compliance. Finally, moderation analysis reveals that the strength of local environmental legal institutions significantly conditions policy effectiveness. These findings provide new insights into how information disclosure-as a governance infrastructure-mobilizes distributed actors and enables polycentric coordination to promote sustainable innovation.