Decoding Policy Directives: An Empirical Study on the Influence of Central Policy Signals on Local Environmental Regulation Behavior

Central policy texts are crucial instruments utilized by the central government for communicating policy directions, guiding behavior, and administering societal affairs. Analyzing central policy texts helps the public to understand policies and attribute real-world issues. Prior studies draw conclusions mainly based on statistics of word frequency while lacking consideration of the semantics content of these texts. This study employs text mining techniques, specifically Whole Word Masking (WWM)-BERT, to analyze 348 central government documents from 2016 to 2018, which enables the extraction of latent semantic meanings from the policy texts. Aligned with signaling theory, our research investigates the roots of environmental policy implementation disparities, examining aspects such as intensity, clarity, and credibility/reliability of communication. The findings indicate that document signals, such as issuing department, wording intensity, and clarified regulatory targets of policy documents significantly enhance local governments’ environmental regulation behaviors. Meanwhile, the absence of standardized benchmarks for ecological environmental administrative penalties may exert negative impact on the effectiveness of credible threats in shaping environmental regulation outcomes.