Topics of green credit policy in China based on text mining and their impact on the stock returns of heavily polluting enterprises

Green credit policy (GCP) is an environmental regulatory policy that promotes economic greening. Existing studies on GCP seldom focus on the policy text itself, but most construct quasi natural experiments to explore the effects of GCP. This paper explores the topics and characteristics of numerous GCP documents and their impact on the stock returns of heavily polluting enterprises by combining the Latent Dirichlet Allocation (LDA) topic model and the econometric method. The results show that GCP predominantly addresses four topic areas: water and soil protection, energy conservation and emissions reduction, high-quality development of industries, and standardized development of green credit. There is spatiotemporal heterogeneity in the prevalence of the total topic and various topics. Regression analyses demonstrate a significant negative and lagged impact of the total topic prevalence on stock returns of heavily polluting enterprises, with obvious spatiotemporal heterogeneity. Specifically, except for the high-quality development of industries topic, all other three topics exert significant negative effects on stock returns. Finally, this study contributes to policymakers' understanding of policy directions, identifies policy gaps, offers insights for optimizing financial resource allocation, and provides historical experience for improving the green financial system. Moreover, these findings assist heavily polluting enterprises in understanding policy intentions, managing their share prices, and adjusting future development strategies to cope with policy shocks.