Climate risk and financial stability research: a computational bibliometric analysis of thematic framing (2010–2025)

Abstract

Research on climate risk and financial stability has grown rapidly, yet few studies jointly address its bibliometric structure, thematic framing (risk versus solution), and underlying data reliability. To address this gap, this study analyzes over 4,100 relevant English-language articles from the Dimensions database (2010–2025). By combining a domain lexicon, a manual audit, and machine learning models (SVM, ClimateBERT, FinBERT), we classify the abstracts' thematic orientation. Our manual audit reveals substantial noise in the raw data, highlighting the necessity of strict relevance filtering. Methodologically, while lexicon-based classification shows weak agreement with human coding, machine learning models deliver robust performance. Empirically, we find no clear temporal trend in the share of solution-oriented research. However, regression models demonstrate that solution-oriented papers receive a significant citation premium (approximately 27% higher expected citations), an association that remains robust across stricter filters and journal fixed effects. These findings emphasize the crucial role of data curation in computational bibliometrics and suggest that solution-focused research attracts greater academic attention.

JEL Classification : C38, C45, G10, G18, Q54

Paper type : Empirical quantitative research (computational bibliometrics)