Sentiment analysis with large language models for credit risk assessment

16 September 2026

Banca d'Italia today publishes 'Sentiment analysis with large language models for credit risk assessment', the new issue of the series 'Markets, infrastructures, payment systems'.

Recent advancements in natural language processing techniques have significantly expanded the possibilities for integrating unstructured textual data into quantitative analyses. In this paper, large language models (LLMs) are used to construct credit sentiment indicators based on financial news sourced from the Dow Jones Factiva database. These indicators are then integrated into Banca d'Italia's In-house Credit Assessment System (ICAS), which is part of the Eurosystem Credit Assessment Framework (ECAF). The results show that LLM-driven sentiment analysis enhances the ability to distinguish solvent firms from insolvent ones, offering a robust approach to incorporating unstructured textual data into credit risk evaluation.