Eurocentral bank (ECB) has applied three AI methods—dictionary counting, FinBERT, and a GPT‑based approach—to analyze the language of its 43 Financial Stability Reviews (FSRs) from 2004 to 2025. The three techniques produced similar cyclical patterns, but the GPT filter captured sharper negative peaks during crisis episodes by focusing on sentences that contain explicit risk judgments. The study shows that sentiment trends align closely with systemic stress indicators and that the tone of recent FSRs has become more concise, assessment‑bearing and forward‑looking. The findings suggest that AI‑derived sentiment can support risk monitoring and drafting consistency, complementing expert judgment rather than replacing it.
© European Central Bank, 2025.
Summary derived from the ECB website (https://www.ecb.europa.eu ).
Made by AI. If you spot anything of concern write us at contact@cybach.com. We’ll promptly correct irregularities.