The role of sentometrics in analysing financial instability

The paper traces the evolution of approaches to the analysis of financial stability, ranging from classical prudential indicators and aggregate financial stress indices to sentiment-based indices capturing the mood and expectations of economic agents. Based on the contemporary literature, the paper provides a systematic review of the key methodological issues related to sentometric approach, namely the construction of specialised dictionaries and the aggregation of textual information into a single index. The paper discusses the trade-offs between the interpretability of dictionary-based methods and the flexibility of models based on machine learning and large language models. It is further emphasised that the choice of text corpus and data processing method has a substantial impact on the interpretation and predictive properties of the resulting indicators. Particular attention is paid to a review of empirical studies based on a thematic graph of articles (Connected Papers), which makes possible to identify the main clusters of work at the intersection of sentometrics, financial stability and macroprudential policy. In the concluding part, the paper discusses the limitations and weak points of sentometric indicators, including problems of interpretation, comparability and data quality, and formulates directions for further research required to integrate sentometric indices into the existing standard frameworks for monitoring financial stability.