A1315
Title: Transformer-based CoVaR: Systemic risk in textual information
Authors: Weining Wang - University of Bristol (United Kingdom) [presenting]
Abstract: Conditional Value-at-Risk (CoVaR) quantifies systemic financial risk by measuring the loss quantile of one asset, conditional on another asset experiencing distress. A Transformer-based methodology is developed that integrates financial news articles directly with market data to improve CoVaR estimates. Unlike approaches that use predefined sentiment scores, the method incorporates raw text embeddings generated by a large language model (LLM). Explicit error bounds are proved for the Transformer CoVaR estimator, showing that accurate CoVaR learning is possible even with small datasets. Using U.S. market returns and Reuters news items from 2006-2013, out-of-sample results show that textual information impacts CoVaR forecasts. With better predictive performance, a pronounced negative dip is identified during market stress periods across several equity assets when comparing the Transformer-based CoVaR to both CoVaR without text and CoVaR using traditional sentiment measures. Results demonstrate that textual data can be used to effectively model systemic risk without requiring prohibitively large datasets.
