ECB Working Paper Links Inflation News Narratives to Inflation Risk Premia in Euro Area & US

Key Takeaways
- An ECB staff working paper (No 3288) finds that inflation risk premia move inversely to the balance of demand-driven versus supply-driven inflation narratives in the Financial Times and Wall Street Journal.
- The paper is research, not regulation, and imposes no obligations.
- It does indicate how market participants may price inflation risk, which is relevant to market risk, model risk, and stress-testing teams.
Deep Dive
An ECB staff working paper (No 3288) finds that inflation risk premia move inversely to the balance of demand-driven versus supply-driven inflation narratives in the Financial Times and Wall Street Journal. The paper is research, not regulation, and imposes no obligations. It does indicate how market participants may price inflation risk, which is relevant to market risk, model risk, and stress-testing teams.
The authors build a measure called NetDemand, the monthly difference between the number of articles attributing inflation to demand factors and the number attributing it to supply factors. They derive it with Causality Extraction applied to Financial Times (Euro Area) and Wall Street Journal (US) articles from 2005 to 2024. Inflation risk premia at the two-, five-, and ten-year maturities are inversely related to NetDemand: they are higher when inflation is narrated as supply-driven and lower when it is narrated as demand-driven. The relationship holds after controlling for composite PMI and VIX. It strengthens with risk aversion in the US and with inflation volatility in the Euro Area, though only at short and medium maturities. It is not subsumed by macro-based measures, professional forecaster surveys, or LLM-derived narratives. The authors also report an F-score of 79% for the Financial Times narrative extraction, and note that the news-based measure is available daily.
The paper suggests that shifts in media narratives about inflation drivers, especially commodity and energy supply stories, can help explain movements in inflation risk premia and therefore nominal yields. This is relevant to market risk, interest rate risk in the banking book, and valuation of inflation-linked instruments. It also supports scenario design in stress testing, where supply-shock and demand-shock inflation scenarios imply different risk premium behavior. Limitations apply: the findings show association, not causation. The US premia estimates rely on survey-based adjustments, and the Euro Area estimates rely on an affine term structure model, so both carry model uncertainty. The paper also notes possible indexation-lag distortions at shorter US maturities.
Recommended action:L
- Circulate the paper to market risk, treasury, ALM, and stress-testing owners for awareness.
- Consider whether scenario libraries distinguish supply-driven from demand-driven inflation, with differing risk premia and yield-curve responses.
- If the institution is considering news-based or NLP indicators, apply existing model risk management and data governance standards. These include validation, documentation of data sourcing and licensing, and monitoring of classification accuracy.
- Note the paper's caution that LLMs can misread non-causal relations as causal. This is relevant to any internal use of generative AI for narrative extraction.
- No regulatory filing or policy change is required at this time. Monitor for any later ECB supervisory communications that cite this work.
Direct compliance risk is low. Indirect market and model risk relevance is moderate, mainly for inflation-sensitive portfolios and for firms developing alternative-data signals.
Source: European Central Bank

