Using Density Forecasts to Measure the Stability of Inflation Expectations

The anchoring of inflation expectations is a major objective of modern monetary policy. This paper estimates variation over time in the sensitivity of inflation expectations to news. The function describing the response of an agent’s expectations to news is fundamentally unobservable since on a given date we only see the single realization of news that actually occurred. However, under the assumption that agents apply Bayes’ rule and that they believe their signals are Gaussian (hence allowing for a wide range of behavioral biases), the marginal response of expectations to signals is proportional to agents’ uncertainty about future inflation. Empirically, both in the time-series and the cross-section, reported uncertainty both contemporaneously explains and also predicts the future sensitivity of expectations to news. The results imply that as of 2025, inflation expectations are 2–3 times more sensitive to news than prior to 2020.