Central Bankers Who Doubt: Measuring Knowledge of the Limits of Knowledge

Abstract

This paper explores the epistemic attitudes of central bankers, focusing on their awareness of the inherent uncertainties in economic modeling and decision-making. Using natural language processing (NLP) techniques, I analyze over 5,000 speeches from the Federal Reserve (Fed), European Central Bank (ECB), and Bank of England (BoE) from 1996 to 2024. Employing topic modeling and Kullback-Leibler divergence, I construct an index measuring each speech’s alignment with a benchmark of uncertainty-aware discourse. The findings reveal divergent responses to the 2008 Great Recession: the Fed and ECB exhibited greater epistemic humility post-crisis, while the BoE displayed increased overconfidence. Female committee members across all three institutions were notably more self-assured in their speech tones than their male counterparts. This study contributes to the literature on central bank communication by quantifying shifts in epistemic attitudes, offering insights into how central bankers navigate the tension between confidence and uncertainty in response to external shocks.