Authority, Attention, and the Origins of Monetary Policy Surprises (Banque de France)

Summary

• A speech by the Chair or Vice Chair draws about 65 distinct Reuters articles, whereas a speech by another Board governor draws about 10.p.2
• The compiled corpus contains 4,177 speeches and congressional testimonies delivered between 1961 and 2024 by 66 Federal Reserve Board members.p.8
• Non-Chair speech tone jointly with Chair tone explains 18 percent of the variation in one-year-ahead interest-rate forecast errors, with the predictive content sitting entirely in the non-Chair term.p.4
• Upon promotion from governor to Chair or Vice Chair, media coverage of the same speaker's speeches jumps from about 14 to about 63 articles.p.2
• FOMC transcripts indicate non-Chairs warrant about 54 percent of attention based on decision influence, but they receive only about 40 percent from markets.p.2

Key insights

1. Attention paid to Federal Reserve communication follows institutional authority rather than informational content.p.4
2. The speech tone of non-Chair Board governors predicts consensus interest-rate forecast errors, while the Chair's tone carries no comparable predictive content.p.4
3. The asymmetry in predictive content belongs to the office rather than the personal identity or style of the speaker.p.5
4. Interest-rate forecast errors by professional forecasters correlate directly with FOMC deviations from its empirical Taylor rule.p.24
5. Underweighting non-Chair public communications causes a measurable portion of monetary policy announcement surprises, specifically in forward guidance.p.5
6. Forecasters devote 60 percent of attention to the Chair, matching an implicit assumption that the Chair is far more influential than transcripts demonstrate.p.5
7. Single-governor speech signals predict forecast errors about as well as multi-governor averaged signals, showing the discount is not a statistical averaging artifact.p.23