Authority, Attention, and the Origins of Monetary Policy Surprises

Press coverage of the same speaker rises sharply upon promotion to Chair or Vice Chair

Non-Technical Summary

Between meetings, Federal Reserve policymakers speak almost continuously. By the time a decision is announced, much of the relevant information is already public. Yet announcements still move asset prices substantially. Why does monetary policy remain surprising when so much has already been said?

The answer lies less in what information is available than in which information markets choose to process. Because attention is scarce, agents extracting market signals from a central bank's communication allocate their attention according to the hierarchical rank of the speaker. Using Federal Reserve Board members' communications since 1961, the Reuters news archive, and survey and market measures of interest-rate expectations, I show that attention follows institutional authority rather than measured influence, and that the communications markets discount most carry the information they turn out to have missed.

Attention is concentrated at the top. A speech by the Chair or Vice Chair draws about 65 distinct Reuters articles; a speech by another Board governor draws about 10. The public follows suit: Google searches for a governor's name barely move when they speak and jump when the Chair does. Non-Chair speeches also move market prices far less on the day they are delivered.

Yet the overlooked speeches are the informative ones. I score each speech for how hawkish or dovish it sounds, using a fixed list of policy terms. The tone of non-Chair speeches predicts the errors that professional forecasters and futures markets make about the path of the federal funds rate, and the effect grows with the horizon. It also predicts the forward-guidance component of the next announcement surprise, even after accounting for the public news available at the time. The Chair's tone predicts none of these.

The asymmetry belongs to the office, not the person. Seven Board members were promoted from governor to Chair or Vice Chair during the years the Reuters archive covers, so the same speakers are observed in both roles. On promotion, coverage of their speeches jumps from about 14 to about 63 articles (Figure 1). At the same time, their words stop predicting: before promotion, their tone forecasts consensus errors at every horizon; afterwards, it forecasts nothing. Their expressed stance does not change across the transition. What changes is the attention they command.

Could this allocation nonetheless be rational? The Chair does, after all, set the agenda. FOMC transcripts allow a direct test: they reveal how much each member's stance moves the decision and how variable that stance is. By this benchmark, non-Chairs warrant about 54 percent of attention. They receive about 40. Markets behave as if the average governor carried less than half the weight in decisions that the transcripts assign them.

Two implications follow. First, high-frequency announcement surprises are widely used as exogenous monetary policy shocks, on the grounds that they cannot be forecast from prior public information. A measurable part of the surprise concentrated in its forward-guidance dimension, is in fact predictable from speeches delivered months earlier. Their exogeneity is a property of how markets allocate attention, not of the information environment.

Second, for central banks, the reach of communication depends on who is heard as well as on what is said. A decentralized committee supplies a stream of individually informative signals, but the price system absorbs mainly those carrying the right institutional label. More communication from under-attended members is not, by itself, more transparency.

Keywords: Central Bank Communication, Monetary Policy Surprises, Limited Attention, Forward Guidance, Textual Analysis, Federal Reserve
Codes JEL : D84, E58, G14