- 22 APRIL 2026 · RESEARCH BULLETIN NO. 142
Messaging to a public with its own view on central bank confidence
Central banks’ announcements attract intense attention from markets and reflect decades spent finetuning their communications (see Jansen et al., 2022). A key challenge is uncertainty. A central bank’s confidence in its view of the economy adjusts fast as new information comes in. However, public perception of a central bank’s confidence may differ from the bank’s view, resulting in markets over- or under-interpreting messages. Our research (Gáti & Handlan, 2026) finds theoretical solutions and compares them to current Federal Reserve System practices. We find that by drafting a carefully calibrated range of alternative statements the central bank can move market expectations in exactly the right way.
Central bank confidence and market reputation for confidence
In our study “confidence” is the inverse of the central bank’s uncertainty around the single most likely economic outcome.[2] The greater its uncertainty, the lower the central bank’s confidence. Its “reputation for confidence”, however, is the markets’ view of the central bank’s confidence. That reflects how the markets see the central bank’s track record, based on observing the accuracy of its past forecasts. Imagine the central bank as a doctor and the economy as a patient. Markets factor in the doctor’s reputation – built on their diagnostic track record with other patients – when deciding how seriously to take an individual diagnosis, but the doctor’s true ability and accuracy in making this diagnosis is still something that is unobservable.
The communication game set-up
To study how confidence and reputation for confidence shape optimal communication, we set up a theoretical communication game between the central bank and the markets. We assume a Phillips curve relationship between output, inflation and inflation expectations.[3] An economic shock occurs that neither the central bank nor the markets can observe directly.
The central bank gets a “noisy signal” on this shock through real-time data and judgement (based on modelling the data and discussing it at its policy meeting). The inverse of how noisy the signal is serves as the model’s measure of how confident the central bank is in its forecast. Alternatively, confidence can be interpreted as the degree of agreement between members of the policy committee at the policy meeting. The markets, meanwhile, gradually update their own beliefs about the central bank’s reputation for confidence by observing actual inflation outcomes over time.
The focus of the model is the central bank’s communication strategy. The central bank announces its assessment of the economy in order to effectively manage the expectations feeding into the Phillips curve. In our communication game, all announcements are true but the degree of precision can vary.
In our model, the central bank’s communications strategy involves drafting – and publishing, with a delay – a range of alternative statements like the alternative FOMC[4] statements of the Federal Reserve System (the “Fed”). However, our model could be applied to any central bank – all go through the process of considering alternative wordings, even if these are never made public. The drafting stage involves two dimensions of announcement precision.
The first dimension is the “span” of announcements. This captures how broad the range of possible economic situations is at the time of drafting. For example, if the central bank drafts one announcement on hyperinflation and another on hyperdeflation the span is broad, but if the announcements only disagree on whether inflation will be 10 or 20 basis points above target the span is narrow. A small span allows the central bank to prepare announcements that are more precise.
The second dimension is the “count”, which is the number of different draft announcements needed. For a given span, a larger count increases the precision of the announcements. Suppose the draft set covers the possibility of hyperinflation as well as hyperdeflation. If there are only two draft announcements, there might be one expecting inflation and another expecting deflation. But if there are five, they can specify whether the bank expects “hyperdeflation”, “some deflation”, “inflation right on target”, “some inflation” or “hyperinflation”. The higher count allows for more precise announcements.
Managing market misunderstanding
The model implies that the higher the central bank’s reputation for confidence, the more a given announcement will move market expectations. Going back to the doctor and patient example, the patient will follow a treatment plan from a doctor with a good track record more fully than if the patient had doubts about the doctor’s ability. Our main theoretical result is therefore that if the central bank’s reputation for confidence is above or below its true confidence, the market over- or underreacts to its announcements. In that case, the central bank should correct for the way the markets will “misunderstand” the announcement. If its reputation is too high, the central bank needs to draft more vague announcements to avoid a market overreaction. If reputation is too low, it needs to draft more precise announcements to increase the impact on market expectations. In short, our model suggests that when reputation is not equal to confidence, the result is central bank communication with a deliberately chosen level of precision. Notably, we find a novel justification for imprecise communication as coming from reputation being higher than confidence.
This is an important result because past studies on communication found that imprecise communication was only optimal if the two parties had sufficiently different objectives (Crawford and Sobel, 1982). This implied that imprecise announcements by central banks could only be justified if the central bank had an incentive to trick the public (Barro and Gordon 1983, Moscarini, 2007). Instead, we find that deliberate imprecision can be the optimal way for central banks to manage how the markets over- or underestimate the central bank’s degree of certainty when responding to an announcement.[5]
Model predictions
The model predicts that announcement precision increases when confidence is higher or when reputation is lower. When confidence is high, the central bank is very certain about the economic outlook and can increase announcement precision. It will primarily do so by focusing on a narrower range of likely economic outcomes in its announcement (a narrower “span”). When reputation is low, market expectations underreact to the bank’s announcements, so increasing precision is how the central bank can increase the responsiveness of market expectations. The model predicts it will do so by drafting more alternative statements to increase precision – increasing the “count” – so that each alternative statement can focus more specifically on a particular economic outcome.
Taking the theory to data
Our study tested the theory on data from internal Fed communication deliberations – the alternative FOMC statements stored in the “Tealbooks”.[6] We used the number of alternative draft statements for the “count” and measured the “span” at each meeting using a large language model. We then quantified reputation using the policy uncertainty measure from Baker et al. (2016) as the number of words referring to uncertainty in newspaper coverage of the monetary policy discussions.[7] In the data, both our measures – Fed reputation and confidence – are high. Reputation tracks confidence, suggesting that markets indeed learn about the Fed’s confidence over time. Reputation tends to dip somewhat in times of crisis, such as the Great Financial Crisis, and subsequently builds up again.
We found that the model’s predictions are largely borne out by the data. Announcement precision is higher when confidence is high or reputation low. Also, confidence is more strongly associated with the span of alternative drafts and reputation more so with the count. It appears that the Fed does indeed adjust announcement precision to compensate for the markets factoring in its reputation for confidence. However, we also discovered that in the data the Fed’s communication responds less to reputation than our model would predict.
A new rationale for deliberate imprecision
Our study uncovers a new rationale for deliberate imprecision in central bank communication: optimizing the precision of announcements to counter anticipated misunderstanding by the markets. This need arises whenever the markets over- or underestimate the central bank’s confidence about the economic outlook. We find that the more certain the central bank is, or the less certain the markets perceive it to be, the more precise the central bank’s announcements are. Imagine being diagnosed by a doctor who has seen a clear test result, but has mixed reviews from past patients – they must choose their words carefully to offset your doubts about them. Adapting the level of precision in announcements can be a useful element in central banks’ communication strategies. While beyond the scope of our paper, the extent to which the ECB engages in similar behavior is an important empirical question.
References
Baker, S., Bloom, N. and Davis, S. (2016), “Measuring Economic Policy Uncertainty”, The Quarterly Journal of Economics, Vol. 131, No 4, pages 1593-1636.
Barro, R. and Gordon, D. (1983), “Rules, discretion and reputation in a model of monetary policy”, Journal of Monetary Economics, Vol. 12, pp. 101-121.
Crawford, V. and Sobel, J. (1982), “Strategic Information Transmission”, Econometrica, Vol. 50. No 6, pp. 1431-1451.
Gáti, L. and Handlan, A., (2026), “Reputation for Confidence”, Working Paper Series, No 3141, ECB, March.
Jansen, D.-J., de Haan, J., Ehrmann, M. and Blinder, A. (2022), “Central bank communication with the general public: Challenging, yet worthwhile”, VoxEU.org, 6 October.
Moscarini, G. (2007), “Competence Implies Credibility”, American Economic Review, Vol. 97, No 1, pp. 37-63.
This article was written by Laura Gáti (Directorate General Research, European Central Bank, and CEPR) and Amy Handlan (Brown University). The authors gratefully acknowledge the comments of Alex Popov and Zoë Sprokel. The views expressed here are those of the authors and do not necessarily represent the views of the European Central Bank or the Eurosystem.
If the central bank’s uncertainty about the probability of the single most likely economic outcome in its forecasts is x, then “confidence” is 1/x
Such a relationship means, first, that prices and economic activity move together and, second, that how people expect prices to change is a factor in how they actually do change.
FOMC stands for the Federal Open Markets Committee, which takes decisions on the Fed’s open market operations, thereby setting the Federal Reserve System’s monetary policy
A subtle point is that having a reputation higher than its true confidence is not a good thing from the point of view of the central bank, because it means that the markets overreact to central bank announcements. Therefore, a vague announcement today, by lowering reputation tomorrow, actually corrects a market misperception. A richer model framework could speak to how to jointly manage announcement precision and the evolution of reputation over time.
The Tealbooks contain internal FOMC meeting materials, such as the alternative statement drafts, and are released to the public with a five-year lag.
For confidence, we apply the Baker et al. (2016) methodology to the FOMC transcripts, effectively measuring the frequency of uncertainty words in its internal policy discussions. Using these variables, we run regressions of span or count on reputation and confidence, with various controls.