The European Central Bank published Working Paper Series No 3141 by Laura Gáti and Amy Handlan modelling how a central bank communicates noisy forecasts when it faces both its own uncertainty and the public’s perception of that uncertainty, termed “reputation for confidence”. The paper argues that disagreements between the bank’s true confidence and the public’s perceived confidence create an interpretation mismatch that can make partial transparency and deliberate imprecision an optimal communication strategy. In the model, the central bank drafts an endogenous “message space” of candidate statements and releases one public announcement as an interval around its signal, with the interval width capturing precision. Point-like full revelation is possible only when the public’s reputation assessment equals the bank’s true confidence (or when the signal is zero), implying that misperceived confidence can itself generate coarse communication even without an incentive to surprise markets. Using text data from internal Federal Reserve meeting materials (including alternative drafted statements in Bluebooks and Tealbooks since 2005), Federal Open Market Committee transcripts, and newspaper-based measures of monetary policy uncertainty, the authors find patterns broadly consistent with the model: higher confidence and lower reputation are associated with more precise statements, while the Federal Reserve’s communication appears less responsive to reputation than the model predicts.