The European Central Bank published research by Laura Gáti and Amy Handlan that models how central banks should calibrate the precision of their communications when markets’ perceived “reputation for confidence” diverges from the bank’s actual uncertainty about the economic outlook. The paper argues that even when announcements are truthful, optimal communication may involve deliberate imprecision or extra precision to prevent markets from overreacting or underreacting. In the model, announcement precision is determined by both the “span” of alternative draft statements (how wide the range of contemplated outcomes is) and the “count” (how many alternatives are prepared). If reputation is higher than true confidence, markets would overreact, so the optimal response is to communicate more vaguely; if reputation is lower than confidence, markets would underreact, so the bank should communicate more precisely. The authors test these predictions using Federal Reserve System alternative FOMC statements from the Tealbooks, measuring “count” directly, “span” using a large language model, and reputation using a policy uncertainty-based proxy drawn from newspaper coverage. The results broadly align with the theory: precision increases when confidence is higher or reputation is lower, with confidence more closely linked to narrowing span and reputation more closely linked to increasing count, although the Fed’s communications appear less responsive to reputation than the model predicts. The paper notes that whether the European Central Bank exhibits similar behaviour is an open empirical question beyond the study’s scope.