On Processing Central Bank Communications: Can We Account for Fed Watching?

Abstract

Central bank communications need interpretation. We contribute to the communications literature by focusing on the effort expended on deciphering central bank communications. We build a model economy in which banks provide a deposit/insurance function for consumers subject to idiosyncratic liquidity shocks. Though ex ante identical, banks exhibit ex post heterogeneity by choosing different predictors that vary in terms of accuracy with respect to the expected future return on money. In the literature with heterogeneous forecasts, the modelling approach has relied on stochastic costs as the primary force accounting for the coexistence of different predictors in equilibrium. Here, model the problem as a willingness to pay for different predictors; each predictor has a different forecast accuracy with more accurate predictors resulting in higher expected utility. By the concavity of the consumer's utility function, there exists a willingness-to-pay which satisfies an indifference condition. More accurate forecast predictors correspond with greater willingness-to-pay amounts. The resources expended to obtain a more accurate forecast correspond with the bank's processing of the central bank's communications. Hence, we interpret the willingness to pay as Fed watching. The model with Fed watching exhibits local stability, while we derive conditions in which no Fed watching results in local instability. We further apply this approach to a banking economy in which the returns to one asset are subject to a fractional externality; that is, the return to one asset is negatively related to the fraction of banks holding that asset. The approach is designed to capture how herding and the regulatory settings are related to what the central bank knows (and communicates) about bank operations.