Abstract

Central Banks regularly publish fancharts of macroeconomic variables, communicating forecasts for several horizons. Although fancharts contains three types of information: point forecasts (the path), likelihood of the path (bands around it), and variable’s dynamics across horizons; the latter is neglected by existing absolute evaluation approaches. Practitioners indeed evaluate the calibration of fancharts testing the forecast accuracy horizon by horizon, not considering any time-dependency among them. This paper describes fancharts as density path forecasts, discusses the impact of time-dependence in evaluation and proposes calibration tests to assess whether or not Central Banks publishes reliable forecasts.