Reconsidering the Optimality of Federal Reserve Forecasts

We investigate the econometric properties of the Federal Reserve Greenbook forecasts with an integrated real-time database of U.S. macroeconomic data that more precisely characterizes the information sets available to Fed policy makers in advance of the Open Market Committee meetings. Our data set associates historical vintages of NIPA and labor market data with the exact dates of the Greenbook forecasts, allowing us to examine multivariate relationships among the real-time forecast errors for inflation, output growth and the unemployment rate. For most forecast horizons and subsamples we examine, the hypotheses of unbiasedness and efficiency can be individually and jointly rejected. Our results suggest that shifts in lower-frequency components of the forecasted series are an important source for the rejection of forecast rationality. JEL Categories: C53, C82, E58.