Testing for forecast consensus

A panel of forecasts may be defined to be in consensus when individual fore casters place identical weights on a common latent variable. We suggest thi s definition and formulate a dynamic latent-variable model to test for cons ensus. This method also tests whether it is valid to use the mean forecast as the consensus forecast. In applications to surveys of U.S. macroeconomic forecasters, there is greater consensus in forecasts for output growth tha n for inflation or unemployment, but idiosyncratic forecast autocorrelation from year to year is present for most forecasters.