The relationship between two sets of GNP data, the earliest and the fi
rst revisions, is examined. This comparison enables us to determine wh
ether the early numbers are valuable to forecasters. The analysis uses
two methods. The first is based on a technique which has been used to
evaluate whether financial and economic forecasts are valuable to the
users. The second approach uses time series techniques. We conclude t
hat revisions to the 15-day estimates that occur when the 45-day numbe
rs are released are successful and that the 15-day numbers are useful
to forecasters.