Testing the value of lead information in forecasting monthly changes in employment from the Bureau of Labor Statistics

This article examines the value of lead information by investigating the predictive power the automatic data processing (ADP) report has on nonfarm payroll employment data released by the Bureau of Labor Statistics (BLS) 2 days after the ADP. We find that updating a vector autoregression (VAR) forecast with the ADP data improves the forecast accuracy relative to a standard VAR forecast. However, this informational advantage disappears if real-time comparisons are made with the Bloomberg consensus forecasts of the BLS which are available prior to the ADP. We explore the confounding effects of data revisions and the potential pitfalls in testing the value of lead information based on the accumulated historical data.