Can We Improve Upon Preliminary Estimates of Payroll Employment Growth?
研究能否利用初步估计和同期可得信息预测美国劳工统计局就业增长数据的后续修正，发现初步估计并非修正值的有效预测，尤其是基于年度基准的修正中不可预测部分减少了22%以上。
Abstract We explore the feasibility of improving upon the preliminary estimates of payroll employment growth from the U.S. Bureau of Labor Statistics by predicting subsequent revisions to these estimates, using the preliminary estimates themselves and other information available concurrently. Results of statistical tests suggest that the preliminary estimates can be improved upon; that is, they are not "efficient forecasts" of the revised estimates. The improvement of preliminary estimates as indicators of estimates of employment growth following annual benchmarks is particularly large; the unanticipated component of the revision based on the annual benchmarks is reduced by more than 22%. KEY WORDS: Data revisionsEfficient forecastsMeasurement error