U.S. flag

An official website of the United States government

Department of Labor Logo United States Department of Labor
Dot gov

The .gov means it's official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Bureau of Labor Statistics > Office of Survey Methods Research > Publications > Browse Research Papers

How large are long-term revisions to estimates of U.S. labor productivity growth?

John Glaser, Peter B. Meyer, Jay Stewart, and Jerin Varghese

Abstract

This paper examines long-term revisions to official estimates of quarterly US labor productivity growth and to its components, output and hours growth, for 2000-2015. Estimates of output (GDP) and hours growth are revised substantially in the first months after the reference quarter. The data continue to be revised long after the end of the reference quarter, although the magnitudes of the revisions are negligible after 5 years. Revisions are due to the incorporation of additional microdata, benchmarking, adjustments to seasonal factors, and (for output) changes to definitions and methods—all of which are assumed to bring the estimates closer to "truth." We take a close look at these revisions and examine the extent to which these early estimates are good predictors of the estimate 5 years later.

We find that revisions to output growth are substantially larger than revisions to hours growth. Revisions are larger for Q1 and for recession quarters. Long-term revisions to growth rates tend to be smaller than revisions to levels because revisions to current quarter and prior quarter levels tend to be in the same direction and of approximately the same magnitude.

Following earlier research, we estimate Mincer-Zarnowitz regressions to examine whether these long-term revisions are "news" and whether they eliminate "noise." We find that the initial revisions to output and hours are news, while later revisions are not. Early revisions eliminate noise only for hours, while later revisions do not eliminate noise. Further investigation indicates that these findings are the result of the differing nature of early revisions and subsequent revisions.