
An official website of the United States government
Hover over the red dot to see historical information.
The U.S. Bureau of Labor Statistics (BLS) first released industry-level labor productivity measures at the end of the 19th century. The following is a timeline of events from then onward.
BLS has produced studies of labor productivity since the 1800s. In an early attempt to understand the detailed productivity effects of mechanization, BLS did a 5-year study precisely comparing tasks done by hand to those done with mechanical assistance. Data from the resulting 1600-page report to Congress in 1899 still provides new findings.1
BLS Commissioner Ethelbert Stewart began to advance productivity measurement when in 1922 he signed an agreement with Babson Statistical Organization for a productivity study of the construction industry. Commissioner Stewart then brought in Ewan Clague from the University of Wisconsin to lay the foundation of an industry productivity program to measure output per man-hour for the steel, automobile, shoe, and paper industries. A year later, 11 more industries were added. Clague would later go on to become the sixth BLS commissioner.2
By 1935, BLS made a proposal to the Works Progress Administration (WPA) to study productivity in 50 industries. In cooperation with the WPA National Research Project on Reemployment Opportunities and Recent Changes in Industrial Techniques, BLS conducted several labor productivity surveys. The main concern at the time was about labor displacement due to technological advances such as the telephone and automation developments in the steel industry. At the urging of unions, Congress authorized BLS to continue the labor productivity studies and devoted funds to support that effort resulting in the Productivity and Technological Development Division in 1940.3
After a brief pause during World War II, productivity studies resumed. However, the funding for surveys was cut, forcing the productivity program to rely on secondary sourcing to compile statistics. That tradition carries on today, as the BLS productivity program relies entirely on other agencies and programs for data to measure the nation’s productivity growth.
Labor productivity measures were published for aggregate manufacturing in 1955 covering the period from 1939 to 1953. That work expanded to publication of labor productivity for the total private economy in 1959 covering the period from 1909 to 1958.4 The productivity measures were later extended to quarterly indexes of labor productivity, compensation per hour, and unit labor costs for the business, nonfarm business, manufacturing (and its durable and nondurable goods-producing subsectors), and nonfinancial corporations sectors in 1967. The primary source of labor input for productivity measurement has always been the BLS Current Employment Statistics program data on employment and paid hours of employees in nonfarm establishments. To better reflect the labor being used in production, BLS moved from using an hours-paid measure (which includes paid time off) to an hours-worked measure in August 1989.5
BLS has improved labor productivity estimation as data availability has increased. In 1996, the labor productivity program incorporated improved BEA measures of real output in the U.S. business and nonfarm business sectors based on superlative indexes rather than constant dollars. The Current Population Survey had long provided employment and hours worked data for the self-employed and unpaid family workers. In June 2005, the BLS productivity program began to use information from this survey to take proper account of multiple jobholders. This improvement led to an entirely jobs-based employment measure being used in labor productivity measures.6 In addition, the productivity program introduced an improved method for estimating hours worked by nonproduction workers. Another improvement was made in 2017 when a new method for calculating the hours-worked-to-hours-paid adjustment ratios based on detailed industry data from the BLS National Compensation Survey was introduced.
In 1983, BLS developed and began publishing a broader measure of productivity, multifactor productivity (MFP), which is now referred to as total factor productivity (TFP), for the private business, private nonfarm business, and manufacturing sectors. These measures of TFP use capital and labor for combined inputs and were among the first to use a changing weight Törnqvist superlative price index.7
Over the years, the BLS productivity program increased the number of TFP series it publishes. In 1987, BLS expanded its TFP measures to include detailed coverage of the manufacturing sector with the release of total, durable, and nondurable manufacturing and 20 Standard Industry Classification (SIC) manufacturing industries. These measures fit the KLEMS growth accounting framework by using a sectoral output measure and measures of energy, materials, and services in addition to capital and labor for combined inputs.8
Improvements to the measurement of total factor productivity were also made along the way. The effects of industry research and development (R&D), conceived of as intellectual capital, was incorporated into multifactor productivity measures in 1989, and estimated effects on other industries (R&D spillovers) were incorporated in 1994.9
In 1994, BLS also began publishing measures of labor composition for the private business and private nonfarm business sectors.10 In the mid-2000s, the BLS productivity program conducted research about measuring labor composition on a more detailed industry basis.11 In 2016, BLS began estimating labor composition measures at a more detailed level for the 60 NIPA industries in the private business sector.
BLS first introduced nonmanufacturing industry productivity measures in 1999 and as prices and data expanded with a BLS-wide and statistical community-wide service initiative in the early 2000s, continued to improve on these measures. In 2006, BLS published its 60 industry measures on a North American Industry Classification System (NAICS) system for the first time. A total economy measure of TFP was introduced in 2009.
As part of the 2013 National Income and Product Accounts comprehensive revision, BEA started classifying intellectual property products as private fixed investment (PFI). Intellectual property products include software (purchases of prepackaged software and own-account software), research and development, and artistic originals. Starting in 2015, BLS included intellectual property products in the asset composition for the private business and private nonfarm business sectors as well as the detailed industries to better align capital services with BEA value-added output. In 2021, BLS replaced the term “multifactor productivity” with “total factor productivity.” This was a change in terminology only and did not affect data or methods. The change was made to improve the visibility and accessibility of BLS data and was accompanied by changes to the BLS website and all future productivity new releases. Also in 2021, BLS first published annual labor productivity measures for 4 regions, the 50 states, and the District of Columbia.
In 2008, a collaborative project between the BLS productivity program and the BEA National Accounts began to measure productivity at the total economy level. The first effort produced a concordance from the National Accounts measures to the BLS productivity program measures.12 Subsequent efforts produced a 63-industry decomposition of GDP by the capital, labor, energy, material, services, and multifactor productivity measures back to 1987 and research into a historical series back to 1947 was completed, meeting the desires of many key macroeconomist stakeholders.
In 2019, a collaborative project between the BLS productivity program and the U.S. Census Bureau first published experimental productivity dispersion statistics for manufacturing industries. These are a complement to the official BLS industry-level productivity statistics. Data were first released for manufacturing industries from 1997 through 2016. Subsequent releases have expanded the published data to provide greater detail about the industry productivity distributions. Dispersion statistics are updated as the data become available.
Also in 2019, BLS commissioned the Committee on National Statistics (CNSTAT), part of the National Academies, for a project evaluating how statistics should adapt to fundamental changes in the retail trade sector. CNSTAT’s 2021 report, “A satellite account to measure the retail transformation,” suggested developing data on retail margin productivity, pursuing more timely data releases, and including activity from supporting industries. BLS released experimental measures of these in 2023.
1 Jeremy Atack, Robert A. Margo, and Paul W. Rohde, “Automation of manufacturing in the late nineteenth century: the hand and machine labor study,” Journal of Economic Perspectives 33, no. 2 (2019): pp. 51–70; and Jeremy Atack, Robert A. Margo, and Paul W. Rohde, “De-skilling: evidence from late nineteenth century American manufacturing,” Explorations in Economic History 91 (2024).
2 Joseph P. Goldberg and William T. Moye, The First Hundred Years of the Bureau of Labor Statistics, Bulletin 2235 (U.S. Department of Labor, September 1985), https://fraser.stlouisfed.org/files/docs/publications/bls/bls_2235_1985.pdf.
3 The authorization is shown in 29 U.S.C. §2, dated June 7, 1940, which states: “The Bureau of Labor Statistics of the United States Department of Labor is authorized and directed to make continuing studies of productivity and labor costs in the manufacturing, mining, transportation, distribution, and other industries.” It is online at https://www.law.cornell.edu/uscode/text/29/2b. The Division’s launch is noted in the Annual Report of the Secretary of Labor for 1941, on p. 88 (https://babel.hathitrust.org/cgi/pt?id=uiug.30112032741677&seq=100).
4 Ibid., p. 203.
5 Mary Jablonski, Kent Kunze, and Phyllis Flohr Otto, “Hours at work: a new base for BLS productivity statistics,” Monthly Labor Review, February 1990, https://www.bls.gov/opub/mlr/1990/article/hours-at-work-a-new-base-for-bls-productivity-statistics.htm.
6 To learn more about job-based employment estimates, see “Construction of employment and hours for self-employed and other nonfarm workers and for all farm workers, using Current Population Survey data for primary and secondary jobs,” (U.S. Bureau of Labor Statistics, March 31, 2006), https://www.bls.gov/productivity/technical-notes/methods-major-sector-productivity-and-costs-employment-hours-self-employed.pdf.
7 Methods were described in a report, Trends in multifactor productivity, 1948–81, Bulletin 2178, (U.S. Department of Labor, September 1983). For a review, see Edwin R. Dean, Michael J. Harper, and Mark S. Sherwood, “Productivity measurement with changing-weight indexes of outputs and inputs” (paper presented at the OECD Expert Workshop on Productivity: International Comparison and Measurement Issues, May 2–3, 1996), https://www.bls.gov/productivity/technical-notes/productivity-measurement-with-changing-weight-indexes-of-outputs-and-inputs-1996.pdf.
8 Robert M. Solow, “Technical change and the aggregate production function,” The Review of Economics and Statistics 39, no. 3 (August 1957), pp. 312–320.
9 The Impact of Research and Development on Productivity Growth, Bulletin 2331 (U.S. Department of Labor, September 1989), https://www.bls.gov/productivity/articles-and-research/the-impact-of-rd-on-productivity-growth.pdf.
10 Labor Composition and U.S. Productivity Growth, 1948–90, Bulletin 2426 (U.S. Bureau of Labor Statistics, December 1993), p. 122, https://www.bls.gov/productivity/articles-and-research/labor-composition-us-productivity-growth-1948-1990.pdf.
11 Cindy Zoghi, “Measuring labor composition: a comparison of alternate methodologies,” Labor in the New Economy, (University of Chicago Press, October 2010), pp. 457–485, https://www.nber.org/chapters/c10834.pdf.
12 Michael J. Harper, Brent R. Moulton, Steven Rosenthal, and David B. Wasshausen, “Integrated GDP-productivity accounts,” American Economic Review 99, no. 2 (May 2009), pp. 74–79, https://www.aeaweb.org/issues/100.