Topic 05
Manufacturing employment in BLS data: how to read it
Manufacturing employment is one of the most quoted indicators in debates about the American economy, but the numbers alone are easy to misread. What exactly do the BLS data show, and what should you watch for before drawing conclusions? Below we explain where the data come from, how they change over time and what employment does not tell you about output.
What are the BLS manufacturing employment data?
BLS stands for the Bureau of Labor Statistics, the American agency for labor statistics. The series CES3000000001 covers all employees in United States manufacturing. Its source is the survey of establishments, known as the CES, not a survey of households.
The values are seasonally adjusted and given in thousands. That means a figure of 12,632 corresponds to about 12.6 million people. The unit can be confusing, so check it before comparing employment data with other sources.
The data are revised every month. The chart on this site shows annual averages calculated from monthly data downloaded from the public BLS API. Later revisions may therefore shift individual points slightly.
How did manufacturing employment develop?
Annual average employment peaked in 1979 at 19,428 thousand, or 19.4 million. In 2000 it was already 17,265 thousand. The largest fall came in the first decade of the 21st century, because in 2010 employment stood at 11,514 thousand, or 11.5 million. That is the lowest annual average in the whole 1960 to 2025 series.
Between 2000 and 2010 about 5.8 million manufacturing jobs disappeared. This period shapes the overall pattern of the series more than any other, so it is worth examining on its own and not only in comparison with the 1979 peak.
A partial recovery followed. In 2019 average employment was 12,779 thousand. In 2020 it fell to 12.1 million (12,125 thousand), and in 2025 it was 12.6 million (12,632 thousand). That last figure is lower than the 2019 figure but clearly higher than the 2010 figure.
| Year | Employees (thousands) |
|---|---|
| 1979 | 19,428 |
| 2000 | 17,265 |
| 2010 | 11,514 |
| 2019 | 12,779 |
| 2020 | 12,125 |
| 2025 | 12,632 |
Show all annual values, 1960 to 2025
| Year | Employees (thousands) |
|---|---|
| 1960 | 15,437 |
| 1961 | 15,010 |
| 1962 | 15,497 |
| 1963 | 15,632 |
| 1964 | 15,889 |
| 1965 | 16,618 |
| 1966 | 17,681 |
| 1967 | 17,897 |
| 1968 | 18,211 |
| 1969 | 18,572 |
| 1970 | 17,847 |
| 1971 | 17,171 |
| 1972 | 17,664 |
| 1973 | 18,584 |
| 1974 | 18,512 |
| 1975 | 16,913 |
| 1976 | 17,538 |
| 1977 | 18,174 |
| 1978 | 18,936 |
| 1979 | 19,428 |
| 1980 | 18,732 |
| 1981 | 18,634 |
| 1982 | 17,364 |
| 1983 | 17,049 |
| 1984 | 17,921 |
| 1985 | 17,818 |
| 1986 | 17,552 |
| 1987 | 17,608 |
| 1988 | 17,905 |
| 1989 | 17,984 |
| 1990 | 17,695 |
| 1991 | 17,068 |
| 1992 | 16,800 |
| 1993 | 16,776 |
| 1994 | 17,024 |
| 1995 | 17,244 |
| 1996 | 17,237 |
| 1997 | 17,418 |
| 1998 | 17,560 |
| 1999 | 17,323 |
| 2000 | 17,265 |
| 2001 | 16,441 |
| 2002 | 15,257 |
| 2003 | 14,508 |
| 2004 | 14,315 |
| 2005 | 14,224 |
| 2006 | 14,152 |
| 2007 | 13,871 |
| 2008 | 13,395 |
| 2009 | 11,836 |
| 2010 | 11,514 |
| 2011 | 11,710 |
| 2012 | 11,908 |
| 2013 | 11,997 |
| 2014 | 12,160 |
| 2015 | 12,308 |
| 2016 | 12,324 |
| 2017 | 12,406 |
| 2018 | 12,653 |
| 2019 | 12,779 |
| 2020 | 12,125 |
| 2021 | 12,312 |
| 2022 | 12,768 |
| 2023 | 12,873 |
| 2024 | 12,789 |
| 2025 | 12,632 |
What does employment say about output?
Little, if you look at it alone. Employment shows the number of jobs, not the volume of production. Labor productivity rises, so the same output needs fewer workers.
A fall in the number of employees therefore does not have to mean a fall in output. To judge industry, employment data must be set beside measures of production, which are not part of this overview. This overview also has no data on wages or hours worked, so no conclusions can be drawn about changes in them.
How to read employment data in practice
First, check the unit and the type of series: thousands of people, seasonally adjusted, annual averages. Second, compare similar points in time, for example 2010 and 2025, instead of looking only at the 1979 peak. Third, remember the revisions that the BLS makes every month.
Manufacturing employment is a useful indicator of the labor market, but it is not a measure of output. Read it alongside the other articles at Freight & Floor to put these numbers in a broader context.