Federal Reserve Economic Data: Your trusted data source since 1991

The FRED® Blog

Is the decline in manufacturing economically “normal”?

Deciphering the phases of economic development

The FRED graph above tracks the proportions of employees working in three industries—construction, mining and logging, and manufacturing—since 1939. Construction (the blue line) has remained roughly horizontal. Mining and logging (the green line) has steadily declined. And manufacturing (the red line) has noticeably declined as well. This trend may look like weakness for the U.S. economy, but is it something to worry about?

Let’s take a step back: Historically, economic development has led to a declining share of workers in goods-producing sectors. The first sector to decline is agriculture,* whose workers moved to manufacturing and mining during the Industrial Revolution (which pre-dates our graph by a century or so). In the 19th century and beyond, the U.S. economy grew further and progressed to the next phases of development, with mining and manufacturing losing relative importance.

So if the U.S. economy is growing, where is it growing? The graph below shows the service sector has taken up the slack. At the start of the graph, in 1939, this sector had already made up 50% of non-farm employees, and it has continued to grow. The remaining sector, government, has remained relatively flat over the 80 years of this data series. Clearly, the U.S. economy is now much less focused on “making things.” Rather, the emphasis is now on education, health, leisure, retail, information, and finance.

How these graphs were created: Search the Current Employment Statistics release table and choose Table B-1 (seasonally adjusted); select the series you want and click “Add to Graph.” From the “Edit Graph” panel, for each line add series “All employees, non-farm” and apply formula a/b*100.

*Why don’t we show agricultural employment here? For one thing, it’s really hard to count: Many are part-time/seasonal workers and relatives that work on family farms.

Suggested by Christian Zimmermann.

View on FRED, series used in this post: CES0800000001, MANEMP, PAYEMS, USCONS, USGOVT, USMINE

A plateau for manufacturing?

After steady growth, manufacturing productivity seems at a standstill

The Bureau of Labor Statistics’ productivity and costs release provides data that can help us better understand the state of U.S. manufacturing. The graph above shows the evolution of manufacturing output since 1987. Notice the slow but steady growth in output since the Great Recession’s big dip.

What’s behind this slow and steady growth? The first suspect we’ll look at is manufacturing employment. The graph above shows there’s been a strong downward trend, which has accelerated during each recession. Yet, since 2010, manufacturing employment has been slowly making its way back up.

Next we’ll look at how much each worker produces in the manufacturing sector. Here, the story’s different: The general trend has been continuous increases in productivity per worker, but something seems to have broken with the Great Recession. First a major drop in productivity, then some progress getting back to trend, and then no progress since about 2010.

What if, since the Great Recession, manufacturing jobs have offered fewer hours of work or more part-time work? Maybe productivity per hour worked is growing. But the graph above, which shows productivity per hour instead of per person, shows no difference. The cause of this productivity standstill is thus either lack of technological progress or (more likely) a change in the composition of the manufacturing workforce toward lower-productivity work.

How these graphs were created: Search for “manufacturing sector” and each of the discussed series should be among the top choices. Simply choose them and click “Add to Graph.”

Suggested by Christian Zimmermann.

View on FRED, series used in this post: OPHMFG, OUTMS, PRS30006013, PRS30006163

How’s manufacturing?

Depends on the sector

The industrial production (IP) index is a popular metric of economic activity because it’s available relatively quickly. This monthly data series covers only a part of economic activity, however. In particular, it misses the service sector and the government sector. The graph above shows its evolution since 1972 along with a subcomponent that covers only manufacturing. Note that the index is set at 100 in 2012, meaning that all the indexes will always cross in 2012. A particularly healthy sector will start lower before 2012 and then rise higher after 2012. The graph shows that manufacturing has done well compared with overall industrial production before 2012 and a little less well afterward. This hides considerable sectoral differences, though.

In this second graph, we highlight some sectors within manufacturing. The sector that appears to have suffered massive losses is apparel and leather goods. Indeed, clothes manufacturing largely migrated abroad during this time span, with a decrease in production of about 80% since the mid 1990s. On the other extreme is computer manufacturing; It was insignificant in the first years but has increased by 1200% since the mid 1990s. All other sectors lie somewhere in between, and they average out to the manufacturing index shown in the first graph, which does not look as dramatic as the second graph. Some other interesting observations in this second graph: The primary metal industry has remained essentially unchanged over the past 45 years, with its index hovering around 100 throughout the sample period. The furniture industry incurred great losses from the Great Recession that it has not yet recovered from. And the car industry is doing pretty well.

How these graphs were created: For the first, search for industrial production, select the two series (likely the top choices), click on “Add to Graph,” and adjust the time period to start on 1972-01-01. For the second, go to the industrial production release, and select the monthly and seasonally adjusted tags. In the list, choose the series according to industry, and click on “Add to Graph.”

Suggested by Christian Zimmermann.

View on FRED, series used in this post: INDPRO, IPG315A6S, IPG331S, IPG334S, IPG3361T3S, IPG337S, IPMAN


Subscribe to the FRED newsletter


Follow us

Back to Top