Making (and Building, and Digging Up) Stuff Is Back in Fashion

The U.S. economy’s long-running shift to services takes a pause.

Let’s make some stuff.

Photographer: Luke Sharrett/Bloomberg

One of the most important economic changes in the U.S. over the past half century has been the shift away from making goods to providing services. But the May jobs report released Friday shows this particular economic recovery continuing to buck that trend, sort of.

This Recovery Has Got the Goods

Percentage change in nonfarm payroll employment since February 2010

Source: U.S. Bureau of Labor Statistics

Seasonally adjusted. Goods-producing sectors are mining, logging, construction and manufacturing. Service-providing sectors are everything else, including government.

I say “sort of” because employment in the service-providing sectors was so much bigger to start with that it has added many more jobs than the goods-producing sectors since February 2010 — which I used as the start date because that’s when payroll employment stopped falling after the Great Recession. Look at employment, rather than its rate of change, and it’s clear that the U.S. job market is not in the midst of some kind of major shift back to mining, construction and manufacturing.

The Big Shift to Services

Nonfarm payroll employment, seasonally adjusted

Source: U.S. Bureau of Labor Statistics

Still, the employment outperformance of the goods-producing sectors in this recovery really is something the U.S. economy hasn’t seen in decades.

U.S. Bureau of Labor Statistics

Some of this outperformance, especially in the early years of the recovery, was simply manufacturing and construction employment rebounding from an especially awful recession. But now, eight years into the recovery, it’s got to be more than just that. Here’s one more way of slicing the numbers that shows just how unique the past few years have been. 1

Services' Employment Share Has Flatlined

Employment in service-providing sectors as a percentage of total nonfarm employment

Source: U.S. Bureau of Labor Statistics

Note that the y-axis does not go to zero.

What’s causing this? It’s partly that, at more than 86 percent of nonfarm employment, there may not be all that much room left for services’ share to grow. We apparently still need some people to make things, build things, dig things out of the ground and cut things down. It’s also that, after big employment cuts during the first decade of the 21st century due to automation and offshoring to China, U.S. manufacturing has been making a modest comeback. Labor-cost differentials between the U.S. and China have shrunk, and some manufacturers have been rethinking far-flung supply chains. The fracking-enabled boom in U.S. oil and natural gas production (which falls under mining in the employment numbers) has added jobs, too. And policy decisions have surely helped, with the 2009 auto industry rescue and stimulus package boosting manufacturing and construction employment early in the recovery and last year’s business tax cuts seemingly boosting both now.

This goods-producing renaissance is a healthy development for a number of reasons. For one thing, it’s providing some jobs for men without college educations, who have been struggling in the labor market for decades (in the goods-producing sectors, 78 percent of jobs are held by men; in services, men are a minority at 46 percent). I also get the sense that a lot of goods-producing jobs are locating in places outside of big metropolitan areas that have been struggling economically. 2 And jobs in goods-producing sectors tend to have higher multiplier effects (that is, they stimulate more economic activity) than service work.

The U.S. economy is never going back to the era where most jobs involve producing tangible things. But the epic shift from goods to services may have passed an inflection point.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

  1. Why does one chart start in 2010 and the others in 1939, 1990 and 1960? For 1939, it's because that's how far back the data series goes. With the rest, it's simply because I chose the time periods that best illustrated the points I was trying to make. The long, almost uninterrupted rise in services' share of employment began in the late 1950s, for example, so 1960 seemed like a logical place to start.

  2. This is something that it's possible to get more than a sense for, but I fear that digging up the data to back up that assertion would take the rest of the day, if not longer.

To contact the author of this story:
Justin Fox at

To contact the editor responsible for this story:
Brooke Sample at

Before it's here, it's on the Bloomberg Terminal.