If the people in even those places are making more today than in 2000, ‘bringing back jobs’ actually means ‘paying people less.’

One narrative goes like this: China joined the WTO with U.S. support in 2001. This worked great for the U.S. economy as a whole and helped China get rich, but it hollowed out specific U.S. communities affected by trade, especially those with many manufacturing jobs, which have been turned into economic wastelands where no one can find work, and the jobs that do exist are poorly paid. This “China shock” is responsible for dissatisfaction with the U.S. economy and needs to be corrected through robust protectionism and manufacturing subsidies.

Jeremy Horpedahl tests this narrative against the facts from the ten metropolitan areas that “China shock” scholars said were most affected by trade with China. Most of them have more jobs today than they did in 2001, and all of them have higher real wages for workers at every income level.

David Autor, an economist from the Massachusetts Institute of Technology, is the leader of the “China shock” literature in economic journals. In 2021, he and his coauthors identified which U.S. communities were most affected by trade with China. Some are very small and lack sufficient economic data, so Horpedahl looks at the ten metropolitan statistical areas (MSAs) as defined by government statistical agencies that are included in Autor’s list.

The ten most affected MSAs by the “China shock,” based on the list from the guy who started the economics literature on the subject, are:

If you’re surprised by some of the inclusions (San Jose? Austin?), keep in mind that there is a yawning chasm between what the “China shock” literature actually says and what people who often talk about it think it says. For example, the most famous paper does not even attempt to measure the net effects of trade with China on manufacturing employment or any kind of employment, U.S. policy changes were not the primary driver of the shock (it was China’s market reforms), and the literature does not show that trade with China is responsible for most of the manufacturing job losses in recent decades (they were more due to automation and the structural trends that happen as countries get richer).

Horpedahl stresses that he is steel-manning the “China shock” case by picking the ten worst-hit MSAs with available data, according to the people who coined the term. Every single one has had real wage growth since 2001. That’s true for the median worker in all ten areas, but it’s also true at the 10th percentile, 25th percentile, 75th percentile, and 90th percentile. In fact, wage growth at the 10th percentile was faster than wage growth at the median for all ten areas, just as it has been for the U.S. as a whole, since 2001.

Real wage growth for the median worker in the U.S. is 11.5 percent since 2001 (and remember, “real” means “adjusted for inflation”). Six of the ten “China shock” areas had median real wage growth higher than that (Sioux City, Lynchburg, Huntsville, San Jose, Austin, and Jonesboro).

All ten areas have seen declines in manufacturing employment, as has the U.S. as a whole, since 2000, the year before China’s joining the WTO (and, more important, the 2001 recession that hit manufacturing especially hard). But six of the ten areas have seen greater manufacturing job recovery, as a percentage of manufacturing jobs in 2000, than the U.S. as a whole (Huntsville, Sioux City, Jonesboro, Fort Wayne, Raleigh, and Austin). Huntsville and Sioux City have nearly the same number of manufacturing jobs as they did in 2000, while the U.S. on average only has 74 percent as many. One of the biggest losers, with 50 percent of its manufacturing jobs gone, is San Jose — not the Midwestern towns that populists imagine.

Were the job losses in manufacturing replaced by other jobs? Horpedahl checks that as well: Mostly yes. Total employment is higher in eight of the ten areas, and four of them (Austin, Raleigh, Huntsville, and Jonesboro) outpaced U.S. employment growth. The only two with fewer jobs are Lynchburg (just barely) and Hickory.

So if real wages are up in all ten areas, and employment overall is up in almost all of them, that means the jobs that replaced manufacturing were not low-paying. Most of the jobs that replaced them were service-sector jobs, which grew in all ten areas since 2000. This conclusion matches with national data, which show that the average hourly wage for service jobs is $1 higher than for manufacturing jobs ($36 vs. $35).

Horpedahl, who lives in Arkansas, gives the example of Jonesboro as a success story that goes against the narrative:

Employment grew faster than the national average, and wage growth also compared favorably. While Jonesboro is home to a mid-sized research university (Arkansas State University), it didn’t have a very high proportion of its population with college degrees in 2000: just 14.6 percent, less than half of the proportion in places like San Jose and Austin. Nonetheless, despite a 13 percent drop in goods-producing jobs in Jonesboro from 2000 to 2024, service-producing jobs grew by 55 percent. And this is not merely a “small base” effect, as service-sector employment was almost three times as large as goods-producing employment in 2000 in Jonesboro.

Are these communities perfect places to live? No, but nowhere is. The image conjured by populists of hollowed-out communities where no one can find work is not real in the places that “China shock” researchers say are most affected by trade with China. If the people in even those places are making more today than in 2000, “bringing back jobs” truly means “paying people less.”