By Paulina Phan, Senior Editorial Contributor
In 2000, about 17.3 million Americans worked in manufacturing. By 2011, that figure had fallen to roughly 11.4 million, a loss of nearly six million jobs in just over a decade. Economists continue to debate how much of that collapse came from trade, how much from automation, and how much from the two severe recessions of the period. What the research record now shows clearly is that one episode, the surge of imports from China after 2001, hit specific American places with a force that national statistics understated, and that many of those places have not fully recovered a generation later.
The policy turn and the import surge
Two decisions framed the era. In 2000, the United States granted China Permanent Normal Trade Relations, removing the annual congressional review of China’s tariff status. In December 2001, China joined the World Trade Organization. The bilateral US trade deficit with China widened from about $83 billion in 2001 to a peak of $419 billion in 2018, before narrowing to about $279 billion in 2023, according to US trade data.
The imports were concentrated in labor intensive goods: apparel, textiles, footwear, furniture, and consumer electronics. American communities that specialized in making those products suddenly competed with a much lower cost producer inside a newly predictable trade framework.
What the research found
The most influential work is by economists David Autor of MIT, David Dorn of the University of Zurich, and Gordon Hanson of Harvard. In their 2013 paper “The China Syndrome,” and in a 2016 study with Daron Acemoglu and Brendan Price, they compared American commuting zones, local labor markets of a few hundred thousand people, with different levels of exposure to Chinese imports.
Their central estimates: rising import competition from China caused net losses of about 2 to 2.4 million US jobs between 1999 and 2011, including roughly one million in manufacturing. Exposed areas saw higher unemployment, lower labor force participation, and reduced wages, with the effects falling hardest on workers without college degrees. The Economic Policy Institute, using a different method based on the trade deficit and input-output linkages, estimated 3.7 million jobs displaced through 2018, including 2.8 million in manufacturing.
The findings challenged a standard economic assumption: that workers displaced in one industry or town would move or retrain relatively quickly. In the most exposed commuting zones, they largely did not.
Why the damage lasted
In a 2021 follow up study, Autor, Dorn and Hanson examined whether the hardest hit areas had adjusted by 2019. They found they had not. Employment rates and earnings in the most exposed commuting zones remained depressed nearly two decades after the shock began, and a meaningful share of the population in those areas experienced absolute declines in real income even after accounting for the lower consumer prices that imports delivered. Adjustment, where it occurred, was measured in decades rather than years, and population loss made recovery harder: as younger workers left, local tax bases and services weakened, which encouraged further departures.
The places
Detroit is the largest example, though its decline began earlier with automation, suburbanization and shifts within the auto industry itself. The city’s population peaked at about 1.85 million in 1950 and fell to 639,111 in the 2020 census. Youngstown, Ohio, a steel center, began its steep slide with the 1977 “Black Monday” mill closures and lost most of its industrial base in the decades that followed; its population fell from a peak near 170,000 to about 60,000 in 2020, and local estimates show it still shrinking, to roughly 58,800 in 2025, with a median household income of about $34,000. In the Southeast, the furniture and textile towns of North Carolina and neighboring states ranked among the commuting zones most exposed to Chinese competition in the Autor, Dorn and Hanson research, and entire categories of domestic furniture production largely disappeared within a decade.
Steel towns such as Gary, Indiana, followed a similar path as steel employment contracted from tens of thousands of jobs to a fraction of that level.
The other side of the ledger
Not all economists assign trade the leading role. Researchers who emphasize productivity note that American manufacturing output continued to grow even as employment fell, which points to automation as a major long term driver. Analysts at the Cato Institute and the American Enterprise Institute argue that a full accounting shows offsetting gains from trade with China, including lower prices for consumers, cheaper inputs for American firms, and expanded export opportunities, and they caution against attributing the entire manufacturing decline to imports. The Autor, Dorn and Hanson estimates themselves assign China roughly a quarter to a half of the manufacturing job decline during the 2000s, depending on the period and measure, not all of it. Recessions in 2001 and 2008 to 2009 also destroyed large numbers of factory jobs independent of trade.
Where the literature broadly agrees is on the geography: whatever the national balance of gains and losses, the costs were unusually concentrated, and local labor markets adjusted far more slowly than earlier models predicted.
Where things stand now
Manufacturing employment has stabilized well below its 2000 level, and recent years have brought a wave of announced factory investment, much of it in the South and parts of the Midwest. Population data offer a mixed picture of revival. Census Bureau estimates released in May 2026 show Detroit grew for a third consecutive year, adding about 5,000 residents in 2025 to reach an estimated 649,095, after gains of 6,791 in 2024 and its first increase since 1957 in 2023. City officials attribute the growth to housing investment and neighborhood redevelopment. The gains are real but modest against a loss of 1.2 million people from the peak.
Other former industrial centers show fewer signs of a turn. Youngstown’s population continues to edge down, and across the most exposed commuting zones, incomes and employment rates remain below where pre-2001 trends would have put them. Policy responses have spanned administrations: tariffs on Chinese goods beginning in 2018, semiconductor and clean energy manufacturing incentives in 2022, and continued tariff measures since. Economists note that new plants tend to be far more automated than the factories they replace, so investment announcements have not translated into manufacturing employment on the scale of the losses. The research consensus, two decades on, is less about whether the shock happened than about its duration: for the towns at its center, it never entirely ended.
FAQ
How many jobs did the China shock cost? The most cited estimate, by Acemoglu, Autor, Dorn, Hanson and Price, is 2 to 2.4 million total US jobs between 1999 and 2011, including about one million in manufacturing. Other methods produce higher and lower figures.
Why did some towns recover and others did not? Research points to industry mix, education levels, and whether new employers arrived. Places able to attract new investment or diversify into health care, logistics and technology fared better than single industry towns.
Does automation or trade matter more? Both mattered. Trade explains a large share of the losses in the 2000s in the most exposed places; automation explains much of the longer run decline in factory employment even as output rose.
Sources and methodology
Employment figures are from Bureau of Labor Statistics data as summarized in the National Bureau of Economic Research digest of Acemoglu, Autor, Dorn, Hanson and Price (manufacturing employment of 17.2 million in 1999 and 11.4 million in 2011; an estimated 2 to 2.4 million net job losses from Chinese import competition, 1999 to 2011). The 2013 “China Syndrome” findings and the 2021 persistence study are by Autor, Dorn and Hanson (the latter NBER Working Paper 29401). The Economic Policy Institute estimate of 3.7 million displaced jobs through 2018 reflects that organization’s trade deficit based methodology and is included for comparison. Trade deficit figures are US Census Bureau data (2001, 2018, 2023). Detroit population figures are US Census Bureau decennial counts and annual estimates, including the May 2026 release covering July 1, 2025. Counterarguments are attributed to published analyses by the Cato Institute and the American Enterprise Institute.