The myth of the China shock

The decline in US manufacturing jobs began decades before China's WTO entry. The bigger challenge is helping workers and regions adapt to economic disruption

Few issues in the public square are uncontested. But the narrative that the United States’ decision, driven by elite considerations, to open trade with China led to the loss of a large number of manufacturing jobs and deindustrialisation comes close. Democrats and Republicans, the mainstream media, commentators of all stripes, and even economists hold, advance, and defend this version of events. And yet every part of it is off base. 

Did growing US trade with China following the 2000 decision to permanently normalise trade relations and China’s accession to the World Trade Organization in 2001 lead to a large reduction in manufacturing employment? In their 2013 paper, economists David Autor, David Dorn, and Gordon Hanson find that rising exposure to Chinese import competition is associated with a net reduction in US manufacturing employment of 1.5 million from 1990-2007. In work with Daron Acemoglu and Brendan Price, they find that up to 2.4 million jobs were lost through 2011 due to competition with Chinese imports. 

 

To assess whether these are large numbers of job losses, place them in the context of broader US labour market dynamism. From 2000 to 2007, more than five million workers, including around 425,000 manufacturing workers, separated from their employers in a typical month. There was nothing special about that period — these magnitudes have been similar over the last five years. 

Moreover, when it comes to trade liberalisation, import competition is only half the story. In the 1980s, 1990s, and 2000s, trade with China, and globalisation more broadly, led to growing opportunities for US exporters. 

Economic theory suggests that trade liberalisation should have little effect on aggregate US employment because job losses from import competition can be balanced by job gains in export-intensive firms and sectors. The economist Robert Feenstra and his coauthors attempt to account for both sides of the ledger. 

In a 2019 paper, Feenstra and his colleagues confirm the “China shock” result, finding that 1.9 million jobs were lost between 1991 and 2011, owing to import competition from China, with more jobs lost to competition from global imports. But they also find that a roughly equivalent number of jobs were gained due to export expansion. 

One should also consider that manufacturing’s share of total US employment followed a relatively smooth downward trend from the early 1950s until the 2008 financial crisis, when falling productivity actually caused the trend to slow. The decline predates the “China shock” by decades. And there was no obvious trend break in manufacturing’s employment share in 2000 or 2001, which is consistent with the view that, over the long term, declines in manufacturing employment have been driven mainly by productivity growth, not by trade competition. 

Finally, the US did not decide to open trade with China in the 1990s in the same way that I decided to have a third espresso this morning. The decision wasn’t nearly so simple or singular. 

Yes, China was granted permanent normal trade relations in 2000 and entered the World Trade Organization in 2001. But the US had annually renewed China’s normal trade relations status since 1980, and US trade with China grew rapidly over two decades prior to its WTO accession. According to my calculations, China’s share of total US imports grew during the 1980s, hit 2.5 per cent in 1989, had more than doubled to 5.4 per cent by 1993, and stood at 8 per cent in 1999. 

The trend continued following China’s WTO accession. China’s share of total US imports doubled again, from 8.2 per cent in 2000 to 16.4 per cent in 2007. But even this overstates the role of US policy in facilitating the so-called China shock. China’s exports continued to grow in part because the US eliminated the uncertainty created by the pre-2000 annual renewal of trade-policy parameters. They also grew because of China’s internal, pro-market reforms — including a reduction in its own tariff rates. 

Nor was the US decision to trade with China made by a shadowy elite. China’s exports to America grew as a result of millions of decentralised, individual decisions. During the 1980s and 1990s, US consumers and businesses increasingly chose to purchase goods made in China, a trend that continued following China’s entry into the WTO. 

It is wrong to present the “China shock” as evidence that trade liberalisation hurts the working class, or that a powerful, murky, nefarious elite is making deliberate, isolated choices that hurt the majority of Americans. 

The right lessons to draw from the 2000s are that it is harder than economists and policymakers had thought for workers to move from a sector with declining economic opportunities into a sector with expanding opportunities. Moreover, workers are less willing to relocate from regions with declining opportunity than they had been in the past — let alone at a pace that would establish a new aggregate labour-market equilibrium. 

These two lessons are broadly applicable, and policymakers should keep both top of mind as advances in generative AI continue. They should be open to new types of policies — for example, public relocation assistance for workers in localities hit hard by economic disruption, and potentially large earnings subsidies to help workers in transition. 

But policymakers should not build walls around the economy or attempt to slow the pace of technological change. They should approach the future with optimism, not with fear — unburdened by the incorrect view that dynamism and economic liberalism are obstacles to long-term prosperity, rather than its key drivers.

---

The writer is director of economic policy studies at the American Enterprise Institute  © Project Syndicate, 2026