Case Study 2 — The China Shock and the Politics of American Manufacturing

In 2001, China joined the World Trade Organization. The accession was the culmination of fifteen years of negotiation and represented a substantial commitment by China to lower its trade barriers, accept WTO dispute resolution, and integrate more deeply into the global trading system. From the perspective of American policymakers at the time, it looked like a clear win: China would lower its tariffs (good for U.S. exporters), commit to standard trade rules (good for U.S. negotiators), and become a more reliable trading partner (good for everyone).

What happened next was not what most economists predicted. Chinese exports to the United States grew at an extraordinary rate — much faster than the standard models had estimated. Over the decade from 2001 to 2011, the value of Chinese imports to the U.S. tripled. Specific sectors of U.S. manufacturing — furniture, certain textiles, electronics, some chemicals — were particularly hard hit. American factories closed. American workers lost their jobs. American communities that had been built around those factories struggled to recover.

By the mid-2010s, this was being called the "China shock." It became one of the most studied episodes in the history of empirical trade economics, and the findings have reshaped how the field thinks about the costs and benefits of free trade.

This case study walks through the China shock as a phenomenon, the empirical work that documented it, and what we now know about its consequences. The picture that emerges is one of a major economic event whose costs were larger and more persistent than anyone predicted, but whose aggregate gains were also real.

The size of the shock

To get a sense of what we're talking about: the value of U.S. imports from China rose from about $100 billion in 2001 to over $400 billion by 2011. As a share of U.S. GDP, Chinese imports grew from about 1% to about 2.5%. As a share of U.S. manufactured-goods consumption, Chinese imports rose from about 6% to about 18% over the same period.

The growth was concentrated in specific industries. Furniture imports from China rose nearly tenfold. Apparel imports rose substantially. Consumer electronics — though much of the actual value-added was elsewhere in the supply chain — saw enormous flows. Certain chemical products, certain metals, certain machinery: all saw rapid Chinese penetration of U.S. markets.

For comparison, these flows were larger than the post-NAFTA growth in trade with Mexico. They were larger than the post-EU-expansion growth in trade with Eastern Europe. They were one of the largest single trade-related shocks the U.S. economy has experienced in modern history.

The Autor-Dorn-Hanson research

David Autor (MIT), David Dorn (University of Zurich), and Gordon Hanson (now at Harvard, then at UC San Diego) began studying the China shock in the late 2000s. Their key methodological insight was to exploit geographic variation in exposure to Chinese import competition.

The U.S. has many local labor markets. Some of these markets specialize in industries that compete heavily with Chinese imports (furniture in North Carolina, textiles in the Carolinas and Georgia, certain electronics in the Midwest). Others specialize in industries that don't compete with China (services, agriculture, oil production, government work). By measuring how exposed each local labor market was to Chinese import competition (based on its industry mix), Autor et al. could see whether more-exposed markets fared worse than less-exposed ones.

The methodology let them isolate the causal effect of Chinese import competition from the many other things that were happening to the U.S. economy in the 2000s.

What they found

Finding 1 — Manufacturing employment fell. In communities heavily exposed to Chinese import competition, manufacturing employment fell substantially — by about 1 million jobs net over the 2000s, attributable specifically to Chinese imports. (The total decline in U.S. manufacturing employment over this period was much larger, but only about 1 million was attributable to China specifically; the rest was due to automation, other trade, and broader structural change.)

Finding 2 — The effects spilled into other sectors. Counties hit hard by Chinese imports didn't just lose manufacturing jobs. They also lost jobs in adjacent service sectors (restaurants, retail, construction) that depended on manufacturing payrolls. Total employment in heavily exposed counties fell by more than just the direct manufacturing job losses.

Finding 3 — Wages fell for workers who kept their jobs. Even workers in heavily-exposed counties who didn't lose their jobs saw wage growth slow significantly. The competition for the remaining jobs depressed wages across the board.

Finding 4 — Workers did not relocate as expected. The standard model assumed displaced workers would move to other regions or other industries. The empirical record showed that this happened much less than expected. Workers in displaced communities stayed put. They drew on unemployment insurance, then disability, then welfare. Many retired early. Some never worked again.

Finding 5 — Government transfer spending rose substantially. Communities heavily exposed to Chinese imports saw large increases in disability claims, food assistance, and other social safety net programs. This represents real economic damage that the simple welfare calculation does not capture.

Finding 6 — Marriage rates fell. Particularly among working-age men, exposure to Chinese import competition was associated with measurably lower marriage rates. Why? Lower employment and lower wages reduced men's marriageability in the marriage market, particularly in communities where local social norms favored male earners.

Finding 7 — "Deaths of despair" rose. Affected communities saw increases in suicide, drug overdose, and alcohol-related deaths. The connection to trade-related job loss was not the only factor (the opioid crisis was a major contributor), but trade exposure was a meaningful predictor of community-level health outcomes.

The political follow-up

In 2020, Autor, Dorn, Hanson, and Kaveh Majlesi published a follow-up paper looking at the political consequences. They found that:

• Counties more exposed to Chinese import competition shifted toward more polarized politics
• Specifically, in the 2016 presidential election, areas heavily affected by the China shock were more likely to vote for the Republican candidate (Donald Trump), and the shift was statistically significant
• The estimated effect was large enough that, plausibly, the China shock affected the outcome of the 2016 election in several swing states (Pennsylvania, Michigan, Wisconsin)

This finding is contested in some details — the exact magnitude of the political effect is hard to estimate precisely, and other factors (cultural concerns, immigration anxiety, other economic stresses) also mattered. But the general finding — that trade-related economic damage produces political backlash — is widely accepted.

What the China shock means for trade theory

It does not refute comparative advantage. Aggregate gains from trade with China were real. American consumers benefited from lower prices on consumer goods. American workers in export industries (especially in agriculture and high-end manufacturing) gained from larger Chinese markets. The textbook story of "trade increases the size of the pie" is empirically supported by the China shock data.

It does refute the naive version of trade adjustment. The simple comparative-advantage story assumes that displaced workers find new jobs in expanding industries reasonably quickly. The empirical record on the China shock shows this happened much less than expected — particularly for older workers, workers in regions with weak labor markets, workers without easily transferable skills, and workers in the depths of the Great Recession.

It does suggest that policy needs to do more. The U.S. response to the China shock — primarily Trade Adjustment Assistance, which is small and underfunded — was clearly inadequate to the scale of the disruption. The political backlash that followed was, in part, a response to the gap between what economists predicted and what affected workers actually experienced.

It does suggest that economists were wrong about something specific. Most economists in the 1990s and early 2000s would have predicted smaller and faster-resolving labor market effects from trade with China. The empirical record shows the predictions were wrong. This is the kind of update — where the evidence forces a revision of professional consensus — that the field should be honest about.

What policymakers should learn

The lessons from the China shock are not "stop free trade." They are:

Lesson 1 — Adjustment costs are larger than the simple model predicts. Trade liberalization should be accompanied by substantial investment in worker assistance, retraining, regional aid, and infrastructure for affected communities. Not as a side benefit but as a core part of the policy.

Lesson 2 — The "long run" is sometimes very long. Some workers and communities never recover from major trade shocks. Policies should not assume that adjustment will happen automatically. They should plan for the cases where it doesn't.

Lesson 3 — Distributional effects can be politically destabilizing. Trade liberalization that produces large concentrated losses in specific communities can generate political backlash that reverses the policy and harms broader political stability. The China shock contributed to the political environment in which both Democratic and Republican candidates in 2016 promised to be tough on trade.

Lesson 4 — Pace and predictability matter. A trade liberalization that gives affected workers and communities time to adjust is much less damaging than one that produces sudden, large shifts in import volumes. The China shock was so disruptive partly because it happened so fast.

Lesson 5 — Compensation must be funded adequately. Trade Adjustment Assistance, the main U.S. compensation program, has had budgets in the hundreds of millions of dollars per year — utterly inadequate for a shock that affected millions of workers. If we're going to keep doing free trade, we need to fund the compensation programs at a level commensurate with the disruption.

The lessons of the China shock are not anti-trade. They are pro-honesty about trade. The aggregate gains are real. The distributional costs are real and have been underestimated. The right response is not to abandon trade but to do a much better job of compensating its losers — and the political stability of free trade depends on getting this right.

Discussion questions

1. The Autor-Dorn-Hanson findings have been widely accepted in the economic literature. Why have the policy responses been so weak?

2. The China shock is a historically unusual event. Are the lessons applicable to other trade liberalization, or were the magnitudes specific to that one episode?

3. Imagine you are designing a Trade Adjustment Assistance program from scratch, with adequate funding. What would it include?

4. The 2020 paper suggested the China shock affected the 2016 election outcome. What does this tell you about the relationship between economic policy and political stability?

5. The case study argues that "the lessons of the China shock are not anti-trade." Do you agree? Are there reasonable ways to read the same evidence and conclude that the U.S. should restrict trade more aggressively?