The question of trade and inequality is as old as the discipline. Adam Smith and David Ricardo argued that countries gain from trade, and they were right. They said much less about who, inside each country, gains and who loses. The answer turns out to depend on what each country has more of, and the answer is sometimes uncomfortable. Workers in industries that compete with imports tend to lose. Workers in industries that benefit from exports tend to gain. Owners of factors that are scarce locally and abundant abroad lose. Owners of factors that are abundant locally and scarce abroad gain. Trade makes a country richer in total. It also redistributes income across people inside the country in ways that can be sharp.
The formal version of this insight is the Stolper-Samuelson theorem, published in 1941. For most of the next fifty years, the theorem was a textbook curiosity. Trade flows were small, the workers affected were dispersed, and the political tension around trade focused on national producers rather than individual workers. That changed in the 2000s, when China’s accession to the WTO produced a wave of imports into advanced economies that was concentrated in time, in industries, and in regions. The empirical work that followed, particularly David Autor, David Dorn, and Gordon Hanson’s 2013 paper on what came to be called the “China shock,” moved the trade-and-inequality debate from theory to evidence. The result has reshaped how economists, policymakers, and voters think about trade.
The Stolper‑Samuelson Foundation
The Stolper-Samuelson theorem follows from the Heckscher-Ohlin model of trade. Countries export goods that use intensively the factors of production they have in abundance, and import goods that use intensively the factors they have less of. When a country opens to trade, the relative price of its abundant-factor good rises, and the relative price of its scarce-factor good falls. Stolper and Samuelson showed that these price changes translate into real income changes for the factors themselves: the abundant factor gains, the scarce factor loses, and the losses are real, not just relative.
Applied to a typical advanced economy in the 1990s, the theorem produces a sharp prediction. The United States, Europe, and Japan are abundant in skilled labor and capital. They are scarce in unskilled labor. Opening to trade with developing economies abundant in unskilled labor should raise the wages of skilled workers in advanced economies and lower the wages of unskilled workers. The mirror image holds in developing economies: unskilled wages should rise, skilled wages should fall.
The theorem does not predict that trade is bad. The country as a whole gains. It predicts that some workers lose, and that the losses are not psychological discomfort about change but actual reductions in real income. The implication is that trade requires either redistribution from winners to losers, or political conflict, or both.
The Trade-versus-Technology Debate of the 1990s
By the early 1990s, the United States had experienced two decades of rising wage inequality, especially between college-educated and high-school-educated workers. The Stolper-Samuelson framework offered a ready explanation: trade with developing economies was lowering the relative wage of unskilled workers, exactly as the theorem predicted. The timing roughly matched: the wage gap widened as trade with developing economies expanded.
Most labor economists, however, were skeptical. They pointed to a competing explanation: skill-biased technological change. Computers, automation, and information technology disproportionately complemented skilled workers and substituted for unskilled tasks. The skill premium was rising because demand for skilled workers was rising relative to supply, and this was happening for technological reasons that had nothing to do with trade.
The trade-versus-technology debate dominated the 1990s. The empirical balance settled on technology as the larger force, with trade playing a smaller but real role. The reasoning was that trade flows in the 1980s and early 1990s were too small to account for the bulk of the observed wage divergence, and trade-exposed and non-exposed industries showed similar rises in the skill premium, suggesting a common cause that was not industry-specific. The consensus held through the 1990s that trade mattered for inequality but was not the main story.
The consensus was correct for its time. It became incorrect after 2001.
The China Shock
China’s accession to the World Trade Organization in December 2001 produced a wave of Chinese exports into advanced economies on a scale that the 1990s data could not have anticipated. US imports from China rose from approximately 1 percent of GDP in 2000 to over 2 percent by 2007, and the share of US manufacturing employment exposed to Chinese import competition rose sharply. Within manufacturing, sectors such as furniture, textiles, electronics, toys, and metal products faced direct competitive pressure from Chinese producers.
The work that turned this episode into a research program was Autor, Dorn, and Hanson’s 2013 paper “The China Syndrome: Local Labor Market Effects of Import Competition in the United States,” published in the American Economic Review. Their methodological contribution was to exploit variation across US commuting zones in their exposure to Chinese import growth. Some local labor markets, defined by their pre-existing industry mix, were highly exposed because they specialized in goods that China rapidly came to produce. Others were barely exposed because their local industries were not in the affected categories.
By comparing exposed and non-exposed commuting zones, they could estimate the local labor-market effects of the China shock more cleanly than aggregate national time-series analysis would allow. The results were striking. Exposed commuting zones experienced sharp declines in manufacturing employment, falls in wages, and rises in unemployment, disability claims, and government transfers. The effects persisted for over a decade. Workers who lost manufacturing jobs in exposed zones did not, on average, transition smoothly to other sectors. They stayed in the same regions, and many never recovered their previous earnings.
| Era | Dominant view | Key papers | What changed |
|---|---|---|---|
| 1941–1980s | Theoretical: scarce-factor losers exist but are dispersed and politically manageable | Stolper & Samuelson (1941) | Limited cross-country trade flows; effects modest in practice |
| 1990s–early 2000s | Technology, not trade, explains rising inequality in advanced economies | Krugman (1995); Berman, Bound, & Machin (1998) | Skill-biased technological change identified as the larger driver |
| 2013–today | Trade shocks produce large, persistent, concentrated local labor-market effects | Autor, Dorn, & Hanson (2013); Autor, Dorn, Hanson, & Song (2014); Pierce & Schott (2016) | Quasi-experimental evidence from commuting-zone variation; longer time horizon allowed dynamic effects to be measured |
|
|
|||
The Magnitude of Manufacturing Decline
US manufacturing employment had been on a slow secular decline since the 1970s. The China shock years accelerated that decline dramatically. Between 2000 and 2010, US manufacturing employment fell by roughly 5.8 million jobs, or about 30 percent of the manufacturing workforce. Subsequent research using the Autor-Dorn-Hanson methodology and similar approaches attributed roughly a quarter to a third of the gross manufacturing decline in this period directly to the China shock, with the remainder driven by automation, demand shifts, and broader structural change.
The chart below shows the stylized trajectory of US manufacturing employment over the period, with the acceleration around the early 2000s clearly visible.
The Failure of Regional Adjustment
Standard trade theory assumes that displaced workers transition to other industries, possibly with some adjustment costs, and that the aggregate gains from trade more than compensate the losers in principle. The China shock literature demonstrated that, at least in the case of US commuting zones in the 2000s, this assumption broke down.
Workers displaced from manufacturing did not easily reallocate. Several patterns emerged in the follow-up research. Displaced workers tended to stay in the same commuting zones rather than relocating, partly because housing market conditions, family ties, and home equity made moving expensive, and partly because the alternative regions did not absorb them easily. Many displaced workers transitioned not to other jobs but to disability insurance, early retirement, or long-term unemployment. The communities themselves declined: tax bases shrank, public services degraded, family formation rates fell, and so-called “deaths of despair” from opioid overdoses, alcohol, and suicide rose sharply.
The China shock also produced second-order effects. Local non-tradable sectors that depended on manufacturing wages, retail, restaurants, services, contracted as the manufacturing wage base shrank. Workers in those non-tradable sectors lost jobs even though they were not directly exposed to import competition. The spillovers meant that the total local effect was larger than the direct exposure measure would suggest.
These patterns mattered for policy and politics. The political backlash against trade in the United States and parts of Europe in the mid-2010s, including the support for Brexit and for tariff-focused presidential candidates, mapped reasonably well onto the geography of trade exposure. Researchers found correlations between China-shock exposure and shifts in voting patterns, although the size of the causal effect remains debated. What is not debated is that the local pain produced political pressure that the standard “gains from trade with compensation” framework had not anticipated.
The compensation problem. The classical defense of free trade has always included the caveat that winners should compensate losers. In practice, US Trade Adjustment Assistance and similar programs in Europe were small, narrowly targeted, and reached only a fraction of displaced workers. The aggregate gains from trade were real but were not effectively redistributed to the local labor markets that bore the concentrated losses. The compensation theorem became, in practice, an unfulfilled promise.
Connections to Modern Trade Theory
The China shock literature did not overturn Stolper-Samuelson. It refined it and added empirical depth. Three theoretical developments connect the older and newer literatures.
Heterogeneous firms within sectors. The Melitz heterogeneous firms model showed that trade liberalization affects firms within an industry differently. The most productive firms expand and pay higher wages. The least productive firms exit. Workers at exiting firms suffer; workers at expanding firms benefit. The within-industry wage divergence predicted by the Melitz framework matches the firm-level evidence that exporters in trade-exposed industries paid higher wages while non-exporters in the same industries paid lower wages or shut down.
Tasks rather than goods. The trade-in-tasks framework from Grossman and Rossi-Hansberg refined the Stolper-Samuelson logic by noting that what gets traded is not whole goods but tasks. Workers performing offshorable tasks compete with foreign workers regardless of which industry they work in. The wage effects are concentrated on workers performing routine or offshorable tasks rather than on entire industry workforces.
Local labor markets. The Autor-Dorn-Hanson framework formalized the importance of geographic concentration. Trade shocks fall on commuting zones rather than abstract national factor markets, and the geographic structure of adjustment matters as much as the sectoral structure. This insight has influenced research on automation, climate transition, and other structural shocks where local concentration drives the political and social consequences.
Beyond the China Shock
The intensity of the China shock peaked in the late 2000s. By the early 2010s, Chinese wages had risen, US import growth from China had slowed, and manufacturing employment had stabilized at a much lower level. The shock as a discrete event was largely complete. But the patterns it documented have reshaped debates about subsequent structural changes.
Automation. Research on automation and AI has used the China-shock methodology, comparing local labor markets exposed to robot adoption or to AI deployment. Some studies have found persistent local employment effects similar in structure to the trade case, though typically smaller in magnitude. The lesson appears to be that any large, concentrated shock to local labor markets produces dynamics that look more like the China shock than like the smooth aggregate adjustment of textbook trade theory.
Reshoring and supply chains. The debates around supply chain restructuring after COVID-19 and geopolitical tensions have drawn on the China-shock literature to argue both sides. Supporters of reshoring point to the local labor-market gains of bringing manufacturing back to the United States or to allied economies. Critics point out that reshoring creates the mirror-image problem in the countries losing the production, that the costs to consumers of higher prices may exceed the local employment gains, and that adjustment goes in both directions.
Tariff policy. The 2018-2019 US tariffs on Chinese imports, and the broader tariff wave that emerged in 2025-2026, were partly motivated by the China-shock evidence. Empirical studies of these tariffs have generally found that they did not restore manufacturing employment in the affected regions, that they raised consumer prices, and that retaliation hit US export-exposed regions. The lesson is that the China-shock evidence makes a case for stronger labor-market policy, not necessarily for tariffs.
What the Evidence Now Supports
The combined research from Stolper-Samuelson to the China shock now supports a small number of robust conclusions.
The aggregate gains from trade are real. Consumers benefit from lower prices and greater variety. Firms benefit from access to larger markets and to cheaper intermediate inputs. National output rises in both trading partners.
The distribution of gains is uneven and partly predictable. Stolper-Samuelson correctly identifies the broad pattern: scarce-factor workers in each country lose, abundant-factor workers gain. The Melitz framework refines this to within-industry firm heterogeneity. The trade-in-tasks framework refines it further to task-level offshorability.
Local labor-market concentration matters. Geographic and sectoral concentration of trade exposure produces persistent local effects that aggregate analysis misses. The decade-long persistence of the China-shock effects on US commuting zones is the empirical anchor for this conclusion.
Compensation is hard. Aggregate gains from trade do not automatically reach the workers who bear the losses. Without active policy intervention, the geographic and sectoral concentration of losses produces political backlash that affects subsequent trade policy.
The China shock was exceptional but instructive. It was unusual in its size, speed, and concentration, but the dynamics it documented apply more broadly to large, concentrated shocks of any kind, including those from automation, AI, and climate transition.
Explains
Four concepts that extend the trade-and-inequality framework
Continue exploring international trade and labor-market economics on the MASEconomics blog.
Explore the MASEconomics BlogConclusion
Trade and inequality is not a recent discovery. The Stolper-Samuelson theorem identified the basic mechanism in 1941: the scarce factor loses, the abundant factor gains, and the losses are real reductions in income rather than just relative changes. For most of the postwar period, this prediction sat in textbooks while trade flows remained modest enough that the political stakes were manageable. China’s WTO accession changed the scale. The Autor-Dorn-Hanson paper of 2013 then changed the evidence. Both shifts together moved the trade-and-inequality debate from theoretical possibility to documented reality.
The picture that has emerged is sober but not catastrophist. Trade still raises national output. Consumers and firms still gain. But the distributional consequences are larger, more concentrated, and more persistent than older models predicted. Local labor markets exposed to large trade shocks do not adjust smoothly. Compensation mechanisms have not delivered. The political backlash against trade in advanced economies over the past decade is a direct legacy of these dynamics. Modern trade theory, modern empirical work, and modern policy debates all sit inside the framework that Stolper and Samuelson sketched eighty years ago, given empirical weight by the China shock literature and given urgency by the politics that followed.
Frequently Asked Questions
What does the Stolper-Samuelson theorem say about trade and inequality?
The Stolper-Samuelson theorem predicts that when a country opens to trade, the real income of the abundant factor of production rises and the real income of the scarce factor falls. For an advanced economy abundant in skilled labor, this means trade with developing economies should raise skilled wages and lower unskilled wages. The losses are real, not just relative, even though the country as a whole gains from trade.
What is the China shock in trade economics?
The China shock refers to the rapid rise in Chinese exports to advanced economies following China’s WTO accession in 2001. The term was popularized by David Autor, David Dorn, and Gordon Hanson’s 2013 paper documenting that US commuting zones more exposed to Chinese import competition experienced sharper falls in manufacturing employment, lower wages, and rising unemployment, with effects persisting for over a decade.
Why did the trade-and-inequality consensus shift after 2013?
The 1990s consensus that technology rather than trade explained most of the rise in wage inequality was based on relatively small trade flows. China’s accession to the WTO in 2001 changed the scale of trade exposure dramatically. The Autor, Dorn, and Hanson methodology of comparing local labor markets with different exposure to Chinese imports provided cleaner causal evidence than the earlier aggregate time-series approaches and showed effects much larger than the 1990s consensus had estimated.
Did trade with China cause US manufacturing decline?
Trade with China contributed to but did not fully cause the US manufacturing decline. Estimates from the China-shock literature attribute roughly a quarter to a third of the gross manufacturing employment decline between 2000 and 2010 to Chinese import competition. The remainder came from automation, productivity growth, shifts in consumer demand, and broader structural change. The China shock accelerated and concentrated a decline that was already underway from other forces.
Why did displaced workers not move to other regions?
The China-shock literature found that displaced workers tended to stay in the same commuting zones rather than relocating. Several factors contributed: housing market conditions and home equity losses made moving expensive, family and community ties anchored workers in place, and alternative regions did not absorb workers easily. Many displaced workers transitioned to disability insurance, early retirement, or long-term unemployment rather than to new jobs, and the local communities themselves declined as the manufacturing wage base shrank.
What policy lessons came out of the China shock literature?
The main policy lesson is that the standard “free trade with compensation for losers” framework did not work as advertised. Aggregate gains from trade did not reach the workers and communities bearing concentrated losses, and existing labor-market adjustment programs were too small and narrowly targeted to make a meaningful difference. The literature has supported calls for stronger active labor-market policy, place-based interventions in affected regions, and more attention to the geographic concentration of shock impacts, though there is much less consensus on whether tariffs are an effective response.
Thanks for reading! If you found this helpful, share it with friends and spread the knowledge. Happy learning with MASEconomics
