Deindustrialization
What actually happened to making things in developed economies, why it now matters strategically, and whether reshoring is real or rhetorical.
(down from ~19.5M in 1979 peak; manufacturing output is now higher than 1979 in real terms)
(up from ~5% in 1995; the largest manufacturing country by a wide margin)
(IRA, CHIPS Act, Bipartisan Infrastructure Law over 10 years; partly contested in implementation)
A note on framing. Deindustrialization is one of those topics where the political conversation has often been ahead of the empirical research, and the empirical research has slowly caught up to validate parts of the political concerns while complicating others. This page tries to separate what actually happened to manufacturing in developed economies from the political rhetoric on either side, and to walk through what the post-2020 reshoring efforts have actually produced versus what they have claimed.
What actually happened
Manufacturing employment in developed economies peaked at different times in different countries: roughly 1979 in the US, 1969 in the UK, 1973 in West Germany. From those peaks, employment has declined substantially in most cases. US manufacturing employment fell from about 19.5 million in 1979 to a low of about 11 million in 2010, recovering to roughly 13 million by 2025. The UK fell from about 8 million to about 2.5 million. Germany held up better - from about 8 million to about 7 million - because its manufacturing structure proved more durable.
Two distinct things were happening simultaneously, and disentangling them matters for thinking about policy. First, productivity in manufacturing rose dramatically: factories produced more output per worker each decade. Second, manufacturing employment was lost both to productivity gains (fewer workers needed for the same output) and to offshoring (output moving to lower-cost countries). The relative weight of these two factors has been contested across decades of empirical research.
The "China shock" research by David Autor, David Dorn, and Gordon Hanson documented that the surge of Chinese manufacturing exports after China's 2001 WTO accession produced larger and more concentrated US labour-market effects than mainstream trade theory had predicted. Specific US regions - the upper Midwest, parts of the Southeast, manufacturing-heavy areas - lost employment that did not recover for over a decade. The political effects of this concentration (visible in the 2016 election geography) are a substantial part of why deindustrialization moved from economist-debate to political-priority.
Manufacturing output, in contrast to employment, has not collapsed in most developed economies. US manufacturing output is now substantially higher than at the 1979 employment peak in real terms; output per hour has roughly tripled. Germany's manufacturing output remains a major economic and export driver. The UK has seen manufacturing output growth slow more than the others. The picture is not "manufacturing left"; it is "manufacturing employment fell while output continued to grow, with concentrated regional effects."
Why it now matters strategically
For decades, mainstream economic policy treated where manufacturing happened as a question of comparative advantage and consumer prices. Whatever was lost domestically was made up for by lower-cost imports. The framework that dominated in Washington and most Western capitals from roughly 1990 to 2015 emphasised this view.
Several developments since 2015 have produced a substantial reassessment. The Chinese economic, technological, and military rise made manufacturing concentration in China look more strategically consequential than the comparative-advantage framework had treated it. The COVID pandemic exposed specific supply-chain vulnerabilities (medical equipment, pharmaceutical precursors, semiconductors) that had not been visible in normal times. The 2022 sanctions wave following the Russian invasion of Ukraine demonstrated how supply-chain dependencies could become weapons in geopolitical confrontations. The Chinese 2023-24 export controls on critical-minerals processing and equipment confirmed that the dependency runs in both directions and could be used.
What changed was not the empirical reality of manufacturing geography but the assessment of what that geography meant. The defence-industrial base, the capacity to produce specific items required for national security, became a recognised concern. The argument that "we should not depend on a strategic adversary for items we may need in a confrontation" gained traction across substantial parts of the political spectrum. The question of whether economic efficiency should be the only consideration in industrial location reopened.
This is not a uniquely American concern. The European Union's strategic-autonomy debate, the UK's industrial-strategy revisions, Japan's economic-security framework, and South Korea's K-Chips Act all reflect parallel reassessments. The G7 has institutionalised some elements of this through the Minerals Security Partnership and other coordination mechanisms.
The reshoring effort
The post-2020 industrial-policy turn has been the most ambitious in developed economies since at least the 1970s. The headline programmes:
US CHIPS and Science Act (2022). $52 billion in direct semiconductor manufacturing subsidies, plus $24 billion in tax credits, plus substantial R&D commitments. Major commitments announced from TSMC (Arizona), Intel (Ohio, Arizona), Samsung (Texas), Micron (New York), with partial construction underway by 2026. Implementation has been slower than initial timelines.
US Inflation Reduction Act (2022). Roughly $370 billion over 10 years for clean-energy manufacturing, with content rules favouring domestic and "friend-shored" production. Has driven substantial battery, EV, and clean-energy manufacturing investment, especially in the US Southeast. Some implementation challenges around Treasury rules; some pullback under the new administration as of 2025-26.
US Bipartisan Infrastructure Law (2021). $1.2 trillion over 10 years, with substantial industrial-base components beyond traditional infrastructure. Implementation is uneven; some major projects have advanced, others have been slow.
EU Chips Act (2023). €43 billion mobilised for semiconductor manufacturing in Europe. Major announcements from Intel (Germany, Poland - the German site partially scaled back), TSMC (Germany), and others. Implementation timelines have slipped; Europe's semiconductor share has grown but remains modest.
EU Critical Raw Materials Act (2024). 2030 targets for domestic mining, processing, and recycling of critical minerals. Funding mechanisms partially in place; project pipeline is real but small relative to targets.
Japan's economic-security framework. Substantial commitments to semiconductor (TSMC's Kumamoto fab, Rapidus advanced-node project), battery, and other critical-industry investment. Among the most coordinated of any country's response.
South Korea's K-Chips Act and related measures. Tax credits and direct support for semiconductor and other strategic industries. Built on existing strong manufacturing base.
India's Production-Linked Incentive scheme. Subsidies for manufacturing across multiple sectors, with substantial growth in electronics assembly (especially for Apple supply chain) and pharmaceuticals.
What this combined effort has produced is real - new factory construction announced and underway, employment growth in some manufacturing categories, supply-chain diversification accelerating - but smaller than the rhetoric suggests. The 2023-24 picture was dominated by announcements; the 2025-26 picture is increasingly dominated by implementation challenges, scaling issues, workforce shortages, and political-cycle uncertainty about whether commitments will be honoured by future administrations.
What works and what does not
The empirical record on industrial policy is more mixed than either advocates or critics typically present. Several patterns recur.
What works. Industrial policy has demonstrably worked in specific cases: South Korea's chaebol strategy in the 1960s-80s, Taiwan's semiconductor build-up from the 1980s, Germany's mittelstand and apprenticeship system, Japan's MITI-coordinated industrial development. The cases that worked share several features: sustained political commitment across decades, alignment with existing comparative advantage rather than against it, willingness to let unsuccessful firms fail, focus on a manageable number of sectors, and integration with broader education and infrastructure investment.
What does not work. Industrial policy has failed often enough that no serious analyst dismisses the failure cases. Specific sectors picked badly. Unsuccessful firms protected indefinitely. Political-cycle reversals. Lack of complementary investment in workforce development. The British post-WWII industrial-policy record is full of examples; the US 1970s-80s "industrial policy" debates produced few clear successes; many developing-country attempts have produced limited results.
What is genuinely uncertain. Whether the current US-EU-Japan-Korea industrial policies fall into the "works" or "does not work" pattern is genuinely uncertain. The factors that predict success are partly present and partly absent. Sustained commitment is contested by political cycles. Alignment with existing capability is strong in semiconductors (the US has substantial design capability), weaker in batteries (where Asian firms dominate the existing supply chain). Workforce development has lagged investment in physical capacity. The honest assessment is that we will know whether this round worked in 5 to 10 years; we do not know now.
One specific observation: industrial-policy commitments that have endured for decades (Korean chaebol support, Taiwanese semiconductor strategy) tended to involve broad political consensus across parties. Policies that depend on which administration is in office tend to produce announcement waves followed by implementation slow-downs. The 2025-26 US transition is testing this directly.
The political consequences
Manufacturing-employment loss in concentrated regions has had measurable political consequences across multiple countries. The empirical research has documented connections to specific electoral patterns: the 2016 US election geography correlating with China-shock-affected counties, Brexit support correlating with deindustrialised areas of Northern England and Wales, AfD support correlating with East German manufacturing decline, parts of the French gilets jaunes movement geography matching deindustrialised regions, and similar patterns elsewhere.
What this research does not say is that deindustrialization is the sole cause of these political shifts. Multiple factors interact: cultural-identity concerns, immigration debates, aging populations, declining trust in traditional political institutions, urban-rural divergence. Deindustrialization is one strand among several, but it is one whose contribution to political volatility has been substantially documented.
The political feedback loop is now visible: deindustrialisation produced political pressure for industrial policy, which has produced the post-2020 commitments, which are now being implemented under the political conditions that those earlier shifts created. Whether the implementation produces the kind of regional renewal that would address the underlying political grievances depends on choices that are still being made.
One important asymmetry: the political pressure for industrial policy comes from regions that lost manufacturing; the actual investment under the post-2020 programmes is going substantially to different regions (the US Southeast for batteries, Arizona and Texas for chips). Whether this matches the political demand remains to be seen.
The paths from here
Partial reshoring with structural Chinese dominance persisting
Specific strategic sectors see meaningful diversification (chips, some battery components, some pharmaceuticals). Most manufacturing remains globally distributed with substantial Chinese share. Strategic vulnerability is reduced for highest-priority items but not for the full economic picture. This is the most likely trajectory.
Industrial-policy reversal under new administrations
Political-cycle changes lead to scaling back commitments before they fully take effect. Announcements that were made do not translate into completed facilities. Workforce investments do not catch up. The reshoring effort proves to have been a transient response to specific political conditions.
Geopolitical shock accelerates reshoring
A Taiwan-related crisis, a major supply-chain disruption, or a sharper US-China rupture forces accelerated commitment. Costs rise; prices increase; subsidies expand. The reshoring effort gets the political support it needs through external pressure rather than through patient build-up.
Friend-shoring becomes the dominant pattern
Rather than full domestic reshoring, manufacturing diversification proceeds through trusted partners: Mexico for the US, Eastern Europe for Western Europe, Vietnam and India for Western markets generally. Total non-Chinese capacity grows substantially; full domestic capacity grows modestly. This is what is actually happening for many sectors and may be the practical equilibrium.
Automation reshapes the question
Advances in industrial automation, robotics, and AI-assisted manufacturing reduce the labour-cost component of manufacturing competitiveness. Reshoring becomes commercially feasible without the same workforce questions. The political benefit of reshoring (regional employment) is partly decoupled from the strategic benefit (domestic capacity). This is genuinely uncertain in timing and magnitude.
Workforce becomes the binding constraint
The skilled-trades workforce required for advanced manufacturing has been hollowed out over decades and cannot be rebuilt quickly. Specific projects are delayed or scaled back because qualified workers are not available. Apprenticeship, community-college, and immigration policies become the binding constraint. This is already visible in some current projects.
Where serious analysts disagree
Reshoring is essential for security and is feasible
Strategic dependencies on China and other potential adversaries are genuinely dangerous and have been allowed to grow too large. The post-2020 commitments are substantial enough to make a difference and represent the most serious industrial-policy effort in decades. With sustained execution, meaningful diversification is achievable.
Held by: Robert Lighthizer (former US Trade Representative), parts of the Biden Administration economic team that designed IRA/CHIPS, Daron Acemoglu's recent work on industrial policy, the broader "new industrial policy" academic community. The case has institutional momentum.
The economic costs of reshoring are larger than acknowledged
Mass-market consumer goods produced domestically would cost substantially more than imports. The costs would fall on consumers and would likely produce inflation. The strategic benefits are real for narrow categories but do not justify a broader reshoring effort. The political case for reshoring is overstating the achievable economic outcome.
Held by: mainstream trade economists, parts of the Cato Institute and Mercatus Center, and Adam Posen at the Peterson Institute for International Economics. The case has empirical support on the consumer-cost side.
The China-shock research is the underweighted policy lesson
Even if reshoring is partial and costly, the political consequences of unmanaged deindustrialization are large enough that some industrial-policy effort is justified on those grounds alone. The mid-2010s mainstream consensus that trade adjustments would be smooth was empirically wrong; correcting that error is a major task of current policy.
Held by: David Autor (MIT), Gordon Hanson (Harvard), David Dorn (Zurich), and the broader China-shock research community. Their case has been substantially incorporated into mainstream thinking; the policy implications remain contested.
Friend-shoring is the realistic compromise
Full domestic reshoring is unrealistic for most sectors; full continued dependence on China is undesirable for strategic reasons. The middle path is diversification toward trusted partners, even at modest cost premiums. This is what is mostly happening already, and it is the right framework for evaluating success.
Held by: Janet Yellen (former US Treasury Secretary), Jake Sullivan (former US National Security Adviser), substantial parts of the trade-policy community, the Mexico-focused near-shoring research community. The case has empirical support in actual investment patterns.
The discussion neglects what manufacturing is becoming
Modern manufacturing is increasingly capital-intensive and software-intensive rather than labour-intensive. The "bring back manufacturing jobs" framing assumes a model of factory employment that is partly historical. The realistic policy goal is manufacturing capability, not 1979-style manufacturing employment. Conflating these is producing policy that may achieve neither well.
Held by: parts of the productivity-economics community, Susan Houseman (Upjohn Institute), and a strand of automation-focused research. The case is partly contested - manufacturing employment has not collapsed in countries with strong manufacturing, like Germany - but the trajectory is real.
None of these readings is fully right or wrong. What can be said from the available evidence: deindustrialization in developed economies was driven by a mix of productivity growth and offshoring, with concentrated regional consequences that mainstream policy underweighted; the post-2020 industrial-policy turn is substantial and unprecedented in scale but its results will be measured over years rather than months; full domestic reshoring is unrealistic for most sectors but partial diversification is genuinely happening; the political consequences of deindustrialization helped produce the political conditions for the response; and the workforce, automation, and execution variables will determine more about the actual outcome than the headline financial commitments.
What this means for you
If you live in a deindustrialised region
The post-2020 industrial-policy programmes have produced specific local opportunities, but unevenly. Specific places have seen real factory construction (Phoenix, Columbus, parts of the Carolinas, eastern Tennessee). Specific places have not. Whether your area is in the first group depends substantially on accidents of geography (proximity to clean water, electricity, transport) and existing workforce. Tracking specific announcements and their implementation status is more useful than tracking aggregate political claims.
If you are early in your career
Skilled trades (electricians, technicians, machinists, advanced welders, semiconductor process technicians) are facing substantial workforce shortages that the current industrial-policy efforts cannot address quickly. Apprenticeship and community-college pathways into these fields offer compensation and stability that the conventional college pathway often does not. The supply-demand gap in skilled trades is one of the most consistent findings in current labour-market research.
If you invest in industrial sectors
The current policy environment has created substantial valuations in specific industrial-policy beneficiaries. Some of those valuations will prove justified; others reflect optimism that implementation will not deliver. Reading project-level execution rather than headline announcements is increasingly important for separating durable from transient effects.
If you make consumer choices
"Made in" labels increasingly involve more complex supply chains than they suggest. A car "made in America" may have substantial Chinese content; a phone "designed in America" may have minimal domestic content. Practical individual consumer leverage is modest; awareness of the actual structure helps when reading political claims.
If you vote on industrial or trade policy
The current political moment includes substantial bipartisan support for industrial policy in principle alongside substantial disagreement on execution. Holding policy to its empirical record - did the announced project actually break ground, did it create the projected jobs, did the supply-chain diversification actually happen - is one of the more useful forms of political engagement on these questions.
