SCIENCE
Structural Transformation
Abstract. Structural transformation is the persistent reorganization of an economy's productive activities, labor, capital, capabilities, spatial settlement, and sources of income. Its most familiar manifestation is the changing weight of agriculture, manufacturing, and services, but the discipline also investigates heterogeneity within those sectors and the institutions that permit movement toward higher-value tasks. This monograph constructs a systematic account of structural transformation as an interaction among sector-specific productivity, nonhomothetic demand, relative prices, capital deepening, trade, urbanization, labor mobility, organizational learning, public infrastructure, and distribution. It separates changes in employment shares from changes in real and nominal value-added shares; derives exact productivity-growth decompositions and identifies their accounting and causal limits; reconstructs classical and contemporary theoretical arguments; and assesses contrasting trajectories of industrialization and services-led development. The research brings together historical scholarship, peer-reviewed multisector models, cross-country productivity studies, international statistical methods, and the 2025 United Nations national accounts standard. It examines agricultural productivity gaps, manufacturing convergence, premature deindustrialization, skill formation, migration, informality, gender, spatial concentration, shocks, technological change, and environmental transition. The original analytical synthesis establishes the conditions under which observed reallocation can plausibly indicate durable development: persistent productivity improvement, feasible movement of workers, broad diffusion of capabilities, and verifiable gains in living standards. These are interpretive conclusions drawn from the literature, not claims to a newly created economic theory or an original econometric estimate.
Keywords: Structural transformation; structural change; economic development; labor reallocation; sectoral productivity; agriculture; manufacturing; services; industrialization; deindustrialization; urbanization; nonhomothetic demand; relative prices; economic geography; informality; growth accounting; international development; JEL O14
1. Subject and Analytical Boundaries
Structural transformation studies how the composition of economic activity changes as societies accumulate productive capabilities, reorganize work, and alter the goods and services they produce and consume. At the center is a jointly determined process: sectoral output changes relative to total output; people move among occupations, activities, and locations; capital is installed in new uses; firms adopt unfamiliar technologies; and income and relative prices reshape demand. These processes may reinforce one another, occur at different speeds, or move in opposing directions. The discipline therefore treats sectoral shares as observations to be explained and productivity and welfare as distinct outcomes to be assessed.
The historical organizing case concerns the declining employment share of subsistence and commercial agriculture, the emergence of large-scale manufacturing, and the eventual expansion of services. In the review by Herrendorf, Rogerson, and Valentinyi, this sequence is a central stylized fact for long-run growth, while variation across countries and accounting conventions remains a constitutive part of the evidence. Kuznets's historical lecture emphasizes how modern economic growth transformed the organization of production and urban settlement alongside per capita income. A general definition needs room for industrialization, services-led change, movement between formal and informal enterprises, and transitions driven by energy, information, or demographics.
Several adjacent questions belong to other areas while informing this Science. Aggregate growth asks why average production rises; structural transformation examines the composition of that rise. Trade theory explains specialization across countries; structural transformation tracks the consequences for domestic tasks and workers. Development economics investigates poverty, institutions, health, and the conditions of prosperity; this Science studies reallocation and transformation within that wider setting. Industrial organization examines market power and firm conduct; here those mechanisms matter insofar as they affect economywide allocation. A high technology share or a large services share can coexist with underemployment and inequality. The composition of output is therefore an empirical characteristic rather than an automatic development grade.
The economic unit of analysis also changes with the question. A three-sector national model isolates broad transitions, but an industry panel distinguishes construction, finance, retail, transportation, business services, and manufacturing branches. A labor-market study tracks establishments, workers, qualifications, and moves; a geographic study tracks cities and regions; a household study must include unpaid work and own-use production. Each boundary generates different measured transformation, and a research claim must specify which boundary it uses.
2. Historical Foundations of Structural-Change Science
The early political economists described changing occupations and divisions of labor as consequences of specialization, markets, and technological improvement. Twentieth-century development economics transformed these observations into explicit theories of economic dualism, demographic and structural regularities, and uneven sectoral growth. W. Arthur Lewis's 1954 unlimited-labor-supply analysis placed capitalist accumulation and labor transfers from traditional production at the center of the development process. In its canonical formulation, modern-sector expansion can initially draw on workers whose withdrawal from traditional activities has little effect on output; later, the opportunity cost of labor rises and labor-market conditions change. The model illuminates incentives and capital requirements while leaving open the empirical prevalence of genuinely surplus labor.
Simon Kuznets made long-run structural and social change part of the definition of modern growth. His 1971 Nobel lecture linked rapid increases in production and productivity with shifts in the composition of output, urbanization, knowledge, and institutions. His comparative-historical enterprise offered a research orientation rather than a deterministic timeline that every economy must traverse. Hollis Chenery and Moises Syrquin then examined cross-country regularities in Patterns of Development, 1950–1970, distinguishing typical correlations with income from the particular pathways of countries. Syrquin's later survey of structural-change patterns clarified that observed shares emerge from demand, production, foreign trade, and demographic mechanisms.
Nicholas Kaldor's study of British economic growth helped establish the argument that manufacturing can generate increasing returns, learning, and demand linkages. His propositions stimulated empirical tests, but the relative contribution of manufacturing must be evaluated rather than presumed from its sector label. William Baumol's analysis of unbalanced growth introduced a different long-run tension: when labor-intensive services experience slower productivity growth than technologically progressive sectors, their relative costs and nominal spending shares may rise even without corresponding growth in real output. These traditions explain why a move toward services may reflect development, relative price effects, or constraints within services itself.
Later equilibrium growth models made the historical research program more precise. Instead of representing agriculture, industry, and services as stages, they derive sectoral allocation from preferences, technology, factor intensities, prices, and intersectoral consistency. Empirical researchers subsequently tested those mechanisms with harmonized industry accounts and worker-level evidence. The movement from illustrative history to multisector models and detailed databases improved analytical resolution, but the measurement of informal production and the identification of causal mechanisms remain substantial problems.
3. Basic Concepts and Analytical Vocabulary
A sector is an aggregation of economic activities under an explicit classification. An industry's value added is its gross output less intermediate consumption; an industry's employment counts people or jobs according to the labor dataset's definition. Aggregate labor productivity is real output per worker or per hour. Sectoral labor productivity uses the same numerator and denominator within each sector. Total factor productivity is an inferred efficiency residual conditional on a model of capital, labor quality, production technology, and factor shares. These concepts interact but are not interchangeable.
Structural change describes change in the composition of production or factor use. Structural transformation denotes the cumulative, developmentally significant reorganization of production, labor, capabilities, demand, and institutions. Industrialization generally means expansion of manufacturing capacity and often employment or value added, but a country can increase manufacturing real output while its share of employment declines. Deindustrialization ordinarily refers to a falling manufacturing share; its interpretation depends on the numerator, denominator, stage of development, and the distinction between real output and nominal share. Rodrik's premature-deindustrialization analysis deliberately compares the levels of income at which manufacturing shares peak.
Reallocation is the movement of labor or other resources between activities. An employment-share change can reflect workers changing sectors, differential recruitment of new workers, retirement, entry into the labor force, or measurement recoding. Upgrading describes movement toward higher productivity, greater skill intensity, more sophisticated tasks, or better-quality output; these dimensions can diverge. Economic diversification increases the variety of productive activities or export products, although variety alone does not demonstrate efficient scale. Formalization changes the institutional organization of production and can move previously unrecorded activity into measured output without an equivalent change in underlying goods or work.
The distinction between structural change and structural transformation is useful as an interpretive convention, but it does not create a rigid universal taxonomy. Researchers sometimes use the terms synonymously. Throughout this work, statistical statements specify the observed variable and the conceptual term identifies the broader process. This convention prevents a decline in agriculture's nominal GDP share from being mistaken for proof that agricultural workers achieved higher incomes, or a rising market-service share from being taken as proof of technological progress.
4. National Accounts and Sectoral Classification
National accounts provide the principal consistent measure of sectoral value added. A calculation begins with domestic value added classified by economic activity and reconciled to national totals under an accounting standard. The SNA 2025 framework was adopted by the United Nations Statistical Commission in 2025 as an update to the 2008 framework; it extends statistical treatment of globalization, digitalization, emerging financial issues, and the informal economy while retaining the broad accounting structure. Adoption of the standard and actual implementation across countries occur on different schedules. Historical series therefore demand metadata about national revisions, chain linking, price bases, and classification concordances.
Activity classification is another fundamental design choice. The UN's ISIC Revision 5 classification was adopted in 2024; countries using earlier revisions must map activities carefully before comparison. A third-party logistics firm, an internal manufacturer's software division, and an outsourced accounting contractor may perform connected production tasks but appear in different industries. Outsourcing can increase measured business-service employment and reduce manufacturing employment while leaving the physical production network largely intact. Such changes are economically meaningful as organizational restructuring, yet their contribution to technical productivity growth must be evaluated separately.
The ETD sectoral database constructed by UNU-WIDER and the Groningen Growth and Development Centre provides a concrete example of harmonization. Its documented version covers 51 economies, twelve sectors, annual observations for 1990–2018, real and nominal value added, deflators, and employment. These 51 economies represent a defined empirical sample, not a world census or a contemporary 2026 panel. The ETD methods note details country-specific source construction from national accounts, censuses, surveys, and other records. Comparable columns should be interpreted in light of underlying differences in enumeration and statistical capacity.
Comparisons between countries also require decisions about exchange rates and purchasing power. Value added at current local prices permits within-country nominal composition and time-specific sector shares, whereas cross-country levels require a coherent currency conversion or purchasing-power measure. A constant-price series measures volume in the chosen reference framework, which may alter the composition of output compared with current-price GDP. Chain-volume aggregates may not be additive away from the reference period. An analyst who reports 'the agricultural share' without identifying whether it is employment, nominal value added, or a particular real series leaves an essential part of the proposition unspecified.
5. Employment, Production, Prices, and Measurement
The production structure can be observed through employment shares, hours shares, nominal value-added shares, real value-added contributions, establishments, exports, and capital stock. These measures answer different questions. An increase in agriculture's physical harvest accompanied by lower agricultural prices can reduce the sector's nominal value-added share. An increase in hospital employment with slow measured output growth may reflect expansion of valuable labor-intensive care, measurement challenges, or low productivity growth. Sector shares must be analyzed together with levels, relative prices, intermediate inputs, and labor utilization.
Labor series may distinguish persons employed, jobs, wage earners, labor force participation, full-time equivalents, and hours. The ILOSTAT labor data help researchers compare employment across economies but depend on survey instruments and national coverage. Informal household enterprises, unpaid family labor, multiple-job holding, own-use farming, and seasonal work complicate counts. Value added per person engaged can differ from value added per hour even where the underlying technology is unchanged; hours and labor quality are therefore necessary sensitivity checks.
Price measurement is particularly consequential for services. Falling prices for telecommunications or information processing can be difficult to capture when quality changes rapidly. Housing services require imputation conventions; public administration and some nonmarket services rely on input-cost methods. Some products are bundled with manufacturing and recorded in services after outsourcing. The introduction of statistical improvements may generate apparent breaks in productivity. Each sector's nominal share combines its real volume and price relative to the total; a rise in one share cannot be assigned to demand or technology without a model or additional data.
Consider two occupations with constant physical output. If their relative wages rise for institutional or scarcity reasons, a nominal-value-added comparison may change even though aggregate physical output remains fixed. Conversely, an industry can lose jobs as its automation raises output per worker. It is possible for manufacturing value added to rise in real terms while employment falls and nominal GDP share contracts. This is why assertions about 'industrial collapse' require explicit dates, denominators, measurement units, and a distinction between firm closures and productivity-driven labor substitution.
6. Aggregate Productivity and Exact Decompositions
Let sector index i identify mutually exclusive activities at two dates t=0 and t=1. For an additive real-value-added measure Y, let L denote employed persons or total hours, let s_i=L_i/L be each sector's labor share, and let p_i=Y_i/L_i be its real value added per unit of labor. Aggregate labor productivity is P=Y/L. The accounting identity is
P_t = Σ_i s_(i,t) p_(i,t).
This identity describes how the employment-weighted average of sectoral labor-productivity levels generates aggregate productivity. It requires internally consistent sector and total denominators, comparable sectors over time, and an additive real output measure. Where chain-linked volume accounts are nonadditive, an analyst must use a valid decomposition technique or disclose any approximation rather than silently impose the identity on incompatible series.
Write Δx=x_1−x_0 and average x̄=(x_0+x_1)/2. The exact midpoint decomposition for the product of two changing factors is
ΔP = Σ_i s̄_i Δp_i + Σ_i p̄_i Δs_i.
The first term is commonly called the within-sector productivity component; the second is the between-sector or reallocation component. The symmetric midpoint assigns half of each interaction term to each component. The exact base-year alternative is ΔP=Σ_i s_(i,0)Δp_i + Σ_i p_(i,0)Δs_i + Σ_i Δs_iΔp_i. These are accounting decompositions: no unobserved causal effect of labor mobility has yet been identified. Empirical studies such as Timmer and de Vries's sectoral analysis and McMillan and colleagues' comparative decomposition use variations of shift-share accounting to compare countries and time periods.
An illustrative economy has eighty agricultural workers producing two units each and twenty manufacturing workers producing eight units each. With one hundred workers, output is 320 and productivity is 3.2. If ten workers move from agriculture to manufacturing while sectoral productivity remains constant, output becomes 380, and aggregate productivity becomes 3.8, an 18.75 percent increase. The between component is 0.6; the within component is zero. These numbers are a transparent arithmetic example, not estimated historical effects. Its assumptions ignore wages, skills, land, capital, congestion, demand, and the adjustment of productivity to reallocation. In observed economies, these variables frequently respond to the move itself.
A positive between component can arise even when displaced workers obtain only marginally better jobs, if average sectoral productivity differs from the productivity of marginal workers. Conversely, a negative between component can coexist with real wage gains when labor shifts into activities with better hours, safety, social protection, or work-life conditions. A growth decomposition is valuable because it specifies the compositional contribution under an accounting convention; it cannot substitute for a welfare assessment or a structural model of counterfactual labor flows.
7. Interpretation of Productivity Gaps
A sectoral labor-productivity gap is the difference or ratio between measured value added per worker or hour across industries. Agriculture's low measured output per worker in many developing economies motivated the claim that labor transfers may raise aggregate productivity. The agricultural productivity gap study by Gollin, Lagakos, and Waugh estimates large cross-country and cross-sector gaps and examines corrections for hours worked and other measurement issues. Their companion cross-country agricultural comparison emphasizes the considerable productivity disparities among national agricultural sectors. The gaps identify an important research problem, but raw ratios are not direct estimates of what an individual worker would produce after moving.
Differences in capital per worker, land, intermediate inputs, human capital, seasonality, prices, worker selection, and transport costs can all influence measured averages. Lagakos and Waugh's selection analysis demonstrates the importance of occupational selection for observed cross-sector productivity differences. Workers choosing low-wage rural employment may face credit constraints, household responsibilities, risk, or location-specific advantages that are invisible in the sectoral average. Marginal productivity after migration may be below a destination sector's current average, especially when infrastructure or capital becomes congested.
Correct interpretation begins by separating the accounting gap, a potentially attainable earnings gap, the marginal product under feasible relocation, and the full social return after congestion and public costs. Establishing an economywide welfare gain requires prices, worker preferences, transition costs, general-equilibrium wage adjustment, public finance, and the welfare of those remaining in the origin sector. Where the gap reflects inefficient barriers, improved transport, land institutions, skill acquisition, and credit can be productive. Where it reflects fixed factors or selection, an estimated mechanical gain will exaggerate the true opportunity. Structural transformation research gains precision by keeping these quantities analytically distinct.
8. Agriculture, Food, and the Demographic Transition
Agricultural development makes structural transformation feasible by changing both the supply of workers and the conditions under which they can be employed elsewhere. In low-income economies, food typically occupies a large share of household spending, so the availability and real cost of food influence the wage and consumption requirements of urban labor. Higher yields can free land and working time, generate rural income, reduce food-price pressure, and support demand for nonfarm goods. Gollin, Parente, and Rogerson place agriculture's productivity constraints within a growth account in which food requirements and cross-sector factor allocation determine the pace of development.
The mechanism has several variants. Land-augmenting technologies may raise output without reducing rural employment immediately; labor-saving mechanization can produce labor displacement and pressure for migration; better irrigation and storage may stabilize both output and incomes. Agricultural commercialization also increases demand for credit, transport, warehousing, repair services, packaging, and processing. These connections mean the transformation of agriculture frequently generates new activities classified outside agriculture. An exclusive emphasis on the declining agricultural employment share would omit these supply-chain effects and the distribution of rural gains.
The role of labor surplus deserves particularly careful treatment. Lewis's dual-economy argument suggests that modern-sector demand can absorb labor without an immediate collapse in traditional output when the marginal contribution of some workers is small. Seasonal agriculture, gendered household work, dispersed plots, and the concentration of peak labor demand make the empirical assumption difficult to test. A worker who appears underemployed during a dry season may be essential at harvest, and an urban job may offer higher annual pay while exposing the household to housing and health risks. Migration decisions therefore reflect calendars, risk diversification, and household bargaining as well as average pay.
Demographic change influences this process through population growth, age composition, educational attainment, and the timing of fertility decline. A growing working-age population creates potential for production and makes job creation urgent. Falling dependency ratios can support savings and investment, but the benefit depends on demand for productive labor. The transition from large rural households to smaller urban households also changes household production, market purchases, and demand for education and health services. Structural transformation and demography are mutually conditioned; neither follows mechanically from a national income threshold.
A successful agricultural transition preserves the capacity to produce or procure affordable food, raises the productivity of those remaining in agriculture, and provides viable nonfarm opportunities for those leaving. A dramatic exit from farming alongside stagnant agricultural yields may threaten rural livelihoods, food security, or trade balances. The economic importance of agriculture can persist even as its share in GDP diminishes, because agricultural productivity anchors food costs, land use, and upstream and downstream linkages.
9. Manufacturing, Scale, and Industrial Learning
Manufacturing's historical significance rests on several identifiable economic properties: capital deepening, repeatable processes, product standardization, learning-by-doing, opportunities for exporting, and backward and forward linkages with materials and services. Standardized production can permit workers to enter with limited formal schooling and acquire skills on the job. Some manufacturing activities support large establishments and scale economies; others are small, labor intensive, or dependent on imported designs. The sector's advantages therefore vary sharply across industries, places, and technologies.
The Kaldorian approach places manufacturing expansion at the center of cumulative demand and productivity growth. Kaldor's growth study frames industrial production as a potential source of increasing returns and external economies. Productivity increases can make products more competitive, support exports, and expand the market in which learning occurs. Such feedback depends on viable demand, macroeconomic stability, access to inputs, and the capacity to maintain quality. Manufacturing is not a uniform technology; mining, food processing, garments, electronics, automotive assembly, and precision instruments have different capital intensities, spillovers, energy needs, and employment elasticities.
Evidence of manufacturing productivity convergence across countries suggests that certain organized manufacturing activities can adopt established production methods and reduce labor-productivity differences even where the national economy remains heterogeneous. The result concerns manufacturing establishments or sectors in the data, not universal convergence for all workers or a guarantee that manufacturing's share will expand. New technologies may raise output per employee without raising manufacturing employment. A country may increase manufacturing export volume, diversify products, and improve technical competence while seeing a smaller manufacturing employment share.
Industrial upgrading involves more than assembling imported intermediate goods. It requires maintenance competence, engineering, supply-chain coordination, reliability, process control, domestic supplier development, and the ability to satisfy demanding purchasers. The value retained domestically differs from gross exports because imported components, intellectual property fees, and foreign-owned services may account for substantial value. Measuring industrial transformation thus requires domestic value added, productivity, wages, workforce skills, innovation, and the durability of domestic production networks.
The distinctive advantage of manufacturing in one historical setting need not establish a universal sequencing rule. Economies with small populations, high transport costs, constrained electricity, or differentiated skills may find niches in processing, services exports, or tourism. Their challenge is to identify tradable or locally productive activities that generate learning and widely accessible jobs. The legitimate empirical question is which production networks create durable productivity and earnings gains under each economy's constraints.
10. Heterogeneous Services and Economic Modernization
Services are a very large collection of distinct production activities. Wholesale and retail trade, passenger transport, public administration, telecommunications, software, finance, education, hospitality, health care, logistics, cleaning, and professional consulting share neither one production function nor one labor-market structure. Their output may be consumed immediately, embodied in goods, traded remotely, or supplied through public institutions. Aggregating them into a single sector can conceal the central transformation question: where within services are productivity and incomes rising, and who can enter those activities?
Baumol's unbalanced-growth argument explains one reason nominal service spending may rise as economies become richer. Where high-contact activities need fixed time per unit of service, wage increases driven by productivity elsewhere push up relative costs. In contrast, digital communications and certain standardized business services can achieve dramatic productivity improvements through replication and network infrastructure. Demand for care can also rise with longevity and income. Service expansion may therefore reflect consumer income effects, relative prices, labor demand, outsourcing, demographic change, and technological upgrading simultaneously.
The World Bank analysis At Your Service? distinguishes services according to tradability, scale economies, innovation, and the skill levels needed to enter. Digitally deliverable services can reach distant buyers, while local retail and personal services largely depend on nearby demand. High-productivity business services complement manufacturing by improving management, logistics, financing, design, and distribution. Manufacturing can in turn generate demand for specialized services. The relation between the sectors is fundamentally interdependent, even when national accounts place their final output in separate columns.
Employment in a rapidly growing service industry does not automatically indicate that sector's marginal employment opportunities are well paid. Informal retail and casual transport may absorb workers because industrial and formal-service employment is insufficient. Regulated professional services may pay well but absorb relatively few people without substantial education. Public services may create essential welfare benefits yet be difficult to value using conventional market-output measures. A credible services-led development assessment therefore reports subsector productivity and job quality, domestic linkages, export potential, digital infrastructure, skill requirements, and financing needs.
A country's services share can also increase through changes in organization rather than only final demand. Outsourcing payroll, inventory management, cleaning, logistics, and data processing shifts employment between company and industry classifications. Such reclassification may still signify specialization and a new business ecosystem; it cannot be interpreted as identical to newly created output. The research question is how the reorganization affects productivity across the entire production chain.
11. Income Effects and Changing Consumption Patterns
As income rises, households generally devote a smaller fraction of their budgets to basic food and can purchase a wider range of goods and services. Because the proportion of spending allocated to different commodities varies with income, growth changes the composition of final demand. Nonhomothetic preferences provide a formal way to represent such variation. The Beyond Balanced Growth model shows how demand shifts can generate substantial sectoral movement while aggregate growth follows regular patterns. This explanation treats changes in consumption as an endogenous force rather than a passive consequence of investment.
Demand composition depends on more than income. Population aging can increase care needs; falling household size may raise expenditures on housing and market services; urbanization can expand commuting, education, and retail; changes in household technologies can replace market purchases with home production or the reverse. Relative prices determine the affordability of alternative forms of consumption. The same per capita income level can support different output structures when family arrangements, public provision, and infrastructure differ.
Consumer demand and domestic production must also be separated in an open economy. Higher domestic demand for manufactured goods can be met by imports, whereas services are partly purchased from local suppliers. A country's agriculture share may fall as its consumption of food remains high but processing and distribution expand, or as it imports food. A service share may rise because domestic households demand health and education, because local businesses export software, or because the government expands essential infrastructure. Their consequences for productivity, trade balances, and employment differ.
The interaction with inequality is especially important. Sectoral demand does not respond only to average per capita income; it responds to the distribution of household purchasing power. An increase concentrated at the top can generate demand for specialized high-end services while a broad increase in earnings may expand mass-market consumption and formal retail. Structural transformation therefore depends on the distribution and composition of demand as well as on technological feasibility.
12. Sectoral Productivity, Prices, and Capital Intensity
A multisector economy contains production functions with different rates of technical improvement and different factor intensities. In the Ngai–Pissarides growth model, differences in sectoral productivity growth alter relative prices and can generate changing sectoral employment even along a regular aggregate growth path. Industries with rapid productivity growth may experience falling relative prices and changing resource shares, depending on substitution possibilities and the structure of preferences. Higher productivity can coincide with reduced employment in the same sector if demand expands less than output per worker.
Acemoglu and Guerrieri's analysis adds capital-intensity differences: aggregate capital accumulation changes relative sector costs and contributes to nonbalanced growth. A capital-intensive sector may gain a cost advantage when capital deepens, even without a sector-specific technological shock. The allocation of labor depends on whether households and firms substitute between sectoral products or regard them as complements. Labor, intermediate inputs, infrastructure, and trade policy also affect sector-specific cost structures.
Relative price effects make nominal and real shares diverge. Suppose a service becomes more expensive relative to manufactured goods while quantities grow more slowly. Its nominal value-added share can rise, even as its real-volume contribution does not rise proportionately. Conversely, rapidly improving digital goods may account for an expanding quantity of useful services while representing a small nominal share because their prices decline. Boppart's joint price-and-income analysis demonstrates how both relative price changes and preferences can jointly explain structural change in a model consistent with observed growth regularities.
These models also clarify why national-income-based predictions are incomplete. Two countries with similar average income can have different structures if their capital supply, relative productivity, trade access, tastes, population, and institutions differ. The appropriate empirical comparison identifies which model parameters or shocks would produce the observed sector shares and tests the implied predictions for prices, quantities, wages, and intermediate inputs. A model calibrated only to employment shares risks mistaking one mechanism for another.
13. General Equilibrium and Multisector Growth Theory
General equilibrium imposes consistency across households, firms, factors, and markets. Households demand goods and supply labor; firms choose inputs given technology and prices; capital accumulation responds to savings and investment; and factor prices clear markets or reflect specified frictions. A multisector theory explains why resource shares change while respecting aggregate accounting and the feasibility of production. the Handbook of Economic Growth synthesis organizes this literature around sectoral demand, relative prices, and production structures.
The nonhomothetic-demand tradition derives structural change from the responsiveness of consumption to income. The differential-productivity tradition derives it from sectoral cost and price trends. Capital-intensity models add differential responses to accumulation. Buera and Kaboski's scale analysis extends the discussion to specialization, market production, and the changing organization of work. Many observed transitions require more than one mechanism. A household may move from own-production to retail purchases as income rises while a retailer substitutes digital systems for clerical labor; both changes alter counted sectoral activity.
Introducing an open economy changes the solution. World prices and foreign demand influence which goods can be produced competitively, and imports relax domestic supply constraints. Input-output connections transmit shocks across sectoral boundaries. A rise in manufacturing output can expand business services and transportation; automation in one sector can alter relative wages and consumption elsewhere. Aggregate employment changes depend on entry and exit, aggregate demand, and labor-market rules. A theory that fixes aggregate employment may identify allocation but cannot resolve underemployment or the welfare implications of lost jobs.
A model is scientifically useful when its assumptions make predictions that can be contradicted by data. Does a proposed income effect imply a particular consumption trajectory? Does a productivity explanation imply a measurable relative price trend? Does capital deepening predict a systematic response in capital-intensive industries? Do frictional labor markets imply slow adjustment after a shock? Model pluralism is appropriate where different mechanisms operate; the empirical burden is to identify their relative importance and limits, not to proclaim a single mechanism universal.
14. Trade, Globalization, and Economic Complexity
International exchange makes structural transformation sensitive to external demand, comparative costs, imported technology, and the governance of production networks. Trade can accelerate industrial learning by exposing firms to demanding buyers and expanding the scale of a market. It can also intensify competition, displace local workers, and favor sectors that already use a country's most abundant resources. The domestic outcome depends on which tasks become viable, how rapidly workers can adjust, and whether new activities retain and diffuse capabilities.
the globalization and productivity comparison by McMillan, Rodrik, and Verduzco-Gallo documents different historical contributions of structural change across regions and periods. Where labor moves toward high-productivity tradable sectors, aggregate productivity may rise rapidly; where export booms are capital intensive or displaced workers turn to low-productivity informal services, the aggregate effect can be weaker. Commodity exports can raise national income and public revenue without absorbing much labor. Exchange-rate appreciation during resource booms may reduce competitiveness elsewhere, affecting sectoral development through price channels.
The finding of manufacturing's cross-country convergence is especially relevant to globalization because it suggests potential productivity gains from adopting established industrial technologies. Yet the ability to gain market share and the capacity to absorb workers are separate. Supply chains may allocate design, high-value components, assembly, distribution, and after-sales services across different economies. Gross export value exaggerates local production when imported inputs are substantial; domestic value added, learning, wages, and technology transfer provide more informative measures of transformation.
Trade-induced dislocation carries costs. Artuç and colleagues' adjustment analysis models workers' costly transitions across sectors after shocks, and Autor, Dorn, and Hanson's local-labor-market study demonstrates persistent geographic effects from import competition in the United States. These results caution against interpreting aggregate consumer gains as sufficient evidence that all affected workers can quickly find comparable employment. A dynamic transformation strategy must consider worker insurance, retraining, mobility constraints, local fiscal capacity, and policies that attract alternative employers.
Economic complexity in the ordinary sense of the term refers to the coordination of interdependent specialized activities, technologies, and skills. A country's capacity to export complex manufactures or sophisticated services reflects more than one successful firm; it requires reliable inputs, production knowledge, standards, and logistics. Whether an observed export diversification yields broad development depends on the diffusion of those capabilities to domestic suppliers and workers. The composition of trade is informative, but its developmental significance must be traced into national accounts, learning, and livelihoods.
15. Cities, Migration, and Spatial Transformation
The movement of workers from agriculture to nonfarm activities frequently accompanies movement from rural areas toward towns and cities. Density supports matching among employers and workers, learning through proximity, specialized suppliers, and access to consumers. Urbanization can reduce the cost of delivering certain services while increasing pressure on land, transport, sanitation, and housing. A productive city requires complementary investment in public services, not only an inflow of workers.
The Harris–Todaro migration model demonstrates that rural workers may move to cities even when urban unemployment exists, because decisions depend on expected earnings and the probability of securing formal employment. Urban wage premiums can reflect productivity, regulation, compensating differentials, or selection. When the expected gain is offset by uncertain employment and high living costs, migration may not raise household welfare. Informal employment, family networks, and circulating migration complicate a simple rural-to-urban progression.
Spatial restrictions can prevent workers from reaching productive firms. In Hsieh and Moretti's housing analysis, constraints on housing supply influence the spatial allocation of labor and, in their model-based estimates, aggregate productivity. The exact aggregate counterfactual depends on estimated elasticities, migration responses, and housing mechanisms; it should not be presented as a directly measured causal gain. The insight is more general: large measured productivity gaps between cities and regions do not automatically disappear when workers wish to move, because housing and public infrastructure can limit access.
The geography of industry can also reverse. Deindustrialized cities may retain accumulated human capital but lose specialized job networks; coastal export zones may flourish while inland districts remain agricultural; logistics corridors may concentrate employment. Regional structural transformation therefore requires geographically disaggregated data. National averages can rise while some localities suffer prolonged wage loss or out-migration. World Development Report 2023 places migration within a framework of matching, rights, and institutions that helps contextualize these flows beyond a narrow production statistic.
An analytical treatment of urbanization joins migration probabilities, worker skills, employment opportunities, housing, land governance, transport times, social protection, and municipal finance. Economic integration is spatial as well as sectoral. The relevant policy objective is an affordable and productive pathway to jobs and public services rather than an arbitrary target urban-population percentage.
16. Firms, Misallocation, and Productive Organization
Sectoral averages are the outcomes of heterogeneous firms. Within the same industry, plants can differ substantially in technology, managerial capability, scale, access to finance, markups, and labor quality. Entry by productive firms, expansion of efficient incumbents, and exit of persistently inefficient firms can change industry productivity without a worker moving across broad sector categories. An economy can undergo significant productive transformation within manufacturing or retail while its three-sector employment shares barely change.
Hsieh and Klenow's misallocation study examines dispersion in marginal revenue products among manufacturing establishments in China and India and uses a model to quantify potential gains from reallocation. Their estimated counterfactual gains depend on the model's assumptions about technologies, distortions, and output demand. Revenue productivity dispersion may reflect genuine differences in market power, quality, adjustment costs, or measurement errors in addition to wedges. The study nevertheless establishes the importance of internal resource allocation for development analysis.
The institutional circumstances of firm entry and scale are essential. A producer may possess a viable technology but remain small because of unreliable electricity, missing titles, costly credit, inaccessible markets, regulatory complexity, or shortages of managerial skill. Credit reform can improve investment allocation, but under weak competition it can also entrench connected incumbents. Public procurement can support new suppliers while raising risks of dependence and capture. Improvements in roads and digital payments can expand market size, reduce transaction costs, and change the composition of firms without changing a sector's nominal share immediately.
Structural transformation has a corporate dimension in the organization of production chains. Internalizing logistics, outsourcing software, contracting independent distributors, and creating specialist suppliers can increase task specialization. National accounts register the industries of establishments, whereas capabilities reside in workers, teams, routines, and relationships that cross those boundaries. Consequently, firm-level analysis should supplement sectoral accounting with measures of entry and exit, plant productivity, wages, market shares, capital investment, supplier links, and task content.
The welfare consequences of exit require attention. Reallocation through the closure of unproductive plants may raise aggregate output per worker but impose transitional earnings losses, destroy location-specific skills, and damage communities. Better competition and mobility policies can increase efficiency while social insurance and place-based interventions address adjustment costs. Productivity and transition losses should both enter policy evaluation.
17. Human Capital, Labor Markets, and Gender
Human capital conditions the direction and pace of structural transformation. Workers require literacy, numeracy, technical abilities, language, digital competence, and occupation-specific knowledge to enter many activities. Firms require managers who can organize production, assess quality, and absorb imported technology. A large pool of workers with limited formal schooling may support labor-intensive manufacturing, but specialized health, engineering, and information services require longer training and institutions that certify competence. Education policy changes both supply-side capabilities and the wage structure of employment transitions.
Matching is costly. Workers may be unable to relocate because of family obligations, housing costs, legal restrictions, or the risk that a new job will disappear. Firms may not train workers if they cannot appropriate the returns. the trade-adjustment evidence demonstrates how switching costs complicate movement between industries. These frictions mean observed aggregate productivity differences may persist long after markets open and new technologies become available. Public employment services, portable benefits, apprenticeships, and credible qualifications can reduce those costs when designed around actual labor demand.
Gender is part of this transformation rather than an auxiliary social variable. Claudia Goldin's historical analysis of female participation explores why measured female labor-force participation may follow a U-shaped relationship with economic development as home production, industrial work, education, and norms change. This stylized pattern is context-dependent: it is affected by statistical treatment of family work, household wealth, occupations available to women, and norms governing market employment. Greenwood and coauthors' household-technology analysis connects durable household innovations to the expansion of female labor-market participation. These mechanisms remind researchers that shifts between unpaid and market work can alter measured employment without an equal change in total socially valuable work.
Productivity-enhancing changes can impose uneven burdens across demographic groups. Young workers may enter new sectors more readily than older workers with industry-specific skills. Migrants may face credential barriers. Women's employment opportunities can depend on safe transportation, childcare, work schedules, land rights, and equal access to education and finance. A policy that succeeds on sectoral productivity averages while excluding large groups from productive employment has a narrower welfare impact than its headline growth accounting suggests.
18. Institutions, Infrastructure, and Public Capacity
Markets for land, labor, capital, energy, and information operate within legal and administrative arrangements. Property rights influence investment and the transfer of land; contract enforcement supports complex supply chains; banking access permits firms to finance equipment and working capital; reliable electricity and water determine which industries can operate at competitive cost. Infrastructure changes the feasible set of economic activities. Roads and ports can make rural production marketable and connect factories with input suppliers, while communications infrastructure enables some services exports.
Structural transformation is especially sensitive to coordination failures. A manufacturer may require a trained workforce, continuous power, transport, and specialized suppliers at the same time; none of the individual investments is profitable until enough other investments occur. Governments can address missing public goods and information failures through infrastructure, training systems, standards, and appropriately designed coordination measures. Structural Change, Fundamentals, and Growth brings together case studies that examine how fundamentals and reallocation interact across developing economies.
Policy quality depends on incentives and accountability. Industrial promotion can help discover competitive activities, yet conditional support may become permanent protection for firms that fail to improve. Export promotion can generate learning while increasing fiscal exposure or dependence on one buyer. Subsidies justified by knowledge spillovers require a plausible transmission mechanism and a credible evaluation strategy. Infrastructure investments require maintenance and the capacity to manage land, environmental, and procurement risks. Governance is therefore part of the economic mechanism, not an administrative afterthought.
The 2024 World Development Report interprets middle-income development through the successive capacities to invest, absorb external technologies, and innovate. Its organizing framework resonates with structural transformation, but broad policy labels need operational definitions in each setting. Investment without productive use, imported technology without learning capacity, and innovation without diffusion can increase expenditure while leaving productivity weak. Institutions matter because they determine whether economic resources produce durable capabilities, competitive markets, and broadly distributed gains.
19. Informality, Small Enterprises, and Household Production
Informal production accounts for substantial employment and value creation in many economies. Small household farms, retail stalls, repair workshops, transport operators, domestic services, and unregistered manufacturers frequently operate with limited access to finance, social insurance, reliable infrastructure, or enforceable contracts. Informality may reflect regulatory barriers, low productivity, risk management, or the economic feasibility of very small-scale enterprise. Its causes and consequences differ among occupations and national legal systems.
The transition from household production to purchased market services can alter measured GDP and employment, even where the total service experienced by the household changes little. the scale-and-specialization account highlights how the organization of home and market production contributes to structural change. If a household begins purchasing meals or childcare previously produced at home, market-service output rises, and the measured sectoral distribution changes. That change may free time for paid work and improve specialization; it should still be distinguished from the physical creation of entirely new consumption.
Informality also complicates industrial analysis. Small workshops and microfactories may not appear in large-establishment manufacturing surveys, and conventional datasets can therefore overstate manufacturing decline when production shifts toward smaller entities. the manufacturing renaissance study provides evidence that developing-country manufacturing trajectories depend on how sectoral coverage and the evolution of establishments are assessed. Contrasting evidence requires harmonized definitions, rather than simply choosing either a premature-deindustrialization or a renaissance narrative.
Formalization is a multidimensional outcome. Firm registration alone does not ensure increased productivity, safety, protection, or survival. A successful transition may combine better electricity and transport, access to financing, simplified tax and registration procedures, public procurement, accounting systems, and worker protections. In some contexts, informal production provides a livelihood buffer after macroeconomic or climatic shocks. Policies should therefore examine the real capabilities and incomes of workers and firms and avoid equating a legal status change with a complete productivity transformation.
Data quality requires triangulation among household surveys, establishment censuses, labor force surveys, administrative registers, and national accounts. ETD's documented methods illustrate the importance of harmonizing source coverage before cross-country comparisons. Improvements in statistical coverage can make measured services or small-scale manufacturing expand even when actual economic activity changes gradually. Historical analyses must flag such breaks.
20. Industrialization and Development in East Asia
East Asian experiences demonstrate a range of pathways involving agricultural improvement, industrial exports, education, urbanization, technology transfer, and expanding producer services. Economies such as Japan and the Republic of Korea experienced sustained increases in manufacturing capabilities before high-end services assumed a larger economic role. Later industrializers, including China and parts of Southeast Asia, entered global production networks under different international conditions. Historical differences in market access, political institutions, demographic profiles, and global technology prohibit a single explanatory template.
Manufacturing expansion often created opportunities for standardized wage employment, supplier specialization, and international learning. But the pace of employment absorption depended on labor intensity and technology. Export-oriented assembly can enter global markets rapidly while retaining a relatively small share of final value added. Upgrading into domestic design, components, process engineering, management, and advanced logistics can increase the local gains from participation. the manufacturing convergence evidence supports the proposition that existing industrial technologies can permit rapid productivity improvements in particular sectors; this is not proof that all industrial export strategies achieve the same national outcome.
Structural change in a given economy may coexist with remarkable within-sector productivity gains. Diao, McMillan, and Rodrik's comparison emphasizes that recent growth episodes cannot universally be explained by the simultaneous operation of favorable reallocation and rapid sectoral productivity. An agricultural exit may happen during industrial employment expansion, while a later productivity acceleration may arise mainly from technological improvement within sectors. Sectoral shares and firm data must therefore be matched by historical period.
The broader lesson concerns the combination of productive absorption and upgrading. A labor-intensive export industry can absorb new entrants while transmitting managerial and technical knowledge, but sustained growth requires progress beyond initial cost advantages. Urban housing, education, infrastructure, domestic competition, and the treatment of displaced workers influence the distribution of gains. Comparative evidence is most persuasive when particular policy interventions are tied to documented capabilities and outcomes rather than when entire national growth histories are assigned to a single strategy.
21. Latin American Paths and the Reversibility of Change
Latin American economies exhibit long histories of urbanization, industrial development, commodity specialization, repeated macroeconomic crises, and changing trade and industrial regimes. Their experiences illustrate that structural transformation can stall, reverse, or take forms that deliver weak productivity performance. Manufacturing can contract in relative terms after tariff liberalization, exchange-rate shifts, debt crises, or technological change, while services employment absorbs displaced workers at varied productivity levels. The relevant comparisons must respect wide differences among countries and decades.
Timmer and de Vries's regional dataset was designed to investigate structural change and growth accelerations in Asia and Latin America. Its contribution is methodological as well as substantive: the pace of aggregate growth cannot be understood solely through sectoral shares without considering the productivity trajectory within each activity. The globalization and structural-change study associates some historical Latin American episodes with labor movements that were less favorable to productivity growth than observed in parts of Asia. This comparison is period-specific and does not imply that every Latin American country followed the same path.
Commodity booms create a distinctive challenge. A highly productive mining or agricultural-export sector can generate foreign exchange and fiscal resources while employing relatively few people. Appreciating exchange rates and rising domestic costs may make labor-intensive tradables less competitive. Services associated with commodity exports can expand, but their quality and resilience depend on domestic linkages. Investments in transportation, technical education, and diversified suppliers can transform a temporary export windfall into productive capabilities; failure to establish such linkages can leave structural vulnerability after commodity prices fall.
Informal urban employment is another important adjustment mechanism. Workers leaving agriculture or manufacturing may enter retail, construction, or personal services because those activities have relatively low entry barriers. Employment transfer may reduce the share of agriculture without producing a sustained rise in real output per worker. A comprehensive analysis therefore combines sectoral labor productivity, real wages, informal employment, export composition, and macroeconomic stability. Structural change can occur under stagnation as well as under growth.
Policy evaluation must distinguish a structural outcome from the political label assigned to an episode. Tariff protection, export incentives, fiscal stabilization, and competition reform can each have different effects depending on implementation, world prices, administrative capacity, and labor-market flexibility. General statements about a region should be subordinate to country-level evidence with specified dates, sector coverage, and robust measurement.
22. African Structural Change and Measurement Challenges
Sub-Saharan African structural transformation reveals the importance of country differences, informal production, employment absorption, and changing global conditions. Agriculture remains a major source of jobs in many countries, while urban growth often occurs through construction, trade, transportation, and other services. Manufacturing employs varying shares of workers and faces constraints related to electricity, infrastructure, finance, input costs, and market access. Extractive exports can dominate measured production without generating a comparable employment base.
de Vries, Timmer, and de Vries distinguish static gains from movement into sectors with higher initial productivity and weaker dynamic gains when destination sectors subsequently exhibit modest productivity growth. This distinction is central: reallocation can raise the aggregate level once, but sustained development requires the productivity of destination activities to improve over time. the comparative growth-boom study identifies cases where recent growth relied on either reallocation or within-sector productivity rather than on their simultaneous reinforcement. These findings call for sector-specific historical analyses rather than a universal African account.
The Economic Transformation Database expands comparative coverage by harmonizing twelve sectors for a defined group of countries, including eighteen sub-Saharan African economies in its documented sample. Such coverage improves on overly broad three-sector aggregates, but it does not automatically represent every country, microenterprise, or change after 2018. Household enterprises, subsistence activities, unpaid family work, and service-output deflators remain difficult to measure. A difference between national accounts and labor surveys can materially change estimated productivity levels and the calculated reallocation contribution.
Recent evidence also complicates a one-way interpretation of industrial decline. the manufacturing renaissance investigation documents manufacturing growth in developing countries and examines coverage differences. It should be set alongside the premature-deindustrialization thesis, because the studies emphasize related but distinct outcomes, measures, and historical samples. Manufacturing employment shares can peak early even as absolute manufacturing employment or real output rises in some economies. Neither a declining share nor a rising number of establishments alone establishes the direction of productivity and welfare.
Regional differences in climate, population growth, conflict, land access, energy, and transport make the feasibility of transformation highly specific. Some opportunities arise from agriculture-related processing, digital services, or regional trade; others depend on lowering the costs of electricity and logistics or strengthening basic administrative capacity. The defensible conclusion is a conditional one: development potential follows the capacity to combine job absorption, productivity improvement, and durable links among sectors.
23. Services-Led Development and Least-Developed Economies
Services-led development has become a central policy question because a growing number of economies urbanize and increase the service share before reproducing the historical manufacturing-employment trajectory of early industrializers. A favorable services-led path is possible when tradable, scalable, knowledge-intensive services grow and when basic services become more productive. It is more difficult when most new employment is concentrated in fragmented low-productivity retail, personal transport, or temporary informal work. The relevant division is therefore among service activities and their interactions with the rest of the economy.
The World Bank's analysis of services-led opportunities distinguishes digitally enabled tradables, enabling services used by producers, and face-to-face activities. It documents the reasons some services possess characteristics historically associated with manufacturing, such as scalability and access to international demand. At the same time, many modern services require advanced skills and reliable digital infrastructure. Their ability to generate exports does not necessarily make them capable of absorbing millions of less-skilled workers immediately.
The most recent major institutional synthesis in this review is UNCTAD's Least Developed Countries Report 2025, publicly launched on February 10, 2026. It finds that services account for almost half of economic output in the average least-developed country but often absorb workers in low-productivity, informal activities. The report states that agriculture accounted for 48.6 percent of employment across least-developed countries in 2023 and estimates that these economies need to absorb approximately 13.2 million additional job seekers annually through 2050. These are UNCTAD's dated group estimates, not new calculations in this monograph. They illustrate why employment absorption and services productivity must be examined together.
Digital delivery provides a potentially important margin of growth, but its prerequisites include reliable electricity, connectivity, payment infrastructure, language skills, contractual enforceability, and access to international clients. A logistics hub can raise trade efficiency yet employ relatively few people; an exportable business-service industry may offer good wages but require extensive training; domestic retail can absorb labor but remain constrained by weak household purchasing power. A coherent policy analysis considers their combined effects rather than assigning all services the same development potential.
A useful test of services-led transformation has four parts: sustained real productivity improvement in the relevant service activities; capacity to employ a sizable and inclusive workforce; strengthened linkages with agriculture and manufacturing; and measurable household welfare improvements after adjusting for working conditions, cost of living, and public-service access. This is an evaluative synthesis of the reviewed evidence, not a newly introduced formal economic doctrine. Under that test, neither service employment growth nor service GDP share can alone certify successful transformation.
24. Premature Deindustrialization and Contemporary Manufacturing
Premature deindustrialization describes the tendency of manufacturing employment and, in some cases, nominal manufacturing output shares to reach their peaks at lower per capita income levels than in earlier industrialization experiences. Rodrik's empirical study identifies such shifts in a cross-country historical sample and considers the roles of global technological change and trade. This pattern raises a question for development: how can countries build productive employment and capabilities if the scope for large manufacturing-employment expansions narrows?
Measurement resolves part, but not all, of the controversy. Absolute manufacturing output can grow while its GDP share falls because services or other sectors expand faster. Manufacturing employment can remain stable in absolute numbers while declining as a share of a rapidly growing labor force. Real and nominal value-added shares can move differently because relative prices change. Definitions of manufacturing, establishment size, informal production, and sample countries also matter. Any statement of premature decline must therefore specify the precise indicator and the reference benchmark.
Kruse and colleagues' developing-country evidence emphasizes that manufacturing activity has continued to expand in a range of countries, challenging excessively uniform deindustrialization accounts. the International Yearbook of Industrial Statistics provides a separate institutional basis for examining manufacturing indicators and cross-country coverage. These sources do not imply that modern industry will absorb labor at the same rate as historical textile or assembly sectors; they support the distinction between the expansion of productive capacity and the labor share it represents.
Automation and capital intensity can reduce the amount of direct manufacturing labor required per unit of output. At the same time, manufacturing stimulates employment in logistics, repair, software, engineering, distribution, and energy provision. Some jobs formerly recorded as manufacturing migrate statistically into services through outsourcing. The full industrial employment footprint requires careful input-output and occupational analysis rather than an industry headcount alone.
For policymakers, the problem is one of creating viable learning-intensive productive systems under contemporary technologies and trade conditions. Maintaining obsolete factories without demonstrable capability gains is expensive; abandoning all industry because of a declining historical employment share would disregard production linkages and heterogeneous opportunities. The research supports a strategy of evidence-based experimentation, competition, infrastructure, technical learning, and social protection, with measurable outcomes replacing automatic sector targets.
25. Technology, Digitalization, and Artificial Intelligence
Technological change transforms both the productivity of sectors and the boundary between them. Mechanization can increase agricultural yields while reducing demand for certain types of field labor. Digitally coordinated production can improve factory scheduling and product quality. Cloud computing, electronic payments, communication networks, and automated translation can change which services are commercially tradable. The impact depends on capital costs, electricity reliability, organizational competence, worker skills, and the relative ease of replacing tasks with technology.
A technological improvement is not equivalent to a net reduction in employment. Automation can substitute for certain tasks, lower prices, create demand, and complement workers in other activities. A factory may employ fewer machine operators but more technicians, logistics specialists, designers, and software engineers. A retailer may reduce cash-handling time while expanding delivery, customer support, and digital marketing. These occupations may be located in different sectors or establishments, so an industry-based employment statistic can miss the movement of tasks throughout the production system.
The theoretical relevance is captured by multisector productivity-difference models and capital-deepening models: sector-specific technology and factor intensity change relative costs and labor allocation. Their general-equilibrium lessons apply to contemporary artificial intelligence as a technology used in multiple sectors, but the models do not provide a ready-made causal estimate of current AI-induced employment changes. New data must establish which tasks are deployed at scale, their effects on quality-adjusted output, the labor and capital they displace or complement, and the distribution of benefits.
Digital technologies can reduce entry barriers for exporters of certain services, yet network effects, intellectual-property control, data access, and capital requirements may concentrate market power. Remote service delivery can detach production from metropolitan consumers while increasing dependence on reliable global platforms and telecommunications. Increased measured service exports can reflect genuine domestic skill development, outsourcing from foreign firms, or the relocation of corporate reporting structures. Domestic value added, skills and earnings must be measured to distinguish these mechanisms.
The statistical challenge is substantial. Fast-changing digital products complicate quality adjustment, while services embedded in manufacturing products may cross classification boundaries. The 2025 national accounts update explicitly addresses changes in the digital economy, but the diffusion of new standards through national statistical systems takes time. Scientific accounts of a technological transition should therefore specify measurement dates, adoption indicators, and alternative explanations before attributing a sectoral trend to one technology.
26. Environmental Change and the Energy Transition
The transition toward lower-emissions production is an economywide structural change because energy prices, regulations, technology, and natural constraints alter the feasible activities of farms, manufacturers, transport networks, and services. The production and installation of renewable equipment may increase employment in particular industries; improved energy efficiency may reduce demand for some inputs; resource extraction regions may lose established employers. The overall allocation depends on resource endowments, technology costs, the design of environmental policy, and the timing of capital replacement.
An energy transition is not adequately measured by the employment share of a named 'green sector.' Electricity generation, grid maintenance, building renovation, electrified transportation, industrial heat, agriculture, and software coordination are spread across ordinary industrial classifications. Classifying employment by task, product, process, or sector will generate different estimates. Internationally comparable evidence requires both industry accounts and detailed environmental or energy-use statistics.
Climate vulnerability also influences traditional structural transformation. Droughts and water scarcity can reduce agricultural labor productivity and induce migration without increasing workers' outside opportunities. Extreme heat can reduce hours or physical output in construction and outdoor services. Floods can damage transport networks and raise the cost of maintaining urban concentrations. In such cases labor may shift because the original activity becomes less viable, while destination earnings remain low. A positive or negative sectoral shift-share term cannot by itself capture the welfare losses associated with climate shocks.
Policy requires attention to distribution, finance, and durable capability formation. Training can help workers move from carbon-intensive industries toward electricity, equipment maintenance, or construction, but retraining without credible employer demand will have limited effect. Public investment in grids and transport can crowd in private activity when institutions maintain reliability. Cleaner production can create benefits that national-accounts output alone omits, including health and reduced environmental damages. Evaluating structural transformation in the energy transition therefore requires productivity indicators alongside emissions, energy security, regional effects, and public finance.
27. Inequality, Wages, Poverty, and Welfare
The economic significance of structural transformation ultimately concerns the capabilities, incomes, and living conditions of people. Growth in GDP per worker can coexist with slow median wage growth, unstable employment, regional decline, or worsening environmental conditions. The transmission from measured output to welfare depends on factor ownership, wage bargaining, taxation and transfers, prices of essential goods, and access to public services. An interpretation that concludes at aggregate labor productivity is therefore incomplete.
The distributional effects of reallocation are multidimensional. Workers leaving agriculture may obtain higher hourly wages but face much higher urban rents and commuting costs. Workers displaced from manufacturing may need years to find similarly paid employment because their skills and social networks are location specific. The China-shock labor-market evidence documents how import competition generated geographically persistent adjustment pressures in the United States. At the other end of the process, improved food productivity can benefit nonfarm households through lower prices while changing rural land rents and labor demand.
Gender and unpaid labor require particular attention. Goldin's historical labor-force research and the household-technology account show that women's paid employment and the boundary of market production vary with education, norms, and home technologies. A rise in formal labor-force participation may represent expanded opportunities, but measured productivity is affected by the prior omission of unpaid work. Welfare assessment should include time use, care burdens, safety, autonomy in earnings, and access to services where appropriate evidence is available.
Distributional analysis also needs sectoral heterogeneity. Average service-sector earnings can be raised by a small highly paid finance or software industry even when a large retail workforce earns little. Average manufacturing earnings can disguise differences by firm ownership, export status, capital intensity, and formalization. A policy evaluation should examine worker-level earnings distributions, employment duration, benefits, occupational mobility, and price-adjusted consumption. National productivity growth is one outcome among several, and empirical social improvement requires evidence at a comparable level of aggregation.
The correct normative conclusion is conditional on objectives and observed trade-offs. A policy that raises average output but produces large uncompensated losses may require complementary insurance and adjustment measures. A policy that preserves every existing job indefinitely can constrain entry into more productive activities and impose costs on consumers and taxpayers. Scientific evaluation makes these mechanisms visible and estimates their magnitudes whenever data permit; it does not substitute a single sectoral target for the multidimensional objectives of development.
28. Research Designs and Causal Identification
Structural transformation research combines historical comparison, national accounts, structural economic models, microeconomic surveys, spatial analysis, and policy evaluation. Each design answers a different question. Long-run descriptive studies identify changing shares and correlations with income; shift-share decompositions allocate observed changes in average productivity; multisector equilibrium models specify counterfactual behavior under explicit assumptions; quasi-experimental designs seek causal effects of particular shocks or policies. Their claims should be compared according to the quantity estimated rather than their apparent precision.
Time-series comparisons require consistent sector classifications, deflators, hours, labor-force definitions, geographic borders, and treatment of informal work. Structural breaks in statistical systems can mimic economic turning points. Cross-country comparisons require decisions about income levels, purchasing power, land and capital, commodity dependence, and whether the sample includes conflict periods or small countries. The ETD methods documentation provides an example of the painstaking reconciliation needed to construct sectorally comparable panels. Published data are not free of measurement assumptions merely because they are machine-readable.
A conventional shift-share decomposition is an identity conditional on data construction. Its 'between' contribution compares observed changes in employment weights and sector productivity; it is not an instrumental-variables estimate of the effect of moving a worker. Positive reallocation can coincide with adverse shocks, selection, or induced changes in capital per worker. Likewise, a country-fixed-effects regression of manufacturing share on income identifies within-country association under assumptions, not an automatically exogenous influence of industrial employment on growth.
Natural experiments and instrumental variables may identify more localized effects when their exclusion restrictions and parallel-trends assumptions are plausible. A trade-exposure shock can support inference about manufacturing labor demand, but other shocks or general equilibrium spillovers can affect the result. The local import-competition study and labor-switching-cost analysis illustrate the gains and limits of designs focused on adjustment. Their estimates should not be generalized to every form of technological or sectoral reallocation without attention to exposure and context.
Structural estimation can discipline counterfactual reasoning by imposing production and demand equations and fitting parameters to observed outcomes. manufacturing misallocation counterfactuals and spatial-allocation counterfactuals illustrate the role of model assumptions in estimating gains that are not directly observable. Sensitivity to elasticity choices, unmeasured quality, institutional constraints, and endogenous prices is essential. Successful quantitative replication requires source datasets, exact sector mapping, price conventions, working code, and documented robustness checks.
A defensible research sequence is to define the estimand, state the statistical unit, validate data concordances, calculate transparently, articulate plausible rival mechanisms, seek exogenous variation or specify structural restrictions where causal inference is needed, and report sensitivity and unresolved uncertainty. Scholarly rigor lies in matching the strength of the conclusion to the available design.
29. Comparative Evidence, Rival Explanations, and Limitations
An important question is whether structural transformation causes growth, accompanies growth, or sometimes follows changes in measurement. The answer depends on the period and mechanism. Reallocation toward high average-productivity sectors can raise measured output per worker arithmetically, but technological improvements within sectors can themselves release workers or create demand for labor elsewhere. Income effects redirect consumption, while trade, migration, institutions, and urban infrastructure constrain feasible transitions. An observed sector share is consequently the joint result of several endogenous forces.
The empirical literature displays meaningful contrasts. Duarte and Restuccia's analysis emphasizes the contributions of agriculture, manufacturing, and services to aggregate productivity patterns and convergence. McMillan and coauthors report regionally differentiated reallocation effects in historical samples. the African static-gains study identifies reallocation gains whose dynamism depends on productivity growth in destination sectors. the recent growth-boom comparison shows that within-sector advances and labor reallocation need not reinforce one another in every growth episode. These results concern different datasets and decomposition windows, so variation among them is substantive evidence rather than a contradiction to be removed through averaging.
Theoretical mechanisms likewise require discrimination. nonhomothetic consumption theory predicts demand shifts with income; sector-specific technical progress predicts relative price movements; capital-intensity theory predicts responses to factor accumulation; marketization and specialization theory predicts changes in the division between home and market production. Each can fit a subset of broad sector trends. Stronger identification comes from jointly examining observed quantities, prices, incomes, wages, input shares, household expenditure, and the organization of tasks.
A recurrent limitation is that aggregate productivity studies value output at market or imputed prices. They omit or incompletely measure care quality, domestic work, environmental damages, social protection, and the experienced value of public services. International comparisons can also be distorted by purchasing-power conversion and classification changes. The international national-accounts framework promotes consistency but does not eliminate the need for judgment in rapidly changing or unregistered activities. Therefore, there is no unqualified sector-share threshold at which a country becomes 'structurally transformed.'
Another limitation concerns contemporary evidence. A dataset ending in 2018 cannot establish the effects of the COVID-19 period, later trade disruptions, or 2026 changes in digital technologies without subsequent observations. A report issued in 2026 and titled 2025 must be dated accurately when presented as a current institutional synthesis. Comparisons of classic industrialization and today's services-led opportunities should identify not only similarities in sector labels but differences in technology, demography, global market access, and factor prices.
30. Implications for Economic Policy and Empirical Research
The policy implications of the literature are rooted in the interaction of capabilities and allocation. Policies should raise the productivity of existing activities while improving access to more productive alternatives. Agricultural research, water management, secure land arrangements, and rural roads can increase rural output and create demand for nonfarm services. Electricity, ports, logistics, fair competition, and technical education can reduce the cost of industrial and service production. Worker mobility requires housing, health, transportation, credentials, and insurance. None of these instruments operates independently of political and administrative capacity.
An effective policy process starts with a diagnostic account. What is the productivity distribution across industries and firms? Which workers are changing occupations, and who is excluded? Are employment shifts caused by opportunity, distress, new entry, or changes in data definitions? Do successful establishments face capital, skills, electricity, transport, or market-access constraints? Can public intervention address a demonstrable externality or coordination failure at acceptable fiscal cost? the comparative growth case studies and the middle-income development framework support an approach that links policy to diagnosed constraints, technology absorption, and institutional capacity.
Industrial policy is one possible instrument, with conditions for credible success: clear objectives, transparent eligibility, time-bound or performance-conditioned support, openness to new entrants, monitoring of domestic capabilities and wages, and an exit mechanism for persistently ineffective programs. A services strategy requires equally concrete criteria for exportability, digital infrastructure, employment absorption, and connections to agriculture and manufacturing. UNCTAD's least-developed-country assessment warns that services expansion with weak productivity and limited digital capabilities is insufficient to deliver sustained development.
Workforce policies should recognize the asymmetry between short-run losses and long-run national gains. Sectoral and geographic shocks can destroy earnings trajectories even when consumers benefit from lower prices. documented trade-adjustment costs and persistent regional responses justify evaluating adjustment assistance, retraining, portable benefits, childcare, and housing alongside investment and trade measures. Their design requires evidence on employment take-up, training completion, wage recovery, and fiscal sustainability rather than program enrollment alone.
The research agenda should move from broad shares toward linked measures of task content, establishment dynamics, wage distributions, domestic value added, environment, and household well-being. Better measurement of informal firms and own-use production is especially valuable. The future empirical value of SNA 2025 implementation and newer industry classifications depends on transparent concordances that preserve historical continuity. An advanced statistical program would make both the production and labor sides of a sectoral comparison reproducible and identify revisions clearly.
31. Original Analytical Synthesis
The reviewed literature supports a comprehensive interpretation of structural transformation as the adjustment of an economy's production system to changing capabilities, relative values, and social requirements. Sectoral composition is the visible record of that adjustment. Its economic significance follows from the productive possibilities created and the opportunities distributed among workers and households. This synthesis integrates the established theories and empirical results without introducing a new named theory, empirical law, or doctrine.
First, the accounting structure imposes a necessary discipline. Aggregate labor productivity depends on sectoral productivity levels and labor shares, so within-sector improvements and between-sector shifts are jointly measurable when the data permit an internally consistent decomposition. The decomposition is exact as arithmetic but conditional as economics: it describes observed change and leaves open the determinants of migration, sectoral productivity, factor prices, and welfare. The first analytical responsibility is therefore to identify what the statistic actually measures.
Second, historical and theoretical diversity is an essential characteristic of the subject. Agricultural productivity can lower food costs and release labor; manufacturing can enable learning and scale; high-productivity services can reduce transaction costs, expand export opportunities, and improve other sectors' performance. The relevance of each mechanism depends on technology, skill requirements, market access, factor endowments, and institutions. In the literature, Lewis's labor-transfer mechanism, demand-driven structural change, and technology-and-relative-price adjustment describe complementary forces that may operate in different combinations.
Third, productive absorption and technological upgrading must be examined together. An economy may move a large number of workers into activities with higher average value added and enjoy a temporary accounting gain, while those activities fail to raise productivity thereafter. Another economy may experience rapid technical improvement within a relatively small high-productivity sector without absorbing workers on a comparable scale. Sustained, inclusive development is more plausible when improvements in the productivity of existing work accompany viable opportunities for new entrants and displaced workers. The contrasting evidence in African structural-change research and developing-country growth comparisons supports this distinction.
Fourth, sector labels are increasingly imperfect proxies for productive characteristics. Manufacturing firms purchase software, logistics, design, and maintenance; services firms use capital equipment and industrial infrastructures; digital activities can be traded at distance; agriculture depends on specialized industrial inputs. Statistical classification remains indispensable, but an analysis of productivity mechanisms must follow tasks, supply chains, and capability diffusion across the sector boundary. The services-led development assessment and latest UNCTAD synthesis show both the promise and the constraints of this interdependence.
Fifth, transformation has a geographic and distributive structure. Labor movement, housing constraints, worker selection, social norms, and adjustment costs determine who can benefit from newly productive activities. spatial misallocation research, agricultural worker-selection analysis, and gendered labor-history evidence illuminate different parts of this constraint structure. Development cannot be inferred solely from national sector averages when regions, cohorts, and genders face sharply different possibilities.
Finally, a responsible assessment combines five observable questions in one research judgment: How does real productive capacity change? How do workers and firms reallocate? How persistent are productivity improvements within destination and origin activities? Which institutions make learning and access possible? How do wages, poverty, public services, and environmental outcomes change? These questions establish an analytical reading of evidence already developed in economics. Their combination supports a substantive conclusion: durable structural transformation is best understood through the joint evolution of productivity, productive employment, capabilities, and human well-being, with sector shares serving as necessary but incomplete indicators.
Conclusion
Structural transformation is a central process of long-run economic development because changes in the allocation and organization of production link technological progress, consumption, trade, geography, demography, and institutions. Its historical manifestations range from agricultural modernization and industrial expansion to the growth of specialized services, digital production networks, and environmentally constrained economic systems. The discipline has developed from the early dual-economy and historical-comparative traditions into a body of multisector equilibrium theory, sectoral growth accounting, and internationally comparative micro- and macroeconomic research.
Three conclusions follow from the evidence. Sectoral shares require exact definitions of employment, output, prices, industries, and time periods. A measured reallocation contribution to productivity is an accounting result whose causal significance must be established through further evidence. The development potential of any sector depends on its productivity trajectory, capacity to employ and upgrade workers, links to other activities, and institutional context. The historical importance of manufacturing remains substantial, while contemporary services and technological change create differentiated opportunities and constraints that demand independent assessment.
The science of structural transformation is therefore strongest when it studies specific productive mechanisms, tracks their effects across industries and regions, and measures gains and adjustment costs for the people involved. This approach explains why economies may grow with changing sectoral shares, why similar shares can represent different productive structures, and why durable prosperity requires both improved productive capacity and feasible access to its benefits. The research record supports a plural, measurement-conscious, historically grounded account of economic transformation and an open empirical agenda for the Artificial Era.
References
Acemoglu, D., & Guerrieri, V. (2008). Capital deepening and nonbalanced economic growth. Journal of Political Economy, 116(3), 467–498.
Artuç, E., Chaudhuri, S., & McLaren, J. (2010). Trade shocks and labor adjustment: A structural empirical approach. American Economic Review, 100(3), 1008–1045.
Autor, D. H., Dorn, D., & Hanson, G. H. (2013). The China syndrome: Local labor market effects of import competition in the United States. American Economic Review, 103(6), 2121–2168.
Baumol, W. J. (1967). Macroeconomics of unbalanced growth: The anatomy of urban crisis. American Economic Review, 57(3), 415–426.
Boppart, T. (2014). Structural change and the Kaldor facts in a growth model with relative price effects and non-Gorman preferences. Econometrica, 82(6), 2167–2196.
Buera, F. J., & Kaboski, J. P. (2012). Scale and the origins of structural change. Journal of Economic Theory, 147(2), 684–712.
Chenery, H., & Syrquin, M. (1975). Patterns of development, 1950–1970. Oxford University Press for the World Bank.
de Vries, G., Arfelt, L., Drees, D., Godemann, M., Hamilton, C., Jessen-Thiesen, B., Kaya, A. I., Kruse, H., Mensah, E., & Woltjer, P. (2021). The Economic Transformation Database (ETD): Content, sources, and methods. WIDER Technical Note 2/2021. UNU-WIDER.
de Vries, G., Timmer, M., & de Vries, K. (2015). Structural transformation in Africa: Static gains, dynamic losses. Journal of Development Studies, 51(6), 674–688.
Diao, X., McMillan, M., & Rodrik, D. (2017). The recent growth boom in developing economies: A structural change perspective. NBER Working Paper No. 23132.
Duarte, M., & Restuccia, D. (2010). The role of the structural transformation in aggregate productivity. Quarterly Journal of Economics, 125(1), 129–173.
Goldin, C. (1994). The U-shaped female labor force function in economic development and economic history. NBER Working Paper No. 4707.
Gollin, D., Lagakos, D., & Waugh, M. E. (2014a). The agricultural productivity gap. Quarterly Journal of Economics, 129(2), 939–993.
Gollin, D., Lagakos, D., & Waugh, M. E. (2014b). Agricultural productivity differences across countries. American Economic Review, 104(5), 165–170.
Gollin, D., Parente, S., & Rogerson, R. (2002). The role of agriculture in development. American Economic Review, 92(2), 160–164.
Greenwood, J., Seshadri, A., & Yorukoglu, M. (2005). Engines of liberation. Review of Economic Studies, 72(1), 109–133.
Harris, J. R., & Todaro, M. P. (1970). Migration, unemployment and development: A two-sector analysis. American Economic Review, 60(1), 126–142.
Herrendorf, B., Rogerson, R., & Valentinyi, Á. (2014). Growth and structural transformation. In P. Aghion & S. N. Durlauf (Eds.), Handbook of Economic Growth (Vol. 2B, pp. 855–941). Elsevier.
Hsieh, C.-T., & Klenow, P. J. (2009). Misallocation and manufacturing TFP in China and India. Quarterly Journal of Economics, 124(4), 1403–1448.
Hsieh, C.-T., & Moretti, E. (2019). Housing constraints and spatial misallocation. American Economic Journal: Macroeconomics, 11(2), 1–39.
International Labour Organization. (n.d.). ILOSTAT data portal. ILO. Accessed November 16, 2026.
Kaldor, N. (1966). Causes of the slow rate of economic growth of the United Kingdom. Cambridge University Press.
Kongsamut, P., Rebelo, S., & Xie, D. (2001). Beyond balanced growth. Review of Economic Studies, 68(4), 869–882.
Kruse, H., Mensah, E., Sen, K., & de Vries, G. (2023). A manufacturing (re)naissance? Industrialization in the developing world. IMF Economic Review, 71, 439–473.
Kuznets, S. (1971). Modern economic growth: Findings and reflections. Nobel Memorial Prize lecture. NobelPrize.org.
Lagakos, D., & Waugh, M. E. (2013). Selection, agriculture, and cross-country productivity differences. American Economic Review, 103(2), 948–980.
Lewis, W. A. (1954). Economic development with unlimited supplies of labour. The Manchester School, 22(2), 139–191.
McMillan, M., Rodrik, D., & Sepúlveda, C. (Eds.). (2017). Structural change, fundamentals, and growth: A framework and case studies. International Food Policy Research Institute.
McMillan, M., Rodrik, D., & Verduzco-Gallo, Í. (2014). Globalization, structural change, and productivity growth, with an update on Africa. World Development, 63, 11–32.
Nayyar, G., Hallward-Driemeier, M., & Davies, E. (2021). At your service? The promise of services-led development. World Bank.
Ngai, L. R., & Pissarides, C. A. (2007). Structural change in a multisector model of growth. American Economic Review, 97(1), 429–443.
Rodrik, D. (2013). Unconditional convergence in manufacturing. Quarterly Journal of Economics, 128(1), 165–204.
Rodrik, D. (2016). Premature deindustrialization. Journal of Economic Growth, 21, 1–33.
Syrquin, M. (1988). Patterns of structural change. In H. Chenery & T. N. Srinivasan (Eds.), Handbook of Development Economics (Vol. 1, pp. 203–273). Elsevier.
Timmer, M. P., & de Vries, G. J. (2009). Structural change and growth accelerations in Asia and Latin America: A new sectoral data set. Cliometrica, 3, 165–190.
United Nations Conference on Trade and Development. (2025; launched February 10, 2026). The least developed countries report 2025: Are services the new path to structural transformation?. United Nations.
United Nations Industrial Development Organization. (2025). International Yearbook of Industrial Statistics 2025. UNIDO.
United Nations Statistics Division. (2024). International Standard Industrial Classification of All Economic Activities, Revision 5. Classification adopted 2024.
United Nations Statistics Division. (2025). System of National Accounts 2025. United Nations; adopted by the UN Statistical Commission in 2025.
United Nations University World Institute for Development Economics Research & Groningen Growth and Development Centre. (2023). Economic Transformation Database (ETD). Database documentation (updated September 2023).
World Bank. (2023). World Development Report 2023: Migrants, refugees, and societies. World Bank.
World Bank. (2024). World Development Report 2024: The middle-income trap. World Bank.