SCIENCE
Economic Geography
Abstract. Economic geography investigates how economic life is arranged across space and how places, connections, institutions, and environmental conditions jointly shape production, trade, employment, consumption, and welfare. Spatial allocation arises from decisions taken by firms, households, governments, and networks under transport and communication costs, increasing returns, localized knowledge, land scarcity, and historically accumulated capabilities. This monograph integrates classical location theory, regional science, economic geography's institutional and relational traditions, the new economic geography, quantitative spatial equilibrium, empirical agglomeration studies, and global production-network research. It specifies the distinction between spatial description, causal explanation, and welfare counterfactuals; examines the sensitivity of conclusions to geographic units, sorting, selection, and cross-boundary spillovers; and evaluates contemporary urbanization, digital concentration, climate exposure, industrial transition, and territorial policy. Its original analytical synthesis organizes existing evidence around an explicit chain from spatial structure and connectivity through economic mechanisms and institutional mediation to distributional outcomes. The synthesis is an interpretive integration of established research, not a newly proposed formal theory. It emphasizes that a place's observed prosperity, local treatment effect, and society-wide benefit answer different questions. The resulting discipline is comparative, multiscalar, historically aware, and empirically testable.
Keywords: Economic geography; spatial economics; location theory; agglomeration; market access; regional development; central place theory; core-periphery; spatial equilibrium; global production networks; migration; infrastructure; spatial econometrics; economic inequality; geospatial methods; urbanization; climate adaptation
1. Economic Geography as a Scientific Field
Economic geography is the science of where economic activities occur, why particular spatial patterns emerge, how they change, and what those patterns imply for productivity, opportunity, power, and well-being. Its observational objects include economic establishments, farms, mines, warehouses, transport links, households, labor-market areas, political borders, cities, regions, global supply chains, and the flows connecting them. The field studies both distributions—where output, labor, investment, income, and knowledge are located—and processes that reproduce or transform those distributions. A map of establishments is an empirical starting point; an explanation requires explicit mechanisms connecting decisions to the mapped configuration.
Geographic space has physical and socially constituted properties. Terrain, climate, coastline, soil, mineral deposits, and travel time constrain feasible locations. Property rights, zoning, migration rules, infrastructure finance, social networks, and political borders shape the effective costs of access. A kilometer can be inexpensive across an integrated metropolitan transit system and economically prohibitive across a militarized border. This is why geographical distance and economically relevant distance must be measured separately. The World Bank's 2009 spatial development assessment usefully emphasizes density, distance, and division as distinct dimensions of economic organization; its policy narrative is a synthesis of development research and should not be mistaken for a universal equation.
The central questions fall into five connected classes. Location asks which site a farm, household, factory, or firm chooses among alternatives. Interaction examines movement of goods, workers, capital, ideas, and information. Organization investigates how firms and institutions coordinate activity inside districts and across continents. Transformation studies how economic landscapes acquire path-dependent specialization and respond to shocks. Evaluation asks who gains, who loses, and whether spatial changes improve welfare after adjusting for land rents, moving costs, environmental losses, and public expenditure.
Economic geography intersects urban economics, regional economics, international economics, development, economic history, labor economics, transportation, environmental economics, and human geography. These connections do not dissolve its distinct object: the joint spatial arrangement of activities and relationships across multiple scales. The worldwide production restructuring described in Dicken's Global Shift illustrates why an analysis restricted to one city or one country's trade accounts misses consequential cross-territorial connections.
Explanation requires more than correlation. High wages in a city might reflect local productivity, costly housing, differences in worker composition, occupational selection, or compensating amenities. A strong geographic correlation between two industries can reflect shared inputs, common labor requirements, technological spillovers, or a third underlying locational advantage. The discipline therefore combines spatial theory with comparative history, geographic description, and research designs that distinguish alternative mechanisms.
2. The Historical Formation of Spatial Economic Thought
The modern discipline developed through distinct intellectual traditions that asked different but mutually illuminating questions. Nineteenth-century classical location analysis treated transport costs, land rents, and production conditions as reasons for differences in land use and industrial siting. Johann Heinrich von Thünen's 1826 agricultural rings examined the bid rent of products surrounding an isolated market under deliberately stringent uniformity assumptions. That idealization established a durable question: how does accessibility alter the highest-valued use of land? The emphasis on an abstract market center is a modeling device, not a historical claim that real landscapes were circular or institutionally homogeneous.
In the early twentieth century, Alfred Weber's industrial location analysis identified transport, labor, and agglomeration forces in the spatial optimization of factories; the historical argument is accessible in the English translation of Theory of the Location of Industries. Walter Christaller's 1933 study of central places in southern Germany considered threshold demand and service range in a settlement hierarchy. August Lösch developed systems of market areas and economic landscapes; the historical legacy of both scholars includes the distinction between ideal geometric arrangements and actual settlement networks conditioned by institutions, transport, and uneven populations.
After World War II, formal regional science drew economics, geography, transportation engineering, and quantitative methods into a common methodological arena. Walter Isard's Location and Space-Economy connected market areas, land use, urban structures, and transportation in a unified research program. Input-output linkages, gravity models, regional multipliers, and interregional accounts made spatial questions more accessible to formal and applied economic analysis. Quantitative geography and spatial science contributed location-allocation models, point-pattern analysis, cartography, and increasingly systematic data processing.
By the late twentieth century, geographical scholarship also challenged the assumptions of frictionless mobility and geographically neutral institutions. Political-economic approaches analyzed uneven capitalist development, labor segmentation, and the distribution of control over production. Institutional and evolutionary economic geography examined regional trajectories, routines, specialized knowledge, and historically inherited industrial capabilities. Relational approaches emphasized the networks through which a location acquires economic significance. This material enriches any reading of global production networks: regional outcomes depend on connections and institutional positions as well as physical coordinates.
A further analytical phase emerged with increasing-returns models of market size, firms, transport costs, and mobile factors. Krugman's 1991 core-periphery model demonstrated that cumulative spatial concentration could arise from decentralized market forces. The formal integration in The Spatial Economy showed how cities, regions, and trade could be understood through related mechanisms. Subsequent quantitative spatial economics made it possible to analyze many locations, multiple sectors, heterogeneous trade frictions, and welfare consequences using observed geographic data. Meanwhile, qualitative interviews, network tracing, GIS, and remote sensing expanded the field's evidence base. These traditions are complementary in scientific practice because they illuminate different layers of the same spatial system.
3. Fundamental Spatial Concepts and Geographic Scale
A location is a site identified by coordinates, boundaries, and attributes, but economic location also includes access to other actors and resources. A region may be administrative, ecological, commuting-defined, functional, or constructed analytically. An industrial cluster is a spatial concentration of related activity; an economic network is a set of relations among actors; a corridor is a geographic concentration of connectivity. The same activity may belong to several regions and networks at once. Precision therefore begins with the unit of observation and the unit to which conclusions are intended to apply.
Distance has multiple representations. Euclidean distance is measured along the straight line connecting two points; network distance follows roads, railways, shipping routes, or digital networks; generalized transport cost combines money, time, reliability, waiting, customs barriers, and risk. Market access summarizes opportunities reachable after accounting for these frictions. Density measures activity per spatial area, but its economic meaning changes with the denominator, population served, mix of activities, and spatial resolution.
A local labor market is better approximated by commuting and job-search patterns than by a political boundary drawn for historical reasons. A functional urban area may include employment-intensive suburbs, settlement fringes, and interconnected municipalities. Eurostat's degree-of-urbanisation methodology provides a harmonized grid-based scheme that permits systematic city, town, and rural comparisons, while the United Nations' 2025 urbanization assessment highlights why national definitions and harmonized classifications require careful distinction. Any report on 'urban population' should identify its classification and vintage.
The scale problem is fundamental. National totals may conceal divergent industrial corridors, and comparisons across very small census units can generate unstable rates. Spatial aggregation changes variance, covariance, and perceived cluster structure, an issue developed in Arbia's regional-data analysis. The modifiable areal unit problem includes scale effects (large versus small zones) and zoning effects (different partitions at the same scale). The ecological fallacy occurs when area-level association is mistaken for the behavior of individuals; atomistic fallacy can occur when individual mechanisms are presumed to describe regional systems without examining interactions.
Spatial heterogeneity refers to differing relationships or conditions across places; spatial dependence arises when nearby or connected observations jointly influence outcomes. Neither is a statistical nuisance alone. Dependence may indicate diffusion, supply chains, social interaction, shared shocks, or omitted variables. Researchers must specify whether 'neighbor' denotes a common boundary, transport-time proximity, trade flow, institutional connection, or another substantive relation. The geography built into a model is an empirical assumption subject to testing, not a neutral coordinate choice.
4. Classical Theories of Agricultural and Industrial Location
Classical location theory begins from explicit production and exchange trade-offs. Consider a farm producing crop c at distance d from a town. The maximum land rent the farm can pay per unit area can be written schematically as R_c(d) = y_c[p_c − k_c − t_c(d)], where y_c is yield per area, p_c a market price, k_c nontransport unit cost, and t_c(d) unit shipping cost. With a linear transport schedule t_c(d) = τ_c d, the rent gradient is R_c(d) = y_c(p_c − k_c) − y_cτ_c d. The activity with the highest feasible bid rent occupies a site in the simplest competitive version. High-value perishables may be located close to the market even where another crop has higher yield. This reconstructs the central analytical intuition associated with von Thünen and demonstrates how observed land-use rings would depend on yields, technology, market prices, topography, and more than one destination.
A factory's location problem may be written min_x Σ_i q_i c_i(x) + w(x)L + F(x) − A(x), with input flows q_i, delivery cost c_i(x), local wages w, labor requirement L, fixed land and regulatory cost F, and a site-specific agglomeration benefit A. Under fixed production quantities and known routes, this is a locational cost-minimization problem. The central historical Weberian transport-and-labor framework uses comparable reasoning to explain why a site may lie near raw materials, final markets, or cheaper labor. The result is contingent on factor intensities, supply chains, processing losses, and institutional feasibility.
These classical models have lasting explanatory value as benchmark counterfactuals. A logistics warehouse near a port reflects generalized freight cost; refrigerated food processing may prioritize reliability over straight-line distance; skilled design services may locate near labor and clients although their final product is delivered digitally. Each case can be analyzed by revising the cost vector rather than treating a nineteenth-century formula as a complete description of present economies.
The models are also limited in identifiable ways. They often represent a single decision maker, a static price structure, and predetermined demand. In actuality, prices, rents, infrastructure, suppliers, and labor pools may change with the location choice itself. A plant can attract upstream suppliers, change local wages, create congestion, and alter the cost function faced by later entrants. Economic geography extends classical analysis by allowing those feedbacks to become endogenous.
5. Central Places, Land Markets, and Regional Science
Central-place theory relates a hierarchy of settlements to the range over which consumers travel to obtain services and the minimum market threshold required to operate them. Everyday products usually sustain numerous small service centers, while infrequently purchased specialized products require larger catchments. Christaller's central-place exposition provided a geometric idealization under regular settlement and transport assumptions. Real service geographies are more irregular because purchasing power, commuting networks, historical institutions, online access, and administrative boundaries alter demand and accessibility.
Regional science formalized these intuitions through land-use equilibria, input-output accounts, spatial interaction models, regional accounts, and location-allocation optimization. Isard's synthesis of space economy anchors this movement. A regional input-output matrix represents flows between industries: with vector of final demand f and coefficient matrix A, gross production is x = (I − A)^(−1) f when the inverse exists and coefficients are fixed. Spatially explicit extensions permit cross-region sourcing, but the multiplier is an accounting propagation under stipulated production relationships. It is not by itself evidence that a new plant causally creates the implied jobs.
Location-allocation models ask where to establish schools, hospitals, distribution hubs, warehouses, or depots subject to travel or service constraints. A p-median problem chooses p facilities to minimize the weighted distance from demand nodes to their assigned facility. A covering model instead chooses sites to serve demand within a specified response-time threshold. Both require weights, travel metrics, realistic network constraints, and an explicit welfare objective. A facility solution optimized for average journey time may reduce access for densely populated centers while worsening it for remote or mobility-constrained groups.
The scientific importance of regional science is methodological: its models force the analyst to state how spatial units, link costs, technology, and budget constraints enter an explanation. The empirical importance lies in translating place-based policy into testable spatial choices. A mathematically optimal map always remains conditional on the costs and population distribution put into the model.
6. Agglomeration: Sharing, Matching, and Learning
Agglomeration economies arise when the proximity of economic actors raises productivity, expands opportunity, or lowers costs beyond what can be explained by the independent characteristics of each participant. The sharing, matching, and learning framework distinguishes three mechanisms. Sharing allows firms to use specialized inputs, large fixed-cost infrastructure, and indivisible facilities. Matching improves the pairing of workers, firms, suppliers, and customers when a thicker market reduces search frictions. Learning facilitates the exchange and recombination of ideas, especially where communication is tacit, repeated, and socially embedded.
Localization economies are associated with concentration of an industry or related industries. Urbanization economies arise from broader diversity and city-wide scale. A stylized plant production equation can represent them as log Y_f = log A_f + α log K_f + β log L_f + γ log D_r + ε_f, with D_r a measure of relevant local density. The estimated γ is a productivity-density correlation unless research design addresses worker sorting, selection of firms, reverse causality, local capital quality, and spatially correlated shocks. A coefficient can be sensitive to spatial radius because learning, commuting, and business-service interactions decline over different distances.
The survey of empirical agglomeration evidence emphasizes variation in mechanisms and spatial extent. Coagglomeration evidence links industries that appear together to input-output relations, labor pooling, and technology flows. French firm-level analysis differentiates city productivity advantages from sorting and market selection; location effects and composition effects must be interpreted separately.
Concentration imposes costs as well as benefits. Rents, land prices, congestion, pollution, waiting time, and competition for labor rise as activity becomes dense. For an additional resident, a private wage premium can be offset by higher housing costs and lower amenity values. Some benefits are internalized through market prices, whereas knowledge spillovers, traffic congestion, and pollution may remain external. To estimate an optimal density, a scholar must consider marginal benefits and costs, land and transport markets, and the distribution of effects, rather than treat higher population density as unconditionally desirable.
There are several reasons not to infer a welfare externality from concentration alone. Productive firms can select cities; skilled workers can sort into attractive places; successful cities may receive infrastructure earlier; regulations may restrict housing and drive up nominal wages; and industrial diversity can reflect a city's historic trajectory. A good agglomeration study seeks variation in access, relocation, historic disruptions, or policy that identifies a mechanism. A correlation between wages and density provides a research question, not its final answer.
7. Increasing Returns and the New Economic Geography
New economic geography combines increasing returns, monopolistic competition, mobile labor or firms, and geographically costly trade to explain self-reinforcing spatial concentration. Consider two regions with manufacturing varieties produced under fixed plus marginal costs and consumers valuing diversity. When trade involves iceberg costs—more than one unit must be shipped for one unit to arrive—a larger local market makes nearby production attractive. An increasing number of firms raises product variety and potentially real income, attracting workers and demand. Firms and people may therefore move in response to each other's decisions.
In the 1991 core-periphery model, these demand linkages provide a centripetal force. Dispersion is supported by the geographic distribution of immobile demand, competition for local customers, and, in more general models, land and congestion costs. The equilibrium may feature a relatively even distribution or a core-periphery arrangement depending on expenditure shares, transport frictions, elasticity of substitution, and the mobility of factors. The model does not establish that every economy must become a single core or that concentration always raises national welfare.
Analytical refinements, including Agglomeration and Trade Revisited, make clearer how forward-looking migration and specific utility structures affect the possible equilibria. The synthesis of urban and international spatial economies develops systems of cities and regions with interacting production and mobility forces. Path dependence follows because the location of existing firms and workers changes the incentives of new entrants. A historically modest market advantage can become persistent when feedbacks are sufficiently strong.
The researcher must distinguish multiple equilibria, historical accidents, and policy-induced transformations. If two configurations are viable, knowing the primitives of preferences, technology, and geography may not predict which one a region will occupy without information about history and expectations. A new transport corridor can increase integration and simultaneously strengthen concentration in the already dominant center. Such an outcome is coherent because lower trade costs change market access asymmetrically and alter competition faced by smaller producers.
The contribution of new economic geography is its transparent treatment of feedback between markets and space. Its limits follow from tractability assumptions: a small number of regions or industries, simplified preferences, stylized mobility, and stylized institutions. The best use of the framework is to derive explicit hypotheses about market access, concentration, mobility, and shocks, then confront them with detailed regional evidence.
8. Trade Costs, Accessibility, and Market Potential
Market access summarizes how strongly a location is connected to purchasing power, suppliers, jobs, and services. A simple descriptive index is MA_i = Σ_j Y_j exp(−λd_ij), where Y_j denotes destination market size and d_ij a chosen measure of distance. This index deliberately omits price-index and equilibrium feedback effects. An equilibrium-based market-access term typically aggregates expenditure weighted by bilateral trade frictions and price-related terms; its exact exponent depends on preferences and trade structure.
Hanson's United States county study examines whether wage patterns correlate with demand linkages derived from new economic geography. It documents spatially limited market-potential relationships; this association is informative about the geographical reach of demand forces, but it cannot be translated into a universal labor-market wage elasticity. Access also varies according to the mode of travel: maritime shipping, air freight, truck routes, digital communications, and labor commuting have different cost functions and reliability profiles.
A generalized accessibility problem can use cost C_ij = monetary fare + value_of_time × travel_time + expected_loss + border/compliance cost. These components vary by commodity, person, and firm. A road can reduce truck costs while leaving low-income residents without reliable public transit. Infrastructure provision therefore requires separate freight, labor-market, and social-service access measures. A single 'connectedness' score is a modeling convenience and must not hide conflicting outcomes.
The World Bank's spatial-development analysis emphasizes reducing distance and division in development processes. This is best understood as an argument about connected opportunity subject to local institutional conditions. Eliminating a physical barrier may connect small producers to a larger market, but competition from established producers can reallocate activity away from the intermediate locality. The comparative prediction must jointly account for import competition, scale economies, alternative employment, factor mobility, and public-service availability.
Accessibility itself is endogenous. New economic activities induce infrastructure investment, and infrastructure investment alters where activities locate. To identify whether a corridor caused industrialization, one must examine route-selection processes, omitted pre-trends, geographic feasibility, and spillovers across affected locations. Evaluations of a transport corridor that compare 'treated' counties with nearby counties can understate the aggregate effect if control counties also gain access or lose activity through relocation.
9. Quantitative Spatial Equilibrium and Counterfactuals
Quantitative spatial equilibrium connects observed wages, prices, rents, populations, trade flows, and movement costs within a model where households and firms respond to one another. Each location has amenities, productivity, land supply, and access to markets. Households select among locations in light of employment income, prices, rents, migration frictions, and amenities; firms choose production sites and input sources in light of productivity and trade costs. Market clearing requires that factor payments, consumption, production, trade, and population movements fit together.
In Trade and the Topography of the Spatial Economy, geographical heterogeneity enters general-equilibrium trade through trade costs and locational fundamentals. Quantitative Spatial Economics surveys the formal structures used to infer changes across many places, including welfare effects of policy and transport improvements. The parameters and closure assumptions matter: labor can be perfectly mobile, partially mobile, or tied to place; housing supply may be flexible or constrained; tax revenues may be redistributed nationally or regionally.
A useful analytical distinction is between partial and general equilibrium. A partial-equilibrium calculation of freight savings for a corridor holds wages, rents, and population constant. A general-equilibrium calculation allows those quantities, competitors' decisions, and trade flows to change. In a multi-region model, summing local project impacts without accounting for displacement, market interaction, and price changes generally does not yield the national welfare effect.
One representation of residential welfare is V_ir = (w_ir − T_ir)/(P_r H_r) + a_r, where w is money earnings, T personal travel and tax costs, P_r a composite consumption price, H_r an appropriately normalized housing-cost term, and a_r an amenity index. The exact formula is illustrative rather than an identified cardinal welfare index: researchers usually use indirect utility with multiplicative prices and an econometrically specified amenity term. The key point is that nominal earnings cannot alone rank locational welfare. Housing, access, risk, and preferences all enter.
Model evaluation proceeds in layers. First establish that estimated parameters and baseline data plausibly reproduce observed migration, trade, employment, and price distributions. Next examine whether estimated relationships correspond to identified mechanisms rather than absorbing unrelated shocks. Then compare counterfactuals under alternative assumptions for movement, land supply, externalities, and fiscal redistribution. Robustness must examine whether an apparent national welfare gain masks losses for immobile workers or households excluded from high-productivity places.
10. Urban Systems, Cities, and Housing
Cities concentrate households, employment, infrastructure, specialized services, and institutions, making them important arenas for agglomeration and conflict over scarce land. Urban size depends on productivity, amenities, commuting costs, housing supply, and the wider hierarchy of cities. Large cities may host complex services and high-value design work while smaller cities specialize in manufacturing, logistics, administration, agriculture services, or research. These roles change with technology, trade, and public investment.
Residential and commercial land markets allocate sites through rents, but land supply is regulated by topography, zoning, planning, construction costs, and property rights. A shock that raises metropolitan productivity may result in higher employment and in-migration where housing expands. Where supply is constrained, much of the benefit can be capitalized into rents or house prices. Spatial models therefore distinguish wage gains from real purchasing power. The Berlin Wall density study uses historical division and reunification to investigate the joint roles of productivity, amenities, and urban structure.
The metropolitan boundary is partly a functional labor-market boundary. Workers live in a suburban or rural municipality and commute to a central employment location; local income tax records or resident population can assign economic activity differently from workplace measures. The harmonization issues in the degree-of-urbanisation manual are crucial when comparing metropolitan changes internationally. Counting an administrative city's population without its commuter belt can misstate the size of the economically integrated urban system.
The 2026 OECD regional and urban report finds housing in urban centers averaged 27% more expensive than national averages across 23 OECD countries with available data; the contrast represents specified OECD coverage and period, not a global or causal estimate. Its analysis highlights that housing scarcity can limit the ability of workers and firms to take advantage of high-productivity places. Accessibility is therefore partly a housing and land-use question. Policies addressing transit, infill, housing supply and local public amenities can change the realized distribution of agglomeration gains.
Cities also generate multiple forms of environmental exposure. Urban heat, flood risk, infrastructure interdependence, and air pollution impose unequal costs by neighborhood and income. IPCC's assessment of urban climate risk identifies interactions between built environments and adaptation capacity. Economic geography must analyze the city as an ecosystem of economic flows and physical vulnerability, not merely as a concentration of jobs.
11. Regional Specialization, Diversification, and Structural Change
A region's production structure can be measured through employment, value added, exports, firm counts, patents, occupational mix, and the diversity of its supplier base. A location quotient for industry s in region r is LQ_rs = (E_rs/E_r)/(E_s/E), where E_rs denotes regional employment in the industry, E_r total regional employment, E_s national industry employment, and E national employment. A quotient above one means the industry has a larger employment share locally than nationally. It does not establish local efficiency, export competitiveness, knowledge spillovers, or positive future growth.
Specialization can support scale, targeted training, suppliers, and institutions. Diversification can spread risks and permit recombination across industrial activities. The economic question is conditional: which specialization supports productivity given local labor, knowledge, infrastructure, global demand, and exposure to shocks? Growth in Cities examines competing arguments about industrial specialization, diversity, and competition in city growth, while the coagglomeration study shows that co-location often reflects actual economic connections among industries. These are separate empirical patterns and their methods of measurement should be kept separate.
Structural transformation occurs when labor, capital, technology, and output shift among agriculture, manufacturing, and services. It has geographic consequences even where national sectoral shares appear smooth. A manufacturing decline may be localized in an industrial belt while knowledge-intensive services concentrate in a few metropolitan areas. A rural region can become more productive through mechanization yet lose population as agricultural labor demand falls. A national sectoral productivity improvement can coexist with declining employment in places tied to that sector.
Regional resilience depends on exposure, adaptive capacity, and the structure of employment relationships. A diverse region may absorb a sectoral shock by transferring labor to related industries; another may struggle because skills, capital assets, and housing are ill-suited to expanding sectors. Regional specialization and regional resilience must be measured over time: a static cluster map cannot show whether enterprises are creating new capabilities, outsourcing core functions, or progressively losing bargaining power.
The OECD's 2026 assessment reports that knowledge-intensive business services were the largest sectoral contributor to labor productivity growth in 17 of 34 OECD member and accession countries with available data in 2024. This is a time- and coverage-specific observation, consistent with the growing importance of regionally concentrated services. It does not license extrapolation to every country or imply that manufacturing and agriculture have ceased to shape regional development.
12. Industrial Districts, Clusters, and Innovation
An industrial district is a localized constellation of firms, suppliers, workers, and social institutions; a cluster adds emphasis on related activities and possible competitiveness benefits. Clusters and the New Economics of Competition presents an influential policy-oriented rationale for geographically concentrated complementarities. An analytical definition must specify a geographic scale, industrial boundary, observable connections, and a counterfactual. Otherwise almost any town with several similar firms could be labeled a cluster after it succeeds.
Deconstructing Clusters cautions that the term has been used for distinct phenomena and frequently overextended into policy promises. A policy can announce a cluster, but naming a group does not create the specialized suppliers, trust, skilled workers, markets, or innovation capabilities claimed as benefits. A cluster assessment should begin with actual supplier-buyer ties, worker mobility, shared services, innovation outputs, and entry rates, and then compare with plausible alternative locations.
Knowledge flows across spatial and organizational boundaries. the local-buzz and global-pipelines analysis emphasizes the interplay of routine local interaction with distant, purposive knowledge networks. A geographically compact industrial community can be dependent on distant technology, foreign demand, diaspora relationships, or multi-site firms. Policies confined to local networking can neglect the international links through which a region upgrades technology and reaches markets.
Proximity and Innovation differentiates cognitive, organizational, social, institutional, and geographic proximity. Geographic co-location can support interaction, but organizational and cognitive compatibility may be more important than kilometers, while excessive closeness can reinforce lock-in. An innovation-oriented region may benefit from both related specialization and links to dissimilar external knowledge. This argument clarifies why a strategy focused only on local density can reproduce technological rigidity.
Cluster policy should consequently be evaluated through explicit mechanisms: does shared training correct coordination failures, do laboratories lower fixed experimentation costs, do open procurement and interoperability widen participation, and do transport investments connect firms to customers? Subsidizing a cluster's incumbent firms can entrench rents without creating a broader regional capability. Firm-level survival, innovation, wages, accessibility, and public cost should be assessed together.
13. Global Production Networks and Value Chains
Global value chains and production networks connect geographically separated stages of design, research, financing, component manufacturing, assembly, logistics, marketing, and after-sale services. Geographic fragmentation occurs when different locations offer complementary capabilities or lower generalized costs, and when coordination technology, logistics, and trade rules permit cross-border production. The final exporter's address is consequently a poor indicator of where all of an industry's value and knowledge are created.
The governance of global value chains distinguishes types of coordination between lead firms and suppliers using transaction complexity, codifiability, and supplier capability. Relational, modular, captive, hierarchical, and market forms imply different demands on local skill, autonomy, finance, and quality standards. A region with many assembly jobs may occupy a low-margin position that offers limited opportunities for design or technical upgrading. A small region hosting research, standards, or platform coordination can exert influence exceeding its physical production volume.
Global Production Networks: Realizing the Potential places territorial embeddedness and multi-actor relations at the center of interpretation. Suppliers respond to customers, labor conditions, finance, logistics operators, state agencies, and industry associations. Regional value capture depends on ownership, bargaining power, skill formation, taxation, intellectual property, and the ability to attract or develop alternative partners. The production network perspective therefore complements equilibrium models that aggregate goods and bilateral trade flows.
Global Shift's industrial geography explains why globalization can simultaneously disperse production stages and concentrate high-value command, specialized services, and research. Dispersed physical assembly does not ensure geographically dispersed profits. Conversely, local upgrading can arise when training and supplier capability development enable entry into more complex stages. Export volumes alone should not be used as a sufficient proxy for locally retained value added.
Risk also travels through networks. A flood in one manufacturing center, closure of a maritime chokepoint, export restriction, or failure of a highly specialized component supplier can propagate to firms far from the initial disturbance. Resilience can improve through diversification, inventories, standardization, or alternative routes, but each adjustment has costs. Research must distinguish gross network centrality from vulnerability to a specific shock, identify supply elasticities, and account for rerouting capacity.
14. International Borders, Corridors, and Connectivity
International borders convert geographic proximity into varying degrees of economic separation. Differences in tariffs, customs procedures, regulatory recognition, visas, currency, security, contract enforcement, and political risk affect otherwise similar trade routes. A factory a short drive from a neighboring country's capital can face greater transaction costs than one served by a stable transoceanic shipping lane. Borders are therefore a legal and institutional component of economic distance as well as physical lines on a map.
Trade corridors organize flows around transport infrastructure, distribution centers, ports, storage facilities, special economic zones, and production clusters. Their performance depends on reliability and coordination, not merely road length. A new rail line may reduce marginal shipping time but have modest economic effect if customs delays dominate end-to-end journey cost. Conversely, coordinated border inspection and digital freight documentation can improve access without major construction. The relevant outcome is a change in generalized cost for users and producers.
Railroads of the Raj investigates historic Indian rail expansion using trade and price data within an explicit equilibrium framework, demonstrating why transport access can alter real incomes through prices and trade opportunities. China's trunk highway study provides evidence that closer integration can redistribute industrial activity rather than mechanically create factories in every newly connected peripheral area. Together these studies reject a simple monotonic policy assumption that every new connection produces identical local industrial effects.
Corridor evaluation must define its geographic boundary. A provincial impact may consist partly of firms relocating from a neighboring province; a country's net effect may differ from either local gross outcome. Trade benefits may accrue to consumers through lower prices or firms through input access even if employment in a protected local industry declines. A complete evaluation needs trade-cost changes, resource allocation, spatial spillovers, public financing, ecological damage, and adjustment costs.
Borders can also produce productive complementarities, where manufacturing, services, and labor markets span jurisdictions. Cross-border metropolitan areas require coordination in infrastructure, land use, taxation, environmental enforcement, and worker mobility. Public law and political decisions become endogenous determinants of the economic map rather than peripheral context for an otherwise purely geographic model.
15. Labor Markets, Migration, and Local Opportunity
Regional labor markets connect residence, workplaces, firms, transport, training, and social institutions. Workers select jobs in response to expected wages, conditions, career prospects, housing costs, family ties, language, and the risks of moving. Firms choose locations in response to skills, wage expectations, business partners, and the costs of recruiting. Labor mobility may be restricted by credentials, legal status, liquidity, home ownership, caregiving, and information. A spatial model with perfectly mobile homogeneous workers is therefore an illuminating benchmark but an incomplete description of observed adjustment.
An operational distinction separates nominal wages, real wages, disposable income, and expected lifetime opportunity. A metropolis with a large nominal wage premium may offer smaller consumption gains after housing and commuting costs. Conversely, low-cost regions may offer weak employment prospects or limited healthcare, education, and social infrastructure. The 2026 OECD place-based indicators compare regional productivity, job opportunities, population aging, and housing, showing how multiple outcomes can vary jointly rather than following one labor-market ranking.
Tradable-sector shocks can have effects on nontradable employers through household demand. Moretti's analysis of local employment multipliers demonstrates the logic of demand-driven local spillovers but also makes clear that a multiplier does not by itself prove a market failure. A worker laid off in one region may find employment nearby, leave the labor force, or relocate; the adjustment path depends on sectoral skills, available housing elsewhere, and local social insurance.
The China Syndrome exploits differences in local United States commuting-zone exposure to import competition to study employment and earnings adjustment. Its regional framework is important precisely because national gains from trade and local labor-market costs can coexist. Geographic concentration of dislocation creates fiscal and social effects not captured by a national average tariff or consumer-price calculation.
Migration changes both origin and destination. It can relieve labor scarcity, transfer skills, and support remittances; it can also change housing demand and public-service needs and reduce the origin's workforce. The welfare of incumbents, migrants, landowners, employers, and future generations should be analyzed separately, followed by a transparent aggregation rule if a social welfare judgment is required. Relocation is a choice with costs and identity-specific attachments, not simply a frictionless flow restoring wage equalization.
16. The Geography of Inequality and Social Mobility
Spatial inequality is expressed in differences in employment, productivity, income, wealth, housing access, amenities, ecological risk, and the ability to leave disadvantaged places. A high-productivity locality may have low living standards for some residents because rents, congestion, and exclusion absorb the benefits of economic concentration. A low-productivity region may have pockets of valuable ecological services, informal production, social support, and affordability absent from recorded value added. The welfare meaning of any geographical disparity requires specified indicators and populations.
There are three distinct objects of equity analysis. Between-place inequality compares average outcomes across regions. Within-place inequality measures differences among persons or neighborhoods inside each region. Origin-based or life-course inequality tracks how a child's or adult's location and mobility shape long-run outcomes. A fall in average regional disparities can coexist with widening inequality inside major cities, and a change in region-level ranking can partly arise from migration rather than improvements for existing residents.
Childhood exposure effects investigates how time spent in different neighborhoods affects later mobility outcomes. The analysis makes neighborhood context a potentially causal component of opportunity, while requiring careful attention to selection, timing, and heterogeneity. It also demonstrates that the same address can have a different effect on children and adults because exposure duration and life stage matter.
Housing supply and transport access mediate inequality. When access to productive jobs requires residence in a high-rent metropolitan region, households with limited resources may be excluded even if firms demand labor. Improving a disadvantaged region's economic base may be beneficial, but this is one of several strategies; allowing access to existing opportunity centers can also increase welfare. The 2026 regional report reports widespread metropolitan population concentration and housing pressure across countries with data, strengthening the case for evaluating both location-based investment and barriers to residential mobility.
Spatial disparities create political-economy feedbacks: tax bases affect local schools and infrastructure, economic shocks alter municipal revenues, and access to administrative power shapes which investments receive priority. Territorial policy choices thus embed judgments about place, people, historical obligations, collective infrastructure, and intergenerational equity. A serious comparison reveals distribution across groups and periods rather than relying solely on the growth rate of a regional GDP aggregate.
17. Natural Resources, Agriculture, and the Environment
Natural resource endowments influence the initial geography of activities but do not determine their ultimate location or economic value. Agricultural production depends on soil, water, climate, land rights, transport and technology; extraction depends on geology, capital intensity, infrastructure, global prices, environmental regulation, and community consent. Processing can locate near resource inputs when materials lose substantial weight in transformation, or near consumers when final outputs are costly or perishable to transport. The industrial transport-cost approach makes this conditional choice explicit.
Resource dependence can increase vulnerability to commodity cycles and environmental degradation. The effect of a resource windfall on regional growth depends on ownership, fiscal institutions, domestic linkages, labor-market demand, and how revenues are invested. An extractive enclave may support export value while purchasing few inputs locally; a diversified production ecosystem can distribute learning and supplier opportunities more widely. National resource exports and locally captured income are different measures.
Land and ecological systems also constrain urban expansion, transport siting, and industrial pollution. Cities and infrastructure can expose households to floods, heat, drought, wildfire, coastal storms, and system-wide cascading failure. Cities, Settlements and Key Infrastructure provides a scientific assessment of differentiated climate risks and adaptation; the impacts vary with urban form, inequality, governance, and physical exposure. A locational productivity premium measured under historic climate should not be extrapolated unchanged if future hazards alter insurance, labor health, water supply, or asset integrity.
The spatial scale of an environmental externality differs from an industry's administrative location. Upstream pollution can damage downstream water supplies, and greenhouse-gas emissions contribute to global climatic effects. Carbon-intensive regional employment can therefore coexist with national or global social costs. Appropriate evaluation considers where damages fall, who benefits from production, whether alternatives exist, and the distributional consequences of transition policies.
Regional environmental policy must work with opportunity cost. Protecting a wetland may preserve flood buffering and biodiversity while limiting developable land; restoring a contaminated industrial site may create health benefits and additional land supply. Economic geography contributes spatially explicit trade-offs, making it possible to identify which populations and ecological functions are affected rather than treating 'the environment' as an undifferentiated national category.
18. Digital Infrastructure, Knowledge Work, and Artificial Intelligence
Digital connectivity changes the location requirements of some tasks but does not eliminate the economic significance of geography. Software and data may move cheaply once networks exist, while training, financing, specialized hardware, energy, legal jurisdiction, research communities, and face-to-face coordination remain geographically uneven. Remote work can separate residence from employer headquarters, but workers' access to broadband, housing, time zones, language communities, and specialized labor markets continues to matter.
Digital service platforms can concentrate control and intangible income while distributing users and suppliers globally. Data centers depend on electricity supply, grid reliability, water use, land prices, network latency, disaster exposure, and legal constraints. Software clusters depend on specialized skills, universities, professional networks, risk capital, and procurement networks. Geographic concentration of innovation is consequently compatible with global diffusion of finished software.
OECD Regions and Cities at a Glance 2026 reports that in 2025, 52% of employment in the OECD regions with the highest shares of tertiary-educated workers was in occupations highly exposed to generative artificial intelligence, compared with 42% elsewhere, according to its covered regional sample. Occupational exposure estimates are indicators of potential task interaction with technology, not observed job destruction, adoption, or productivity changes. The report also finds large shares of patenting and venture-capital funding concentrated in the largest functional urban areas or individual small regions in many reporting countries; those concentrations are defined by its national comparison denominators.
The spatial consequences of AI-mediated production depend on complementary infrastructure and capabilities. Where education, compute, finance, demand, and organizational learning are concentrated, early adoption may reinforce existing spatial advantages. Where low-cost connectivity and training expand, some routine tasks can disperse. The net result requires identifying actual firm adoption, local skill formation, energy demand, service provision, and location choices. A prediction that digital technologies necessarily flatten the world is as incomplete as a prediction that every digital industry must cluster.
A useful empirical comparison distinguishes four geographic variables: where technology is invented, where the provider is legally and operationally located, where the technology is adopted, and where economic value and labor displacement are recorded. The four maps may differ dramatically. Economic geography can assess these differences with firm data, labor-market surveys, patent records, network statistics, and geographically disaggregated productivity measures.
19. Economic History, Path Dependence, and Shocks
Economic landscapes accumulate physical and institutional capital over long periods. Ports, railway junctions, mines, legal systems, housing stock, schools, occupational cultures, and corporate networks can persist after the original reasons for their creation weaken. A region's current comparative advantage is therefore partly inherited. Historical geography does not imply inevitability: technologies, border changes, state investments, and migration can alter the value of inherited assets.
Path dependence has at least three analytic meanings. First, sunk costs and network complementarities make relocation expensive. Second, learning and institutional routines channel incremental innovation toward existing skills. Third, self-reinforcing expectations and demand linkages can sustain a concentration that was initially contingent. The dynamic interpretations of core-periphery concentration and migration in agglomeration models connect some of these mechanisms to explicit economic equilibria.
Historic disruptions are useful for identifying spatial processes because they alter accessibility and institutional arrangements in ways that may be partially exogenous to local firms' contemporaneous productivity. The Berlin Wall study uses division and reunification as a setting for understanding urban density and land use. Historical railroad expansion, examined in Railroads of the Raj, also offers variation in access whose interpretation depends on the actual decision process that generated routes and their timing.
Researchers should resist technological determinism and presentism. A historical manufacturing cluster may have existed because of coal, water power, land law, empire, forced labor, or migration restrictions, none of which can be inferred from today's spatial correlation alone. A historical map and a contemporary map may assign identical names to geographic areas whose boundaries, populations, and legal regimes differ. Historical comparisons need harmonized boundaries and explicit evidence about institutions and methods.
Future trajectories are therefore conditional and open. Physical investment and institutions can create lock-in; competition, public infrastructure, capability acquisition, and environmental shocks can disrupt it. Economic geography explains how cumulative advantage forms and changes, and it specifies which historical facts are necessary before asserting that a current regional pattern results from a particular legacy.
20. Spatial Data, Mapping, and Measurement
The evidence base of economic geography includes administrative employment and tax records, business registers, household and labor-force surveys, land and property transactions, customs and freight records, commuting flows, patent and research data, establishment geocodes, satellite imagery, mobile mobility data, shipping and platform networks, historical maps, and qualitative fieldwork. Each source measures a constructed observable, not economic reality in full. A firm register may locate an enterprise at its headquarters rather than at its plants; a household survey may assign income by residence rather than workplace; trade data may register port of customs clearance rather than site of production.
Geographic data processing begins with coordinate systems, boundary definitions, temporal versions, data provenance, and disclosure risks. Coordinates from different reference systems cannot be combined safely without transformation. Historic municipal mergers alter denominators. A point inside a polygon can change classification when the border moves even if the establishment has not. Spatial joins require documented error tolerance for boundaries, offshore locations, and uncertainty of addresses. Data harmonization is a substantive research task rather than a clerical prelude.
Density maps and choropleths are sensitive to area size and classification bins. Counts can make large-population regions appear to dominate; per-capita rates can become volatile in very small zones. The spatial-configuration problems addressed by Spatial Data Configuration show why researchers should report sensitivity to alternative partitions. A harmonized degree-of-urbanisation manual offers an example of explicit geographic criteria designed for cross-country comparability.
Remote sensing supplies complementary measures where official statistics are weak. Measuring Economic Growth from Outer Space develops the use of nighttime lights to supplement measures of economic activity. Brightness is not GDP: electricity access, lighting technology, policies, gas flaring, saturation, satellite sensor changes, and seasonal conditions can change recorded luminosity for reasons independent of value added. A reproducible study must explain calibration, sensor vintage, geographic aggregation, and the set of official benchmarks.
Qualitative methods remain indispensable where causal mechanisms are organizational, legal, or relational. Interviews may reveal how firms select suppliers, why workers reject relocation, or how local agencies coordinate investments. Network tracing can distinguish formal ownership links from actual flows of orders or knowledge. A credible mixed-method design states when interview evidence supports a mechanism, when it is exploratory, and when an inference requires representative quantitative evidence.
21. Spatial Econometrics, Identification, and Inference
Spatial econometrics addresses regression models whose observations are connected through geography or networks. A basic spatial autoregressive expression is y = ρWy + Xβ + ε, where W is a prespecified spatial weights matrix and Wy measures outcomes among connected units. If the model is structurally interpreted and the relevant inverse exists, a shock may propagate through (I − ρW)^(−1). Such a system requires assumptions on the direction and nature of interaction; a spatial lag is not automatically a causal spillover. Spatial Econometrics: Methods and Models provides foundational analysis of dependence and estimation problems.
A spatial error structure instead represents errors correlated across space, such as ε = λWε + u. Distinguishing outcome interaction from spatially correlated omitted shocks can be difficult. Spatial Durbin formulations incorporate spatially lagged covariates; spatial panel models add unit and time heterogeneity; network methods permit relationships independent of physical adjacency. The weights matrix is therefore a scientific hypothesis specifying exposure pathways. It should be justified and assessed under alternatives, not selected solely to maximize a statistic.
Exploratory measures identify geographic clustering. Local Indicators of Spatial Association introduces local spatial association measures useful for detecting potential hotspots and outliers. Multiple testing, scale sensitivity, and unequal population sizes affect interpretation. A visually striking hotspot map cannot establish causation or a mechanism without design and comparative evidence.
Causal identification requires a clearly stated estimand, treatment, comparison, and interference structure. Random infrastructure placement is rare. Researchers may exploit historical route plans, terrain-based costs, regulatory boundaries, unexpected plant-location decisions, or shocks whose assignment is credibly independent of counterfactual trends. Difference-in-differences requires suitable parallel counterfactual trends under its design; instrumental variables require exclusion restrictions; geographic regression discontinuity requires appropriate continuity near borders. None of these assumptions follows from the mere presence of a map.
Cross-boundary spillovers threaten simple treatment-control comparisons. A new transport route affects neighborhoods near but outside the administrative treatment group, while industrial relocation can make an untreated region lose employment. Standard errors should allow for the likely level of shared shocks and dependence, and estimates should disclose sensitivity to geographic clustering and alternative definitions of treatment exposure. The estimand may be a local average effect, a net place effect, a national aggregate effect, or a distributional effect, and each requires different assumptions.
22. Comparative Evidence on Transport and Local Spillovers
Transport studies provide unusually clear examples of why economic geography requires both causal design and spatial theory. In Railroads of the Raj, changes in rail accessibility are related to costs, trade, agricultural prices, and real income through a structural empirical design. The study demonstrates the inferential value of matching observed regional outcomes to a model of trade frictions while controlling for geographic and historical confounders. Its results are specific to colonial-era Indian production and transportation institutions; they do not yield a transport-return coefficient applicable to every future road or railway.
The implications of connectivity can also be counterintuitive. Trade Integration, Market Size, and Industrialization examines the industrial consequences of China's trunk highway system and shows that improved connections can favor existing production cores over certain peripheral areas. Better access can expose local firms to more efficient competitors while expanding market opportunities for other firms and consumers. A null or negative change in peripheral manufacturing employment is not equivalent to a zero consumer welfare effect.
Firm-location events offer another source of variation. the large-plant openings study compares locations selected by major plants with competing finalist locations to estimate localized productivity effects on incumbent firms. Its design is stronger than an ordinary comparison of counties with and without plants because final candidate places share some observed and unobserved attributes. It still requires a defensible account of final site selection and of treatment spillovers; identifying gains among incumbent firms does not alone evaluate public subsidies used to attract a new plant.
Studies of urban productivity confront firm and worker sorting. the French productivity comparison distinguishes selection from agglomeration in distributional patterns, helping to explain why differences in average productivity across cities should not all be treated as city-caused. the Berlin density study combines historical quasi-experimental variation and quantitative spatial structure; this blend of identification and equilibrium reasoning is powerful because mobility and rents adjust simultaneously.
Local labor-market research makes adjustment costs visible. the import-competition commuting-zone study documents geographically heterogeneous exposure to trade shocks; Local Multipliers tracks secondary employment responses to changes in tradable production. Evaluating a shock therefore requires a temporal horizon. Short-run unemployment, medium-run migration, and long-run industrial reallocation can give different signs and distributions even for the same intervention.
The convergent lesson is methodological. Treat cross-place comparisons as potentially confounded, treat firm or household movement as part of the effect, and use structural counterfactuals only within empirically supported parameter ranges. A credible research paper explains which elements are directly identified, which are imposed by model structure, and which remain uncertain.
23. Spatial Policy, Evaluation, and Institutional Design
Spatial policy is an attempt to alter the distribution of economic access, activity, risk, or welfare across places. Its instruments include transport infrastructure, housing supply, land-use regulation, vocational training, university and research investment, public services, fiscal equalization, industrial subsidies, environmental protection, territorial governance, and cross-border cooperation. A coherent policy argument names the market, coordination, or institutional failure it seeks to address and the welfare criterion used to judge outcomes.
The first evaluation question is whether a proposed intervention corrects an identifiable problem. Agglomeration spillovers may justify supporting shared innovation infrastructure, while congestion and pollution may justify pricing or regulation. A shortage of housing supply can prevent workers from entering high-productivity cities even when jobs exist. Remoteness may call for reliable basic services rather than attempts to recreate the industrial structure of a major metropolitan center. The World Development Report on economic geography argues for integration and connectivity alongside place-specific attention, a useful starting framework whose implementation must be examined against local institutions.
The second question concerns who receives benefits. A place-based grant can raise land values for incumbent owners; a new railway can benefit consumers who purchase cheaper goods; a local training program can help mobile workers who subsequently leave the area. The social evaluation should account for these channels rather than measure success by gross jobs announced within the designated zone. The 2026 OECD report underscores spatially uneven housing access, knowledge work, population change, and risk—conditions that make distribution central to policy assessment.
The third question concerns displacement and aggregate effects. If a public subsidy induces a firm to move across a municipal boundary, the receiving area may report new jobs while the country has no net job gain. If a cluster program mostly funds incumbents who would have invested anyway, additionality is low. A useful evaluation defines a baseline, credible counterfactual, time horizon, relevant spatial spillover radius, program costs, and fiscal consequences.
The fourth question is institutional feasibility. Public procurement, permitting, fiscal capacity, local representation, data quality, and intergovernmental coordination affect whether a formally efficient intervention works in practice. Policies often cross administrative borders while jurisdictions do not. A metropolitan labor market may need regional transport governance, and a watershed may demand coordination across counties or countries.
A defensible spatial policy portfolio can combine broad access to education, healthcare and connectivity with targeted investments where empirically documented complementarities exist. The appropriate mix changes across places. Comparing locations requires attention to initial conditions, project crowding-out, administrative capacity, environmental constraints, and residents' own preferences about moving or staying.
24. Contemporary Research Frontiers and Open Problems
Current research asks how new technologies, global fragmentation, demographic change, and environmental shocks interact with geography. The 2026 regional evidence draws attention to the concentration of generative-AI occupational exposure, knowledge-intensive activities, finance, and patents, as well as uneven aging and climate vulnerability. This combination raises a question about future regional divergence that cannot be answered by occupational exposure rates alone: do adoption, wages, employment, and returns to place follow the same geographical pattern?
Housing and migration remain open research fronts because productivity and amenities attract residents while supply restrictions alter who can take advantage of opportunity. Better linked housing transaction, tax, employer, and commuting data can distinguish household gains from land-price capitalization. Research must also examine people unable or unwilling to move and the role of social networks, age, care duties, and migration policy in shaping mobility. The intersection of neighborhood exposure research with spatial equilibrium analysis offers a way to study both long-term individual outcomes and systemic responses to interventions.
Industrial decarbonization changes the location incentives of energy-intensive activities. Electricity transmission, low-carbon power availability, water limits, climate hazards, and geopolitical supply risks can reorganize production networks. The IPCC urban climate assessment identifies risk inequalities that intersect with urban infrastructure and land-use. A robust spatial model of transition must include fixed investments, network capacity, climate uncertainty, and distributional consequences, not just current local energy prices.
Another frontier is multi-layer network exposure. Economic shocks travel through financial, input-output, labor, transport, and information links, each with a different geography. Interregional dependencies can amplify a disaster or speed recovery if alternative routes and suppliers exist. Global production-network research offers a relational account; quantitative spatial approaches such as the spatial economics synthesis supply disciplined equilibrium counterfactuals. Integrating both requires empirical measurement of actual network edges and governance, including confidentiality and uneven data coverage.
Methodological questions remain difficult: how to account for endogenous boundary definitions; how to identify treatment when every place is connected to others; how to reconcile real-time mobility signals with slow-changing census measures; how to protect privacy in granular place-based data; how to measure informal production and unpaid labor; and how to avoid using satellite brightness or patent counts as if they fully captured welfare. Research transparency demands publishing definitions, code, geographic crosswalks, uncertainty, and lawful access methods when possible.
The decisive unresolved problem is the simultaneous explanation of productivity, access, and distribution under changing institutions and geography. Many studies identify one margin well; fewer provide credible evidence about aggregate welfare and heterogeneous effects over long periods. Progress depends on combining methodological disciplines while retaining explicit source-to-claim traceability.
25. Integrative Scientific Synthesis
A synthesis of the reviewed evidence begins with four observable layers: locational endowments (land, climate, resources, amenities, institutions), networks (transport, trade, labor, production, finance and information), allocations (firms, households, sectoral employment and capital), and outcomes (prices, productivity, incomes, rents, exposure, mobility and welfare). Their relationships can be assessed through existing location theory, agglomeration analysis, quantitative equilibrium and causal research designs. The sequence here is a reader-oriented way to organize findings, not an independent economic doctrine or a new formal model.
The first analytical step is to establish the geographic unit and time frame. Statements about metropolitan wage premiums, regional diversification, neighborhood mobility, and multinational value capture refer to different populations and scales. Reconciliation requires explicit crosswalks, not a generic term such as 'the local economy.' harmonized settlement classifications and spatial configuration research show why even a well-defined numerator becomes misleading when boundaries are inconsistent.
The second step is to separate access from agglomeration. Connectivity changes the effective market size and available options; agglomeration changes productivity or matching through concentration. They interact, but a highway's capacity to reduce transport cost is not identical to the external productivity effect of a local supplier cluster. market-potential estimation examines demand access, while the agglomeration microfoundations specify mechanisms by which proximity changes economic performance.
The third step is to distinguish private returns, local accounting outcomes, and social welfare. A plant opening can raise surrounding productivity, change land prices, redistribute workers, and create fiscal obligations. Summed local gross employment effects can exceed or fall below national net gains. evidence from large plant openings addresses localized spillovers; quantitative spatial welfare analysis illustrates why market responses across locations matter for an aggregate counterfactual. Public appraisal must state whose welfare counts and whether future adaptation and ecological risk are included.
The fourth step is to connect institutions and historical development to economic mechanisms. A region's infrastructure and technological capacity are constructed over time, and firms operate within ownership, labor, legal, and governance relationships. value-chain governance and global production networks demonstrate why local production value cannot be read directly from the geography of recorded exports. The same production volume may generate different local capability growth depending on contracts, supplier relationships, and control of intangible assets.
The fifth step is to treat evidence strength as claim-specific. A historical primary text establishes the architecture of a theory, not its modern empirical validity. An international report establishes indicator definitions and observed sample patterns, not causality. A quasi-experimental study provides evidence for a defined historical setting and identification assumptions, not unlimited generalization. A quantitative model supplies conditional counterfactuals, with scope tied to the credibility of primitives and fit. The journalistic temptation to collapse all these into one 'geography causes growth' claim is analytically unsustainable.
The scientific conclusion is that spatial concentration and territorial unevenness are generated by interacting market, network, institutional, and physical mechanisms. They are historically contingent and measurable; they can support productivity while imposing costs and distributing gains unequally. Economic geography contributes a disciplined explanation of these patterns by connecting locational decisions, networks, equilibrium adjustments, causal evidence, and social evaluation within a common geographic frame.
Conclusion
Economic geography explains the spatial organization of economic activity through location choices, accessibility, agglomeration, specialization, networks, labor and residential movement, land and environmental constraints, and inherited institutions. Its intellectual history proceeds from agricultural and industrial location to settlement hierarchies, regional science, political and institutional geography, new economic geography, and contemporary quantitative and geospatial research. Each tradition contributes methods and questions that remain useful when deployed within their evidentiary scope.
The discipline's central analytical result is the coexistence of concentration benefits and spatially distributed costs. Transport integration can enlarge markets while reallocating manufacturing across regions. Dense labor markets can improve matching while restricted housing raises the cost of accessing opportunity. Production networks can distribute manufacturing while concentrating high-value decision-making elsewhere. Regional income data can show an improvement while the original residents experience little gain if migration and rents are not tracked. These outcomes follow from established mechanisms and documented empirical studies; they demand specific, causal, and welfare-aware analysis.
Methodologically, robust geographic research requires precise spatial units, reliable source provenance, identification strategies suited to interference and mobility, appropriate treatment of spatial dependence, and sensitivity to the boundaries over which impacts are calculated. Policy conclusions require explicit distributional and ecological accounting. Current issues of AI adoption, housing shortages, climate adaptation, and supply-chain restructuring reinforce the importance of these principles.
The monograph's original scholarly contribution is an integrated and critically evaluated reading of the field's main explanatory and evidentiary traditions, organized so that concepts, formal assumptions, measured outcomes, and claims of causal identification remain distinguishable. It makes no claim to establishing a new proprietary economic theory. Economic geography remains essential because every economic process has a geography of production, access, exchange, adaptation, and consequence.
References
Ahlfeldt, G. M., Redding, S. J., Sturm, D. M., & Wolf, N. (2015). The economics of density: Evidence from the Berlin Wall. Econometrica, 83(6), 2127–2189. DOI
Allen, T., & Arkolakis, C. (2014). Trade and the topography of the spatial economy. Quarterly Journal of Economics, 129(3), 1085–1140. DOI
Anselin, L. (1988). Spatial econometrics: Methods and models. Kluwer Academic Publishers. DOI
Anselin, L. (1995). Local indicators of spatial association—LISA. Geographical Analysis, 27(2), 93–115. DOI
Arbia, G. (1989). Spatial data configuration in statistical analysis of regional economic and related problems. Kluwer Academic Publishers. DOI
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. DOI
Bathelt, H., Malmberg, A., & Maskell, P. (2004). Clusters and knowledge: Local buzz, global pipelines and the process of knowledge creation. Progress in Human Geography, 28(1), 31–56. DOI
Boschma, R. (2005). Proximity and innovation: A critical assessment. Regional Studies, 39(1), 61–74. DOI
Chetty, R., & Hendren, N. (2018). The impacts of neighborhoods on intergenerational mobility I: Childhood exposure effects. Quarterly Journal of Economics, 133(3), 1107–1162. DOI
Christaller, W. (1966). Central places in southern Germany (C. W. Baskin, Trans.). Prentice-Hall. (Original work published 1933). Original source
Coe, N. M., Dicken, P., & Hess, M. (2008). Global production networks: Realizing the potential. Journal of Economic Geography, 8(3), 271–295. DOI
Combes, P.-P., Duranton, G., Gobillon, L., Puga, D., & Roux, S. (2012). The productivity advantages of large cities: Distinguishing agglomeration from firm selection. Econometrica, 80(6), 2543–2594. DOI
Dicken, P. (2015). Global shift: Mapping the changing contours of the world economy (7th ed.). Guilford Press. Original source
Donaldson, D. (2018). Railroads of the Raj: Estimating the impact of transportation infrastructure. American Economic Review, 108(4–5), 899–934. DOI
Duranton, G., & Puga, D. (2004). Micro-foundations of urban agglomeration economies. In J. V. Henderson & J.-F. Thisse (Eds.), Handbook of regional and urban economics (Vol. 4, pp. 2063–2117). Elsevier. Original source
Ellison, G., Glaeser, E. L., & Kerr, W. R. (2010). What causes industry agglomeration? Evidence from coagglomeration patterns. American Economic Review, 100(3), 1195–1213. DOI
Eurostat. (2021). Applying the degree of urbanisation: A methodological manual to define cities, towns and rural areas for international comparisons. Publications Office of the European Union. Original source
Faber, B. (2014). Trade integration, market size, and industrialization: Evidence from China’s National Trunk Highway System. Review of Economic Studies, 81(3), 1046–1070. DOI
Fujita, M., Krugman, P., & Venables, A. J. (1999). The spatial economy: Cities, regions, and international trade. MIT Press. DOI
Gereffi, G., Humphrey, J., & Sturgeon, T. (2005). The governance of global value chains. Review of International Political Economy, 12(1), 78–104. DOI
Glaeser, E. L., Kallal, H. D., Scheinkman, J. A., & Shleifer, A. (1992). Growth in cities. Journal of Political Economy, 100(6), 1126–1152. DOI
Greenstone, M., Hornbeck, R., & Moretti, E. (2010). Identifying agglomeration spillovers: Evidence from winners and losers of large plant openings. Journal of Political Economy, 118(3), 536–598. DOI
Hanson, G. H. (2005). Market potential, increasing returns and geographic concentration. Journal of International Economics, 67(1), 1–24. Original source
Henderson, J. V., Storeygard, A., & Weil, D. N. (2012). Measuring economic growth from outer space. American Economic Review, 102(2), 994–1028. DOI
Intergovernmental Panel on Climate Change. (2022). Cities, settlements and key infrastructure. In Climate change 2022: Impacts, adaptation and vulnerability (Working Group II contribution, Chapter 6). Cambridge University Press. Original source
Isard, W. (1956). Location and space-economy: A general theory relating to industrial location, market areas, land use, trade, and urban structure. Technology Press of MIT and Wiley. Original source
Krugman, P. (1991). Increasing returns and economic geography. Journal of Political Economy, 99(3), 483–499. DOI
Martin, R., & Sunley, P. (2003). Deconstructing clusters: Chaotic concept or policy panacea? Journal of Economic Geography, 3(1), 5–35. DOI
Moretti, E. (2010). Local multipliers. American Economic Review, 100(2), 373–377. DOI
Organisation for Economic Co-operation and Development. (2026). OECD regions and cities at a glance 2026. OECD Publishing. DOI
Ottaviano, G. I. P., Tabuchi, T., & Thisse, J.-F. (2002). Agglomeration and trade revisited. International Economic Review, 43(2), 409–435. DOI
Porter, M. E. (1998). Clusters and the new economics of competition. Harvard Business Review, 76(6), 77–90. Original source
Redding, S. J., & Rossi-Hansberg, E. (2017). Quantitative spatial economics. Annual Review of Economics, 9, 21–58. DOI
Rosenthal, S. S., & Strange, W. C. (2004). Evidence on the nature and sources of agglomeration economies. In J. V. Henderson & J.-F. Thisse (Eds.), Handbook of regional and urban economics (Vol. 4, pp. 2119–2171). Elsevier. Original source
United Nations Department of Economic and Social Affairs, Population Division. (2025). World urbanization prospects 2025. United Nations. Original source
Weber, A. (1929). Theory of the location of industries (C. J. Friedrich, Trans.). University of Chicago Press. (Original work published 1909). Original source
World Bank. (2009). World development report 2009: Reshaping economic geography. World Bank. Original source