RESEARCH DOSSIER
Global Value Chains and Systemic Economic Interdependence
Abstract. Global value chains distribute productive activities, income generation, and economic exposure across national borders. This Research Dossier examines their historical foundations, organizational structure, measurement, and mechanisms of disruption. It integrates trade theory, value-added accounting, input-output analysis, network economics, natural-experiment evidence, and institutional studies of resilience and economic fragmentation. Transparent numerical illustrations distinguish demand requirements, binding supply constraints, and cost transmission, and demonstrate the consequences of correlated supplier failures. The empirical discussion differentiates reported findings, harmonized statistical data, experimental nowcasts, and modeled counterfactuals. The analytical synthesis establishes the importance of distinguishing income attribution, network exposure, and operational feasibility when assessing systemic vulnerability. The study identifies the implications for macroeconomic adjustment, development, financial continuity, environmental accounting, and targeted resilience policy. Its conclusions are grounded in established research and conditional reasoning; it does not claim independent econometric replication or a new named economic theory.
Keywords: Global value chains; systemic economic interdependence; trade in value added; input-output analysis; production networks; shock propagation; supply chain resilience; input specificity; macroeconomic adjustment; economic development; geoeconomic fragmentation
Introduction
Modern production is an arrangement of geographically distributed capabilities. A finished product incorporates the results of extraction, engineering, manufacturing, logistics, finance, software, commercial organization, and labor performed in several jurisdictions. Its physical arrival at a border reveals only one transaction in a longer sequence of production. The central problem of this dossier is how that sequence shapes the efficiency, distribution, and vulnerability of an economy. A local disruption becomes systemic when other producers and final users experience consequences through their dependencies, and when responses within the system alter the original disturbance.
The distinction between the value of international transactions and the value generated in production provides the analytical starting point. Gross trade measures transactions across borders; value added measures contributions to output after intermediate consumption has been deducted. The structure of transactions also conveys dependencies, which cannot be inferred completely from the distribution of income. A supplier may capture little of the final selling price while being indispensable in the short run. Conversely, a large value-added activity may have several readily available alternatives. A theory of interdependence therefore requires an account of both quantities and constraints.
The World Development Report 2020 located global value chains close to half of world trade at its reporting horizon. This estimate is historical and depends on the report's definition of chain-related trade. Its significance here is structural: trade is increasingly connected with the organization of production itself. When a disruption affects an intermediate activity, the relevant question becomes how that activity enters the feasible production plans of distant firms. The same architecture that raises productivity by enabling specialization can expose a firm to events far beyond its own physical location.
The inquiry proceeds from the historical foundations of specialization to formal accounting, organization, transmission, evidence, macroeconomic adjustment, and policy. It uses published research and institutional statistics, together with transparent hypothetical calculations. The author's contribution is the integration of established mechanisms and explicit comparisons of what their different measures can and cannot establish. No proprietary dataset, new econometric estimates, or claim of a previously unknown general theory is presented.
Historical foundations of international specialization
Smith's 1776 discussion of the division of labor begins with the fragmentation of a productive operation into distinct activities. The resulting productivity advantage arises from skill formation, the reduction of switching time, and improvements in tools. What is visible in a single workshop is magnified when separate firms acquire specialized capabilities. The contemporary cross-border chain extends this organizational possibility across legal jurisdictions and distances, subject to the costs of coordination.
Ricardo's 1817 chapter on foreign trade establishes a different mechanism. England's hypothetical labor requirements for cloth and wine are 100 and 120, compared with Portugal's 90 and 80. The relative cost of cloth in units of wine is therefore lower in England (100/120) than in Portugal (90/80); the opposite holds for wine. Under the stipulated assumptions, specialization can improve the opportunities available to both. The example also identifies a condition often suppressed in simplified presentations: internationally mobile factors would change the location problem itself.
Extending comparative advantage from final goods to stages of production changes the relevant unit of analysis. A location may have a comparative advantage in packaging, wafer processing, a specific chemical transformation, or a design task without having a comparative advantage in the complete product. Intermediate trade permits such partial capabilities to become internationally useful. Specialization across activities remains subject to local skill availability, fixed costs, transportation, coordination, and the capacity of counterparties. Comparative advantage supplies a framework for opportunity cost, while actual sourcing patterns emerge from many interacting constraints.
Leontief's 1973 lecture on the structure of the world economy supplies the bridge from specialized activities to a formal system of production. Its opening distinction between intermediate transactions, final demand, and value added sets up the accounting framework used below. Together these traditions explain why production can become both more efficient and more mutually dependent as the division of labor deepens.
The economic object and boundaries
A global value chain is an organized set of productive activities contributing to a good or service in which production stages span national economic territories. A production network is the more general representation in which establishments, firms, or sector-country units can have multiple suppliers and buyers and both branching and recurrent linkages. A chain names an ordered sequence; a network describes the relationships among such sequences. Both can be represented at different scales.
A supply relationship may concern a material, energy, a service, access to an intellectual asset, or transport. Ownership is a separate relation. A foreign affiliate may purchase locally, a domestic firm may rely on foreign contractors, and a multinational enterprise may exercise control over assets situated in several territories. The location of production, nationality of the ultimate owner, origin of value added, destination of the shipment, and location of final demand answer different questions. Combining them under a single label such as national dependence produces unreliable inference.
For research purposes the unit of observation should follow the question. Sector-country tables are appropriate for allocating value added and broad resource requirements. Product-level customs records are closer to the geography of specific traded inputs. Firm-level data permit observation of actual buyers and suppliers. Plant-level records and bills of materials are essential when a failure involves a technical constraint or certification process. A relationship derived from a monetary flow records expenditure or sales, whereas a technological dependency may exist even if no recent transaction has appeared in the data.
The temporal boundary also matters. An input held in inventory is available despite a temporary cessation of deliveries. A supplier preparing a new facility may represent a feasible alternative only after construction and approval. Interdependence consequently concerns the state of the system at a stated time and the set of adjustments available over a stated horizon.
Trade in tasks and vertical specialization
Hummels, Ishii, and Yi (1999; published 2001) identify vertical specialization through the foreign intermediates used to produce exports. The mechanism differs from trade in completed goods because an intermediate input may enter successive shipments. When border costs apply repeatedly, small changes at each crossing can have appreciable effects on the attractiveness of fragmented production. A chain involving several customs territories needs reliable coordination as well as low costs.
Grossman and Rossi-Hansberg (2008) formalize trade in tasks and show that reduced offshoring costs can generate a productivity effect for factors performing the affected activities. The result is model-dependent and does not establish that particular workers benefit from every relocation. It suggests that the relevant economic margin includes changes in the efficiency of complementary domestic tasks, alongside conventional substitution and distributional effects.
For a firm, fragmentation is attractive when cost or capability gains exceed the additional costs of coordination. Suppose an integrated plant can perform a particular operation at a resource cost of 20 per unit. An external supplier offers the operation for 12; shipping, quality assurance, contracting, and expected delay add 5. A private cost gain of 3 makes fragmentation attractive at the anticipated scale. If the firm must also commit 300 in relationship-specific investment, it needs an expected volume exceeding 100 units to recover that outlay before financing and risk costs. These numbers are hypothetical. They demonstrate how variable and fixed costs interact and why supply relationships can persist even after modest changes in quoted prices.
A border policy applied to an intermediate may raise costs for domestic downstream exporters as well as foreign suppliers. The incidence depends on market structure, currencies, bargaining, substitution, and alternative sources. The economically meaningful comparison traces its incidence through production rather than assigning the entire tariff burden to whichever enterprise is legally responsible for payment.
Governance and the organization of value creation
Gereffi, Humphrey, and Sturgeon (2005) organize the governance of value chains around the complexity of transactions, the extent to which necessary information is codifiable, and supplier competence. Their five stylized forms range from arm's-length market arrangements to hierarchy, through modular, relational, and captive forms. These are distinct coordination arrangements rather than successive developmental stages. Several may coexist in the production of a single final good.
Where an input is standardized and the market thick, competing suppliers can be compared through price and verifiable specifications. Where a buyer and supplier jointly solve difficult engineering problems, relationship-specific knowledge may matter more than easily measured unit cost. A modular supplier performs a complex activity to defined interfaces; the ability to change suppliers then depends in part on how widely the interface and qualification process are shared. A captive supplier faces dependence on a dominant buyer and can incur sunk costs that limit its bargaining alternatives. Internal production allocates certain contractual problems within a single organization, while leaving the firm exposed to external inputs and public infrastructure.
This diversity helps explain why monetary dependency and switching difficulty diverge. A small order of a unique component may be more consequential to continuity than a large order of standardized materials. Contract duration, documentation ownership, alternative qualifications, and intellectual property rights determine feasible responses. An accounting map cannot recover these dimensions by itself.
Governance also influences the allocation of gains. Lead firms may earn returns associated with design, reputation, distribution, or control over access to markets; suppliers may accumulate expertise, undertake investment, and generate employment. The subsequent distribution of income depends on bargaining positions and institutions. Neither the location of final assembly nor a high export ratio alone establishes who receives the economic surplus.
Gross trade and value-added accounting
Consider three distinct economies. An upstream producer in A generates a component worth 30 entirely from domestic factors. It sells the component to B, which adds 40 through processing and exports the resulting intermediate worth 70 to C. C adds 30 and sells the final product worth 100 to its own consumers. Assume no further inputs, taxes, or margins. World gross exports for this sequence are 30 + 70 = 100; world final expenditure is also 100, but the equality is incidental. The value added of A appears in two export transactions, while the value created by C is absent from gross exports. Value added by origin is 30, 40, and 30.
If C instead exports its final output to a fourth economy, total gross exports rise to 200 even though the total value created remains 100. The distinction arises from repeated measurement of the same previously produced intermediate value as it crosses borders. Gross trade is a legitimate measure of transactions and associated border exposure; it is unsuitable as a direct measure of new production along a chain.
Koopman, Wang, and Wei (2014) develop an accounting framework for value added and double counting that distinguishes the origins of gross exports and additional terms created by repeated counting. Full national decompositions must account for value returning home, intermediate exports used abroad, foreign content, and repeated crossings. Our simple example deliberately omits feedback and reimports. Its purpose is to show why bilateral gross balances and bilateral value-added balances need not coincide.
Gross measures and value-added measures are complementary. A transport authority needs shipment volume and traffic patterns. A labor economist may seek domestic employment embodied in foreign final demand. A ministry concerned with critical supplies requires technical input availability. Each indicator should be interpreted according to the question it was designed to answer. Replacing all gross trade statistics with value-added figures would discard important transaction-level information.
Input-output formulation
Let there be n country-sector units. Let z(i,j) denote intermediate deliveries from supplier i to purchaser j, measured in a common valuation basis over a common accounting period. Let x(j) denote gross output and define technical coefficients a(i,j) = z(i,j)/x(j) for positive output. Let f(i) denote final demand for products of i. With A the matrix of these coefficients and x and f column vectors, accounting consistency requires
x = A x + f.
For a nonnegative A with spectral radius below one, the Leontief inverse exists, L = (I - A)⁻¹, and x = L f. Under that productivity condition the expansion L = I + A + A² + A³ + ... has a useful interpretation: the successive terms include direct final demand, first-round intermediate requirements, and increasingly indirect requirements. Every quantity is conditional on the valuation, classification, and fixed coefficients of the table.
If a value-added coefficient v(j) equals the value-added generated by unit j per unit of its gross output, the attributed value-added vector associated with f is diag(v)Lf. Under a closed global table with compatible valuations, no unallocated imported inputs, and complete accounting of primary inputs, column sums of intermediate coefficients plus value-added coefficients equal one. It then follows algebraically that the sum of attributed value added equals the sum of final demand. This identity does not require causal behavior, although applying the same coefficients to new demand involves a modeling assumption.
The OECD inter-country input-output tables provide a practical infrastructure for these calculations. Its 2025 release covers 1995–2022; the regular table identifies 80 economies plus a rest-of-world region, and the extended table offers additional firm-type distinctions within China and Mexico. Aggregation and balancing necessarily affect observed relationships. The data represent best estimates of monetary flows, not a comprehensive registry of technologically indispensable parts.
A fixed-coefficient demand scenario derives required output, conditional on the availability of that output. A binding constraint on a supplier requires a different closure. Without explicit rationing rules and substitution, mechanically multiplying a supply loss by L confuses a requirement matrix with an actual response of production. This distinction is foundational for systemic risk analysis.
A worked accounting and disruption example
Consider a stylized three-location chain producing one final good. Units are fixed-price monetary equivalents for a common accounting period; each technology uses fixed inputs. Supplier A provides 0.4 units of value per unit produced by B. B provides 0.5 per unit produced by C. A uses only primary inputs. The coefficient matrix by supplier rows and buyer columns is
| Supplier | A | B | C |
|---|---|---|---|
| A | 0 | 0.4 | 0 |
| B | 0 | 0 | 0.5 |
| C | 0 | 0 | 0 |
With final demand of 100 for C and zero elsewhere, A³ = 0 and L = I + A + A². Thus x(C) = 100, x(B) = 50, x(A) = 20. The primary value-added shares are 1, 0.6, and 0.5. The value added attributed to final demand is therefore 20 in A, 30 in B, and 50 in C. Total gross output is 170; total value added is 100.
Now suppose A's available productive output is capped at 12, with no inventory, spare suppliers, or redesign and no competing customer. Fixed proportional inputs imply that B can produce no more than 12/0.4 = 30, and C no more than 30/0.5 = 60. Assuming final demand remains 100 and the capacities of B and C are otherwise sufficient, actual output becomes (12, 30, 60). Gross output falls from 170 to 102, a decline of 68 units, while final deliveries and total value added fall from 100 to 60. The latter losses are each 40 units, not 68.
| Measure | Baseline | Constrained | Difference |
|---|---|---|---|
| A gross output | 20 | 12 | −8 |
| B gross output | 50 | 30 | −20 |
| C gross output | 100 | 60 | −40 |
| Total gross output | 170 | 102 | −68 |
| Total value added | 100 | 60 | −40 |
The physical rationing result follows from the additional capacity assumptions, not from the demand inverse alone. A second scenario clarifies the distinction: if final demand instead falls to 60 with A's capacity intact, exactly the same quantities arise in this fixed-coefficient example. Observed production totals therefore fail to identify the initiating shock. Inventory behavior, orders, delivery records, or other evidence is needed to distinguish the two causal explanations.
In a separate cost scenario, suppose A's primary input cost per unit rises by 0.1 while all input coefficients and other primary costs remain unchanged and cost prices fully adjust. B's unit cost rises by 0.4 × 0.1 = 0.04 and C's by 0.5 × 0.04 = 0.02. These are absolute unit cost changes in the constructed system. They are not predictions of inflation; margins, quantities, expectations, and alternative sourcing have been held fixed.
The same matrix supports three distinct exercises: accounting for demand, finding feasible output under a binding supply constraint, and tracing cost changes. Their equations and conclusions depend on the closure chosen. This example is constructed for explanation and does not estimate any actual national industry.
Network influence and aggregate fluctuations
In a network, indirect paths connect a purchaser to suppliers of its suppliers, and a firm may be linked to many other activities in both directions. A node at the center of the technical dependency structure may influence aggregate output far more than its share of direct transactions suggests. In their 2012 study of network origins of aggregate fluctuations, Acemoglu and coauthors establish that heterogeneous supplier influence and higher-order linkages are important for whether disaggregated productivity shocks translate into aggregate volatility under the specified model.
The distinction is mathematically consequential. For a first-order approximation in which aggregate log-output growth responds as a weighted sum of sectoral shocks, let the influence vector be w and the random shock vector be e. The variance of wᵀe is wᵀΣw, where Σ is the covariance matrix of the shocks. Under equal independent variances σ², variance becomes σ² times the sum of squared influences. If normalized weights are equal across n units, this sum is 1/n. If one unit retains a large influence weight, dispersion of the other weights does not eliminate the contribution of that unit.
With correlated disturbances the off-diagonal covariance terms also matter. Two suppliers in different contractual relationships may be exposed to the same energy system or transport route. Independence of counterparties is not independence of risks. The uncertainty in a network reflects both the structure of influence and the joint distribution of shocks.
This representation is a local analytical approximation. A network calculated from output shares or transactions does not automatically yield the weights of an arbitrary behavioral model. Domar-type weights, final-demand weights, or other influence measures have specific interpretations and can differ in normalization. The variance illustration uses normalized weights by construction; it cannot be interpreted as a universal formula for the level of GDP loss. Large shocks and changes in the network require more than a local linear approximation.
Nonlinearity, complementarity, and substitution
A production function with strong complementarity makes availability of a critical input a constraint on the use of other inputs. At the opposite extreme, easily substitutable inputs allow producers to adjust their mix after a price or availability shock. The appropriate assumption must be investigated rather than chosen because it leads to an attractive numerical result.
In the fixed-proportions case, output is limited by the smallest ratio of available inputs to their required quantities. Suppose a producer needs one specialized controller and two casings per final unit. It has 10 controllers and 100 casings; production is limited to 10 units, even if casings account for most recorded material expenditure. A replacement controller must meet the technical requirements within the relevant production horizon. A market with many sellers of incompatible controllers is not a market with many operational substitutes.
Baqaee and Farhi (2019), in their analysis beyond Hulten's theorem, emphasize that nonlinear macroeconomic effects depend on structural substitution, network relationships, scale properties, and the capacity to reallocate factors. Their results are derived within specific assumptions concerning efficient economies. They motivate attention to large changes and the conditions under which first-order approximations become inadequate.
Substitution is endogenous to time, investment, and organizational decisions. A plant may temporarily operate at lower utilization, devote engineering effort to redesign, and gradually qualify new suppliers. A nominally rigid input relationship can become more flexible after investment, while a technically feasible substitute may remain unavailable when competing firms simultaneously demand it. A credible analysis therefore specifies substitution possibilities in the relevant state of the world rather than treating a long-run elasticity as a guarantee of immediate replacement.
The asymmetric effects of losses and gains have a straightforward economic interpretation. A downward availability shock may make existing capital idle, whereas an equivalent increase in availability may be absorbed with little benefit when other inputs are binding. The existence and magnitude of this asymmetry depend on actual constraints and do not follow from connectivity alone.
The temporal structure of disruption
Stocks and flows are distinct. Let I(t) be usable inventory at the beginning of period t, D(t) deliveries actually received, and U(t) input use. With no spoilage, I(t+1) = I(t) + D(t) - U(t), and feasibility requires U(t) not to exceed I(t) + D(t). For a critical component consumed at ten units per day, inventory of 50 units can absorb five days without replenishment if production continues at its planned rate. A sixth day requires reduced output, new deliveries, or an alternate input.
This arithmetic also explains why a delay can generate a nonlinear cascade even when total shipments over a quarter change very little. The chronological distribution of deliveries affects feasible production. Aggregated monthly flows may conceal a critical week of interruption, and a subsequent inventory-restocking surge can obscure the earlier shortfall. For this reason, a causal account requires delivery timing and usable stock in addition to purchase values.
A firm can respond before the material constraint is reached. It may reduce output to extend inventory coverage, prioritize high-value or socially essential products, or pay for expedited transport. These decisions distribute the shortage over time and across customers. A downstream firm's expectation of scarcity can generate additional orders, even when its underlying final demand is unchanged. Upstream producers observing orders alone may interpret this as permanent demand and expand commitments beyond sustainable levels.
The speed of recovery depends on the persistence of the initiating disturbance, spare capacity, component redesign, lead times, logistics, financing, and coordination among firms. A one-time stoppage may leave a backlog that requires production above its previous rate to clear. If capacity after repair is merely sufficient for normal demand, the backlog remains until demand falls or production capacity increases. A resilience claim should specify the path of output and service restoration, not only whether operations eventually resume.
Identification of causal transmission
Evidence of simultaneous decline among trading partners establishes co-movement; identifying the effect transmitted through their commercial relationship requires a source of variation. A local earthquake, for example, can generate a plausibly external disturbance for distant firms that are connected to affected producers. Treating exposure as predetermined reduces the risk of classifying a consequence of the disaster as one of its causes.
The basic empirical comparison separates firms with pre-event exposure from otherwise comparable firms without it. An event-study specification may regress outcomes on firm and time effects, pre-event exposure interacted with event-relative periods, and carefully selected controls. The identifying assumption is that, in the absence of the disturbance, the exposed and comparison firms would have followed similar trends conditional on the design. The assumption is assessed through pre-event patterns, plausible confounders, placebo periods, and the mechanism's institutional details.
Geographical concentration creates a competing explanation: firms linked to affected suppliers may themselves share other disaster exposure or depend on the same customer markets. Controlling for location and industry may reduce this problem without eliminating all latent differences. A network event study also raises an interference issue because a comparison firm may be indirectly exposed through another link. Standard errors and the definition of the control group must reflect the assignment of shocks and correlated outcomes.
A credible analysis distinguishes the directly affected firm, direct suppliers and customers, indirect relationships, and general-equilibrium consequences. Estimated local effects cannot simply be summed when firms are related through overlapping paths. The choice of outcome matters as much as the research design: lost gross sales include intermediate values; measured value added, welfare, and asset prices are distinct outcomes.
The studies in the next chapter provide examples of research designs and reported findings; the dossier does not re-estimate their underlying microdata. Claims about signs and comparative mechanisms are bounded to the studies' reported settings. Broader policy inference requires triangulation across diverse events and production structures.
Evidence from earthquakes and supplier specificity
The 2011 Tohoku earthquake provides a concrete example of an external disruption with international consequences. Boehm, Flaaen, and Pandalai-Nayar's 2015 Census working paper reports firm-level evidence from Tohoku involving US affiliates of Japanese multinational firms that relied heavily on imported Japanese inputs. The reported output response in the following months was approximately proportional to the reduction in imports in the studied group. This supports the interpretation of strong short-run complementarity in that setting. It does not establish a common substitution elasticity across all countries or industries.
Carvalho, Nirei, and Saito's 2014 RIETI working paper investigates supply chain disruptions using transaction relationships and the geography of the same disaster. The summary distinguishes the results for firms linked to affected areas from the stronger evidence when the affected counterparties exited, and identifies indirect downstream effects. The subsequently published 2021 version with Tahbaz-Salehi extends this research; this dossier relies for detailed empirical descriptions on the accessible 2014 summary rather than conflating versions. The case is relevant because direct damage and network effects are geographically differentiated.
Using natural disasters as a source of firm-level disturbances, Barrot and Sauvagnat (2016) report that input specificity is associated with larger adverse consequences for customers of affected suppliers. The conclusion is consistent with the mechanism of costly substitution. It does not imply that all proprietary inputs produce large losses; inventory, financing, and technical alternatives intervene.
The three studies support a common line of explanation across samples: the loss of access to a particular productive capability can affect firms far removed from the initial event. Their designs differ in the source of variation, the definition of exposure, outcomes, and horizon. They accordingly contribute complementary evidence rather than three measurements of one common multiplier. The important scientific conclusion is bounded: actual commercial relationships transmit disturbances when production adjustment is limited, and the magnitude depends on input characteristics and network position.
Pandemic disturbances and demand rotation
A pandemic can simultaneously reduce labor availability, restrict transportation, change household demand, and alter the allocation of firms' investment. The observed commercial effects are therefore mixtures of supply and demand disturbances. The IMF's 2022 analysis of global trade and value chains describes the relatively rapid recovery of goods trade compared with services, while identifying demand shifts and the costs of containment policies. The episode demonstrates the importance of distinguishing the consequences of international connections from the causes of the original shock.
A household substituting goods purchased online for travel or in-person entertainment may increase demand for imported manufactured products even as local service output contracts. A producer that expands orders in response faces longer delivery times if carriers and component suppliers cannot increase capacity at the same speed. Higher volumes for a particular product coexist with shortages of particular inputs. Aggregate trade resilience therefore does not establish the absence of locally severe disruption.
Policy interventions can themselves form transmission channels. Health measures aimed at reducing physical contact affect labor supply and the operation of transport infrastructure. Income support changes spending capacity; liquidity support may prevent viable producers from closing because delivery or payment has been delayed. Their consequences depend on timing and institutional coverage, and the balance between protecting health and supporting production cannot be read directly from trade data.
The IMF overview identifies sourcing diversification and substitute availability as potential means of strengthening resilience. A well-designed comparison distinguishes those measures from complete localization. A foreign supplier in a differently exposed region may be a useful hedge against a domestic disturbance. A domestic substitute exposed to the same lockdown, electricity network, or shortages may add little resilience. The relevant relation is the joint feasibility of alternative supply, not the nationality printed on the invoice.
Transport geography and chokepoints
Production networks require movement through ports, inland corridors, shipping lanes, and other logistics systems. These infrastructures are shared by firms that otherwise have no direct commercial relationship. A failure at a chokepoint can therefore correlate disturbances across numerous production chains. Marine routes may be especially important for heavy intermediate goods, while air transport can be decisive for time-critical high-value components.
UNCTAD's Review of Maritime Transport 2024 reports a 2.4% increase in maritime trade volume in 2023 and highlights disruption around the Suez Canal, Red Sea, and Panama Canal. It puts the share of world trade volume carried by sea above four fifths. This volume statement should not be interpreted as a share of world trade value or of trade in digital services.
A rerouting decision alters both time and capacity utilization. If a vessel's fixed round-trip time rises from 40 to 50 days while its cargo capacity and operating fleet are unchanged, the number of rotations achievable over a long period falls to 40/50 = 0.8 of the previous rate. In this hypothetical calculation, equivalent throughput requires 25% more vessels (50/40), assuming ports and crews can also accommodate the change. The extra requirement is a property of the stipulated scenario and is not an estimate of actual global shipping capacity.
A transport shock places pressure on working capital as goods spend longer in transit. It also raises uncertainty about usable arrivals. Suppliers may change ordering policies, contract for alternative freight services, or increase local inventories. The short-run options depend on the availability of routes, terminal facilities, storage, and specialized handling. Transport interdependence is thus a constraint on the entire production network and not simply another proportional price attached to each traded product.
Sectoral architectures and critical inputs
The Global Value Chain Development Report 2023 devotes separate attention to energy supply, semiconductors, emissions, and inclusive development. Its structure reflects the heterogeneous economics of productive interdependence. A commodity with widely used specifications and substantial transport capacity may have a different substitution structure from a certified component designed for a particular application.
Energy flows depend on generation and extraction facilities, networks for moving or transforming energy, storage, demand flexibility, and the compatibility of consuming equipment. A fuel shortage can propagate into industry because energy is a pervasive input, and because some equipment cannot be rapidly converted to a different source. A rise in alternative fuel availability helps only to the extent that it can be delivered and used within the relevant time.
Semiconductor production illustrates a different set of dependencies. Design, fabrication, processing equipment, materials, and assembly/testing are complementary activities. Individual process technologies require specialized capital and qualifications. Adding final assembly capacity does not necessarily resolve a constraint in an earlier production stage. Conversely, an investment in upstream capacity may fail to improve system resilience if the necessary equipment, expertise, or downstream qualification remains unavailable.
Food production can combine biological cycles with storage and transport constraints. A missed planting or harvesting period may have consequences long after the immediate disruption; perishability makes inventory strategies product-specific. Health-critical supplies raise questions about continuity of service even when the commercial market is small.
These examples are comparative mechanisms rather than quantified rankings of vulnerability. A uniform concentration threshold or universal stockpile period ignores their different technologies. For each critical input, evidence should describe the specific bottleneck, viable alternatives, correlated dependencies, and time required to restore useful delivery.
Services, intangible assets, and digital dependence
A production system contains service activities even where observable shipments are tangible. Engineering, design, logistics management, insurance administration, information processing, and business services contribute to the capacity to produce and distribute goods. Some can be delivered across borders without a material shipment, while others require access to a particular site or physical infrastructure.
The global chain has an information layer. A specialized software system may coordinate inventories and quality documentation across facilities. A disruption to its availability, its identifiers, or its authorization mechanisms can impair production despite physically intact stock. Digital exchange also expands the geographic scope of activities that can be divided internationally, subject to connectivity, organizational competence, regulation, and trust.
The Global Value Chain Development Report 2025 describes a reconfiguration in which digitalization and services influence value creation, including production changes in the automotive and electric vehicle industries. This is consistent with viewing modern manufacturing as a combination of material and knowledge-intensive processes. The associated income can arise in a different location from the visible final assembly.
A dependency audit should therefore examine access to standards, software updates, maintenance, data pipelines, engineering records, and workforce expertise as well as physical supply. These assets differ in reproducibility and transferability. An open, documented interface may make switching technically feasible; a proprietary interface may require consent or redesign. Formal ownership does not by itself establish operational control over every necessary input. The international division of activities extends to the information and organizational capabilities that make production possible.
Participation, position, and contemporary measurement
Backward participation generally concerns foreign value added contained in a country's exports. Forward participation considers domestic value added incorporated into the exports of other economies. Particular published indicators may use different denominators and accounting definitions; comparisons require consistent measures. A country can exhibit high backward participation because it specializes in downstream processing, and high forward participation because its producers supply inputs used by foreign exporters. These are positions within a division of activities, not direct rankings of development or resilience.
The OECD's April 2026 value chain repositioning study provides experimental nowcasts extending measurement beyond the latest completed ICIO year. Across 41 economies, the export-weighted domestic value-added share was about 77% in 2022 and estimated at 77.6% in 2024. This corresponds to a rise of approximately 0.6 percentage points, rather than a 0.6% proportional increase. The estimate suggests limited aggregate movement in the measured period but cannot establish that the underlying firm-supplier geography was unchanged.
The same study notes substantial heterogeneity and caution about nowcast errors, particularly for small specialized economies. Inferences from projected sector tables should not be treated as confirmed final transaction data. In addition, an aggregate share can rise because export composition shifts toward activities that use more local inputs. No actual relocation of a given production facility is required for this compositional effect.
| Evidence | Reference period | Meaning |
|---|---|---|
| OECD ICIO 2025 release | 1995–2022 | Observed and harmonized statistical infrastructure; estimation and balancing involved |
| OECD experimental TiVA nowcast | 2023–2024 | Model-assisted indicator estimates, subject to revision |
| World Bank WDR 2020 | Report-era evidence | Historical measure of the scale of GVC trade |
| OECD resilience simulation | Published 2025 | Conditional counterfactual policy analysis |
These observations can inform a qualitative conclusion about the persistence of cross-border production. They cannot be pooled as if each measured the same variable. Dates, samples, denominators and the distinction between estimates and simulations remain attached to every quantitative claim.
Geoeconomic fragmentation and foreign investment
Investment decisions determine productive capacity available in future periods. When a firm relocates a specialized factory, the change involves fixed capital, labor skills, suppliers, legal arrangements, and often access to a local market. An investment announcement records an intention; operating capacity requires the successful completion of those steps. Trade patterns can respond before a mature facility exists through inventory and sourcing changes.
The IMF's 2023 chapter on geoeconomic fragmentation and foreign investment examines the reorientation of investment toward geopolitically aligned economies and the risks for developing economies dependent on capital from other blocs. Its output-loss conclusions are conditional on modeled fragmentation and the assumed responses. The underlying issues include loss of investment, technology diffusion, and access to productive capabilities.
Reshoring refers to returning an activity to the home economy; nearshoring typically refers to placing it closer geographically; friend-shoring refers to shifting sourcing or production toward politically aligned partners. These strategies may overlap or proceed separately. A foreign affiliate established in a neighboring country may use raw materials imported from a distant region. A domestically produced intermediate may depend on foreign capital equipment or imported energy.
It follows that changes in bilateral import shares are insufficient evidence of reduced indirect dependency. A change in the country shipping a product may reflect genuine new capacity, processing of imported components, corporate reorganization, or trade intermediation. Examination of origin, value-added content, capacity, and sourcing relationships is required before attributing the change to a particular relocation strategy.
The economic consequences of fragmentation depend on which capabilities are removed from feasible trade and investment relationships, the costs of replacing them, and the degree to which alternative systems develop. The distribution of effects differs among economies according to their initial productive structure and their access to competing sources of investment and demand.
Macroeconomic effects, prices, and adjustment
A network shock changes the allocation and utilization of productive resources. At the firm level it may reduce output or increase unit costs; at the macroeconomic level it can affect real GDP, employment, the price level, investment, trade balances, and the distribution of income. These responses operate through differing measures and should not be conflated.
A temporary increase in imported input costs can raise the domestic price of goods incorporating those inputs. The extent of pass-through depends on contract dates, invoicing currency, market structure, competitive pressure, and the elasticity of final demand. Firms may absorb part of a higher cost in margins or respond by producing fewer units. The price of an essential good can rise substantially even when its share of total output is small; aggregate inflation also depends on the composition of the consumption basket.
A negative supply disturbance differs from a fall in aggregate demand because it can constrain productive capacity while raising some prices. Monetary tightening may reduce total expenditure and the pressure on prices, while doing little to restore missing components. Fiscal transfers can sustain household purchasing power and financial continuity, yet a poorly targeted increase in nominal demand may intensify a rationing problem when the physical constraint persists.
Balance-of-payments effects are likewise ambiguous. A country may import fewer inputs because domestic output is constrained, while exporting less of its final products. Higher import prices can worsen its terms of trade even if physical volumes fall. The current account also reflects income flows and saving-investment decisions; the effect of one disrupted chain cannot determine the overall balance without these other reactions.
For large disturbances, the response includes reallocation of labor and capital among industries. Such reallocation takes time and may incur losses of specific skills and relationships. The relevant macroeconomic object is a joint adjustment of resource supply, production feasibility, demand, prices, and financial positions rather than an isolated trade multiplier.
Financial dependencies and continuity
Production requires financing over the interval between paying for inputs and receiving payment for output. A buyer ordering goods from an overseas supplier may pay an advance and carry inventory through production and transit. The delayed sale of its final good increases the period for which its funds remain committed. A disruption with modest eventual loss of physical output can therefore create a significant liquidity need.
A hypothetical firm spends 1,000 per day on operating costs financed until its customers pay, with a normal cash conversion interval of 20 days. Under the simplifying assumption of uniform spending and no supplier credit, its working-capital requirement is 20,000. If the interval extends to 35 days while daily spending continues unchanged, the requirement rises to 35,000. The additional 15,000 reflects timing and not necessarily a permanent reduction of economic value.
A firm without sufficient liquidity may curtail orders even when it expects its main customer to resume production. That action creates an upstream demand shock. Trade credit can redistribute liquidity between buyers and suppliers; bank lending can meet temporary funding needs, subject to collateral, credit quality, and funding conditions. If many firms are simultaneously exposed to a common disruption, creditors may tighten terms together.
Insurance can transfer specified financial losses under contractual terms. Coverage scope, waiting periods, deductibles, exclusions, and limits determine the actual payout. Compensation may support recovery but cannot itself recreate the missing productive capability. Financial resilience and technical resilience are related because the ability to pay for alternative supply, repairs, and inventory affects operational feasibility.
The accounting approach to production dependence should therefore be complemented by an analysis of balance-sheet constraints. A plant with spare physical capacity may still be unable to increase output if it cannot fund inputs or obtain payment guarantees. The availability of productive capital must be considered alongside the institutions that finance its use.
Distribution, development, and upgrading
Participation in a global production network can enable a developing economy to specialize in a narrow set of internationally marketable activities without first acquiring all capabilities needed to manufacture a complex final product. Entry can generate employment, demand for local services, managerial experience, and incentives for investment. The World Bank's development report on value chains presents such participation as a potential route to growth and poverty reduction, subject to complementary domestic and international policies.
The gains depend on the characteristics of the participating activity and how linkages evolve. A country that provides assembly may initially earn a small share of the final price and still obtain significant benefits if the activity creates productive employment and improves capability. A country supplying high-value intellectual or technical services may capture more income per worker while generating fewer jobs. Upgrading refers to changes in quality, productivity, sophistication, or the set of functions performed; a higher proportion of domestic value added is only one possible indicator.
Learning requires institutions and incentives that allow knowledge to accumulate and spread beyond an individual transaction. Local suppliers may develop capabilities if they can meet demanding standards and have access to credit, training, and reliable infrastructure. Where production involves close coordination, international partners can transfer useful knowledge, but contractual control over intellectual assets and supplier bargaining constraints may limit local appropriation.
Benefits accrue unevenly across occupations, regions, and ownership groups. Employment effects must be considered alongside wages, working conditions, skill requirements, changes in consumer prices, and the costs of adjustment for displaced workers. An export expansion that raises national income may coexist with losses in particular communities. Effective assessment therefore requires examining the distribution of both income gains and economic risks rather than treating a rise in gross exports as a complete measure of development.
Environmental interdependence and resource constraints
The environmental footprint of consumption can be located far from the consuming household. A product may embody electricity use at various stages, emissions generated by freight transport, and raw material extraction in a different jurisdiction. An input-output system can attribute these burdens to final demand by extending the accounting coefficients.
Let g(i) be a direct emissions intensity in physical units per unit of output, and use the previously defined consistent L and f. Then gᵀLf gives the direct and indirect production emissions attributed to final demand within the specified environmental boundary. The result counts emissions from each contributing activity under base-period intensities, provided the same direct emissions are not assigned twice. The calculation depends on the quality of sector-level emissions inventories and the price and technology assumptions embedded in the input-output table.
Production-based and consumption-based emissions accounts serve different analytical purposes. A territorial inventory is needed to understand emissions within a jurisdiction; a consumption footprint tracks the production requirements associated with its final demand. Neither indicates automatically how emissions would change if the activity relocated. That requires marginal energy supply, technology, transport, and the displaced use of resources to be modeled.
The Leontief world-economy formulation explicitly envisages environmental variables within a globally integrated input-output approach. The contemporary extension must also address physical limits and temporal differences: energy availability, water constraints, extraction capacity, and climate-related transport interruptions can bind production in ways not visible from monetary expenditures.
Resilience and environmental performance may coincide when efficient logistics or alternative clean energy reduces dependence on scarce resources. They may conflict when redundant production or longer routes require additional infrastructure and fuel. Evaluating such choices entails quantifying lifecycle burdens, the expected losses avoided, and the specific time and location of resource constraints.
Efficiency, resilience, and the policy frontier
Resilience is the ability of production and delivery arrangements to sustain or restore a specified level of service after disturbance. It involves preparation, absorption, adaptation, and recovery. Efficiency is evaluated against the resources used to achieve an economic objective. A resilience investment is economically justified when its expected benefits, including avoided interruption and relevant social costs, warrant its resource commitments under the chosen objective and risk criterion.
Consider a privately financed alternative supply arrangement that costs 8 per year in additional expense. If it reduces an annual probability of a loss of 100 from 0.15 to 0.05, the expected annual loss avoided under these simplifying assumptions is 10. The net expected gain is 2 before financing and other effects. The calculation says nothing about tail preferences, the possibility of catastrophic social costs, the reliability of the probability estimate, or how other firms might simultaneously use the alternative. It is a transparent decision illustration, not a policy recommendation derived from actual market data.
In the OECD supply chain resilience review of 2025, a modeled relocalization scenario reduces global trade by more than 18% and global real GDP by more than 5%, without consistently improving stability. These outcomes are conditional on the scenario's policy settings and the model. They are not observed losses, a forecast for a particular country, or a demonstration that targeted local capacity is always harmful. They establish that blanket localization can be costly and that domestic proximity alone does not guarantee stability.
Intervention may be justified where firms bear only part of the damage that a shortage imposes on others or where indispensable public services cannot pay a commercial scarcity price. Relevant instruments include shared infrastructure, supplier qualification, interoperability, strategic inventories, information disclosure, and international arrangements that preserve emergency access. Each instrument has fiscal and opportunity costs and can create new forms of dependence.
The appropriate comparison is among feasible interventions targeted at a defined vulnerability. A diversified supplier base, a stockpile, a local facility, and a redesign solve different operational problems. Policy assessment should compare their total costs, service continuity across plausible disturbances, implementation times, and distributional consequences.
A practical empirical research program
A systematic study of a selected value chain begins by defining the final service whose continuity matters and the period over which it is assessed. A broad monetary network can then be assembled from input-output tables, used to identify large direct and indirect economic exposures, and connected to product and firm records where legally available. The mapping should document missing trade sectors, unobserved suppliers, rest-of-world aggregation, and the treatment of reexports.
For a technically sensitive input, the researcher must establish product specifications, qualification rules, capacity, actual delivery times, and stock coverage. The identified bottlenecks can be stress-tested across independent and common disturbances. Rationing assumptions should distinguish proportional allocation, contractual priority, discretionary allocation, and market-clearing prices. Because alternatives consume resources, assigning spare capacity to multiple buyers simultaneously without an allocation rule overstates feasible resilience.
Empirical estimation should separately address likely causal mechanisms. A demand shock can be studied through externally driven changes in final markets; a supply event through predetermined exposure to an initiating disturbance; a policy change through a credible comparison of affected and unaffected pathways. The estimate must be tied to the population, event, time horizon, variables and model that produced it. Sensitivity analysis should vary both observed exposure and assumptions about adjustment.
Report results in commensurable units: changes in final deliveries, domestic value added, gross output, inventories, service continuity, costs and distribution. Distinguish calibration from direct observation. A table recording uncertainty and provenance should permit another researcher to establish which inputs drive the conclusion.
An empirical extension of this dossier would download the relevant ICIO vintage, reproduce selected indicators, access firm-level transaction data under appropriate confidentiality rules, and test alternative substitution and rationing assumptions. The present work supplies the methodological sequence and illustrative calculations, not that empirical extension.
Original analytic synthesis
The analysis establishes three related but nonidentical descriptions of the same production system. Value-added accounting locates generated income and production contributions. Network analysis describes paths through which requirements and disturbances can connect activities. An operational feasibility assessment determines whether the activities can actually continue under a defined change. These descriptions are established tools with different explanatory roles; their integration is the author's synthesis, rather than a new named model.
Their distinction resolves an apparent contradiction in international economic debate. An economy with high domestic value-added content can remain dependent on a small imported input with no immediate alternative; another economy with a high foreign content may draw that content from diverse qualified sources. The former may be more exposed to one specified supply disruption than the latter. Hence a high domestic share can coincide with low operational resilience, without implying either that domestic production is generally undesirable or that foreign sourcing is inherently secure.
A related distinction concerns national interests and the clearing of the whole network. A firm's optimization problem evaluates the private returns to its investments in continuity. The aggregate consequences of its failure may include losses at other firms and essential services beyond its contract. The privately attractive amount of buffer capacity can therefore differ from the socially beneficial amount, particularly where downstream costs are difficult to contract over.
A country seeking resilience should consequently identify the particular function that is exposed, check whether its alternatives are physically and contractually feasible, and assess the interdependent costs of any intervention. A national origin label, sectoral output share, or bilateral deficit is insufficient on its own. Market integration and safety margins can reinforce each other when geographically distinct capacities reduce common exposures. They can conflict when further specialization eliminates technically viable alternatives.
The appropriate scientific conclusion is conditional, not a universal rule favoring either integration or relocation. The economic question is how a network produces valuable output across states of the world and how its institutions allocate that output and its associated risks. Measuring value generation, mapping exposures, and testing the feasibility of adjustment provides a more complete account of systemic interdependence than any single aggregate statistic.
Limits of the inquiry and open scientific problems
The literature synthesis is bounded by the evidence directly available in this research session. Several journal articles were accessible through author-institution records or indexed abstracts rather than full text, and the study's use of those works is restricted to their reported mechanisms and metadata. The official statistical documentation and report pages support the specified dated figures. There is no independent econometric replication, no census of actual cross-border supplier links, and no estimated global loss distribution.
Formal calculations use simplified deterministic matrices and hypothetical probabilities. These examples are checkable illustrations of accounting, feasibility, and correlated risk under stated assumptions. In actual economies firms can alter technologies, substitute inputs, establish new relationships, draw down inventory, change product mix, or reallocate scarce goods. Feedback through demand, finance, expectations and policy can change the network itself. Forecasting such effects requires a specified behavioral model, sufficient data and empirical validation.
Measurement of intangible and digital dependencies remains particularly difficult because monetary transactions may not capture control of operational capabilities. Measurement of rare joint failures is also challenging because historical data contain few comparable extreme events. Climate, policy, technological change, and investment decisions can make previous probability estimates unstable.
Further research should distinguish structural from compositional changes in value-added shares, compare firm-level technical dependencies with sector-level statistical exposure, and estimate how adjustment capacity evolves over different horizons. A second line of inquiry concerns the contractual and institutional allocation of resilience costs across partners. A third concerns the recovery of critical public services when private financial incentives do not fully reflect the consequences of interruption. These questions follow from the analysis and identify directions for subsequent empirical work.
Conclusion
Global value chains organize productive specialization internationally. Their economic importance derives from the capacity to combine activities, resources, and knowledge located in different places. Their systemic significance derives from the fact that the performance of one activity can affect the feasible production of other activities through direct and indirect dependencies.
A complete economic interpretation requires at least three complementary operations: attributing the value created along the chain, tracing the relevant economic relationships, and examining the actual conditions for continued or restored production. Input-output accounting answers the first and part of the second. Firm- and product-level evidence, adjustment behavior, and operational constraints are necessary for the third.
Empirical evidence from external disturbances supports transmission through production relationships under conditions of limited short-run adjustment. Comparative episodes involving transport, pandemic disruption, and investment fragmentation show why the origin, duration, and correlation of shocks matter. International integration can distribute productive opportunities and spread disturbances; it can also diversify supply and provide alternatives. The outcome depends on network topology, specific inputs, institutions, stocks, and available replacement capacity.
The policy problem is therefore the targeted management of dependencies and the costs of adjustment. Economic resilience demands attention to what can fail together, what can be replaced, how rapidly output can recover, and who bears the cost. Domestic value creation and international production integration remain meaningful objectives; their relationship to resilience must be evaluated through evidence about the actual productive system.
The principal conclusion of this dossier is that international economic interdependence is a property of organized production, not merely a measure of the quantity of international exchange. The scientific analysis of that interdependence establishes the basis for evaluating both the benefits of specialization and the systemic consequences of disruption.
References
Acemoglu, Daron, Vasco M. Carvalho, Asuman Ozdaglar, and Alireza Tahbaz-Salehi. 2012. The Network Origins of Aggregate Fluctuations. Econometrica 80 (5): 1977–2016. DOI: 10.3982/ECTA9623. Source.
Asian Development Bank, IDE-JETRO, University of International Business and Economics, and World Trade Organization. 2023. Global Value Chain Development Report 2023: Resilient and Sustainable GVCs in Turbulent Times. Geneva: WTO and publishing partners. Source.
Baqaee, David Rezza, and Emmanuel Farhi. 2019. The Macroeconomic Impact of Microeconomic Shocks: Beyond Hulten’s Theorem. Econometrica 87 (4): 1155–1203. DOI: 10.3982/ECTA15202. Source.
Barrot, Jean-Noël, and Julien Sauvagnat. 2016. Input Specificity and the Propagation of Idiosyncratic Shocks in Production Networks. Quarterly Journal of Economics 131 (3): 1543–1592. DOI: 10.1093/qje/qjw018. Source.
Boehm, Christoph E., Aaron Flaaen, and Nitya Pandalai-Nayar. 2015. Input Linkages and the Transmission of Shocks: Firm-Level Evidence from the 2011 Tohoku Earthquake. CES Working Paper 15-28. Washington, DC: US Census Bureau. Source.
Carvalho, Vasco M., Makoto Nirei, and Yukiko U. Saito. 2014. Supply Chain Disruptions: Evidence from the Great East Japan Earthquake. RIETI Discussion Paper 14-E-035. Revised published version with Alireza Tahbaz-Salehi: Quarterly Journal of Economics 136 (2), 2021: 1255–1321. DOI: 10.1093/qje/qjaa044. Source.
Gereffi, Gary, John Humphrey, and Timothy Sturgeon. 2005. The Governance of Global Value Chains. Review of International Political Economy 12 (1): 78–104. DOI: 10.1080/09692290500049805. Source.
Grossman, Gene M., and Esteban Rossi-Hansberg. 2008. Trading Tasks: A Simple Theory of Offshoring. American Economic Review 98 (5): 1978–1997. DOI: 10.1257/aer.98.5.1978. Source.
Hummels, David, Jun Ishii, and Kei-Mu Yi. 1999. The Nature and Growth of Vertical Specialization in World Trade. Staff Report 72. Federal Reserve Bank of New York. Published version: Journal of International Economics 54 (1), 2001: 75–96. Source.
International Monetary Fund. 2022. World Economic Outlook, April 2022: War Sets Back the Global Recovery. Chapter 4, “Global Trade and Value Chains in the Pandemic.” Washington, DC: IMF. Source.
International Monetary Fund. 2023. World Economic Outlook, April 2023: A Rocky Recovery. Chapter 4, “Geoeconomic Fragmentation and Foreign Direct Investment.” Washington, DC: IMF. Source.
Koopman, Robert, Zhi Wang, and Shang-Jin Wei. 2014. Tracing Value-Added and Double Counting in Gross Exports. American Economic Review 104 (2): 459–494. DOI: 10.1257/aer.104.2.459. Source.
Leontief, Wassily. 1973. Structure of the World Economy: Outline of a Simple Input-Output Formulation. Nobel Memorial Lecture, December 11. Stockholm: Nobel Foundation. Source.
OECD. 2025. Inter-Country Input-Output Tables. Regular and extended editions. Paris: OECD, statistical database and documentation. Source.
OECD. 2025. OECD Supply Chain Resilience Review: Navigating Risks. Paris: OECD Publishing. DOI: 10.1787/94e3a8ea-en. Source.
OECD. 2026. Global Value Chain Repositioning: Insights from the 2023–24 TiVA Nowcasting Exercise. OECD Statistics Briefs 2026/01. Paris: OECD Publishing. DOI: 10.1787/8c97068d-en. Source.
Ricardo, David. 1817 (1821, 3rd ed.). On the Principles of Political Economy and Taxation. Chapter 7, “On Foreign Trade.” London: John Murray; Library of Economics and Liberty online edition, 1999. Source.
Smith, Adam. 1776. An Inquiry into the Nature and Causes of the Wealth of Nations. Book I, chapter I, “Of the Division of Labor.” Edwin Cannan, ed., 1904. London: Methuen; Library of Economics and Liberty online edition. Source.
UN Trade and Development (UNCTAD). 2024. Review of Maritime Transport 2024: Navigating Maritime Chokepoints. Geneva: United Nations. Source.
World Bank. 2020. World Development Report 2020: Trading for Development in the Age of Global Value Chains. Washington, DC: World Bank. Source.
World Trade Organization and publishing partners (UIBE, ADB, IDE-JETRO, World Economic Forum). 2025. Global Value Chain Development Report 2025: Rewiring GVCs in a Changing Global Economy. Geneva: WTO. DOI: 10.30875/9789287085641. ISBN 9789287085641. Source.