Ports' criticality in international trade and global supply-chains

Maritime transport underpins international trade and the global economy. Ports link production centers and consumer markets, creating strong dependencies and feedbacks between economic structure (e.g., trade composition, supply-chain organization) and freight flows through specific ports. Disruptions or changes in the maritime network (e.g., port closures, new infrastructure) propagate through supply chains across countries and industries. Recent events such as COVID-19 highlighted simultaneous shocks to demand and supply, causing bottlenecks at many ports; the 2021 Suez Canal blockage showed how a shock to a key route can affect multiple ports and dependent supply chains. Historical events (e.g., Hurricane Katrina) also illustrate how port disruptions ripple through global commodity markets and supply chains. Traditional notions of port criticality—based on absolute throughput, network centrality, or local economic contribution—do not fully capture ports’ primary function as cross-border connectors of supply chains. This study aims to provide a comprehensive view of ports’ systemic importance by explicitly linking global maritime transport flows to sectoral supply-chain dependencies. The authors present a new modeling framework that integrates a global modal split model, a detailed maritime and hinterland transport flow allocation model (OxMarTrans), and a multi-regional input–output (MRIO) database to evaluate how specific ports connect and sustain domestic and global supply chains, and to quantify their criticality under a consistent, sector-resolved approach.

The study builds on two major strands: (1) macroeconomic analyses of international trade and supply-chain interconnectivity enabled by MRIO tables, which capture inter- and intra-industry dependencies across countries but lack explicit transport mode and network detail; and (2) complexity-science analyses of maritime transport networks, which characterize shipping connections and network structure but do not integrate information on commodity types, origins/destinations, or economic use. This mismatch between economic structure data (trade and supply chains) and observed transport networks motivates the integrated framework here. Prior work has examined container and vessel network centrality and canal roles, but without sector-specific flows and supply-chain linkages, potentially misprioritizing port criticality. The authors’ framework addresses this gap by linking sectoral bilateral trade, predicted modal shares, observed maritime network capacities/routes, and MRIO-based economic dependencies at the port level.

Overview: The framework combines (i) a global modal split model that estimates the maritime share for each bilateral trade flow at the commodity level (HS6), (ii) the Oxford Maritime Transport (OxMarTrans) model that allocates maritime and hinterland routes between specific ports based on observed AIS-derived network capacities and multimodal hinterland connectivity, and (iii) an extension of the EORA MRIO (2015) to the port level to quantify supply-chain dependencies. The analysis covers 1378 ports (transport layer) and 1298 ports across 176 MRIO countries (supply-chain layer) using an 11-sector aggregation aligned with EORA. Modal split model: A discrete choice model predicts mode shares using mode-specific variables (cost, time), OD-specific factors (income, adjacency), and commodity characteristics (quantity, value-to-weight, perishability). It is calibrated with UN Comtrade modal data and applied to BACI trade flows (~8 million HS6-level bilateral records), then aggregated to 11 sectors consistent with EORA. Maritime transport model (OxMarTrans): This global flow allocation model assigns predicted maritime trade between ~3400 subnational regions across 207 countries onto an observed maritime network (from AIS) and multimodal hinterland network, incorporating sector-specific constraints and revealed route preferences (including transshipment) between ~1380 ports. Outputs include port-level import/export/transshipment flows, route usage (e.g., Suez and Panama), and dependencies arising from land-based connections and transshipments. Link to input–output tables and metrics: Port-level trade flows are mapped into the EORA MRIO (2015; 190 economies). Because MRIO and BACI differ, only overlapping flows are modified. Two port-centric metrics are computed: (1) Port-Level Output Coefficient (PLOC) using a variation of the Hypothetical Extraction Method. Trade flows through a port are removed from the MRIO to quantify output changes via demand-driven (Leontief) backward linkages and supply-driven (Ghosh) forward linkages, separating domestic vs. foreign effects. PLOCA measures absolute output and consumption dependent on a port; PLOCR is PLOCA normalized by the port’s throughput. (2) Port-Level Import Coefficient (PLIC) quantifies marginal increases in port-level imports required per 1000 USD increase in final demand (domestic consumption and exports) across the economies a port serves, derived from modified import requirements matrices consistent with MRIO methods (following Hummels et al. and Dietzenbacher). Aggregating PLIC across ports serving a country yields the Country-Level Import Coefficient (CLIC). Base year is 2015.

- Maritime share of global trade: About 9.4 billion tonnes (USD 7.6 trillion) were maritime in 2015; ~75% by weight and ~50% by value. Mining and Quarrying (sector 3) is most maritime-intensive (75.7% by value; 86.0% by weight). Many manufacturing sectors move only 40–57% of trade value by sea. - Country disparities: Small Island Developing States (SIDS) import 86.5% and export 79.8% by sea—about twice non-SIDS. Low-income countries import 1.5 times more by sea than high-income countries (68% vs 45%). Geography, alternative modes, value-to-weight ratios, and logistics quality shape modal shares. - Global freight magnitude: OxMarTrans estimates 90.5 trillion tonne-km at sea and 33.4 trillion tonne-km over land (to connect hinterlands). Mining and Quarrying accounts for 43% of maritime tonne-km; sectors 9–11 together account for ~2.7%. - Canal dependencies: ~USD 1.1 trillion (13.9% of maritime trade) passes Suez and ~USD 0.49 trillion (6.2%) passes Panama (2015), aligning with official statistics. - Cross-border dependencies: 16.4% of global port throughput (value) is transshipped; 19.4% involves foreign imports/exports via hinterland connections. Key dependencies include Djibouti for Ethiopia; Dar es Salaam and Beira for Sub-Saharan landlocked states; Lomé/Cotonou for West Africa; Arica/Ilo for Bolivia; regional hubs in Oceania and the Caribbean. - Trade concentration across ports: Only 4 (imports) and 3 (exports) ports handle 10% of global maritime trade; 56 (imports) and 48 (exports) handle 50%; 378 (imports) and 366 (exports) handle 90%, indicating a small core and many peripheral ports. Sectoral unevenness is highest for exports of Textiles & Wearing Apparel (sector 5), Transport Equipment (sector 10), and Other Manufacturing (sector 11); more even for imports of Agriculture (sector 1), Food & Beverages (sector 4), and Petroleum/Chemical/Non-metallic products (sector 7). - Port-Level Output Coefficient (PLOC): On average, every 1 USD of throughput through a port influences 4.34 USD (5th–95th percentile: 3.84–5.03) of global value (PLOCR). The top 5 globally macro-critical ports—Singapore, Shanghai, Busan, Rotterdam, Antwerp—each handle goods that contribute >1.4% of global industry output. In total, 94 ports are macro-critical (>0.1% of global output depends on them). Forty ports are domestically critical (>10% of domestic output depends on a single port), often in small island economies (e.g., Port Louis 26.9%, Pointe-à-Pierre 24.9%, Reykjavik 23.0%, Sitra 25.3%). - Supply-chain positioning: Ports differ in domestic vs foreign and forward vs backward linkages. Rotterdam/Singapore/Algeciras have large foreign dependencies; Shanghai/Bremerhaven show higher domestic dependencies and backward linkages; Los Angeles–Long Beach has strong backward linkages (end-of-chain imports); Ulsan exhibits strong forward linkages (exports of domestically produced goods); Itaqui and Mina Al Ahmadi mainly show forward linkages (raw materials, oil). - Port-Level Import Coefficient (PLIC) and country import needs: Top-15 ports have PLIC >170 (up to 486). Twenty-seven ports have PLIC >100, including small island ports (e.g., Maldives, Aruba, Mauritius, French Polynesia) and regional hubs (Singapore, Kingston, Marsaxlokk, Freeport). Some ports (Djibouti, Berbera, Cotonou, Maputo; Bar, Burgas) are more sensitive to foreign than domestic demand. At the country level (CLIC), a 1000 USD increase in final demand results in a median 84.6 USD increase in maritime imports (maximum 501.5 USD). SIDS have 1.5× higher CLIC than non-SIDS. Low-income countries have lower CLIC (less integrated/diverse supply chains), while manufacturing-heavy sectors require more maritime imports (sectors 9–11: 112–153 USD per 1000 USD) compared to Agriculture and Mining (about 40 USD per 1000 USD). - Demand–import feedback: Across economies, every 1000 USD increase in final demand yields large increases in maritime imports through certain ports; 30 individual ports experience >100 USD increases, highlighting sensitivity of specific nodes. - Validation and magnitudes align with UNCTAD estimates for tonnes and tonne-km, supporting external consistency.

The study addresses the core question of how critical individual ports are to domestic and global economies by explicitly linking sector-resolved trade flows to observed maritime networks and MRIO-based supply-chain dependencies. By quantifying modal shares, route allocations (including transshipments/canals), and port-level economic linkages, the framework reveals that maritime trade is both substantial and highly concentrated in a relatively small core of ports. The analysis shows disproportionate reliance of SIDS and low-income countries on maritime transport, and maps extensive cross-border infrastructure dependencies via land links and transshipment hubs. PLOC metrics demonstrate that throughput at certain ports supports significant shares of global and domestic output, while PLIC/CLIC quantify how growth in final demand translates into increased maritime imports through specific ports and countries. These findings reframe port criticality beyond throughput or network centrality, situating ports within their supply-chain roles (forward/backward, domestic/foreign) and underscoring potential propagation of shocks through international production networks. The results are pertinent for resilience planning, infrastructure investment prioritization, trade facilitation policies, and anticipation of spillovers from economic growth or network changes (e.g., new corridors) that alter port demand.

This work introduces a comprehensive, port-resolved framework integrating trade, transport networks, and supply-chain economics to quantify ports’ criticality. Key contributions include: (i) consistent estimation of maritime modal shares by sector and country, (ii) detailed allocation of flows on observed maritime and hinterland networks with sectoral specialization, and (iii) port-level economic dependency metrics (PLOC, PLIC/CLIC) revealing both global macro-critical ports and domestically critical ports, especially in small island and trade-dependent economies. The framework highlights concentrated dependencies, cross-border linkages, and the feedback between economic growth and maritime imports. Future research directions enabled by this framework include: allocating trade-linked maritime emissions to countries/sectors; testing policy scenarios (infrastructure investments, border facilitation, carbon pricing) and their impact on flow allocation; quantifying future port capacity needs under changing trade and demand; and coupling with disaster impact models to estimate domestic and global losses from port/network disruptions and climate change. Overall, sustainable and resilient port infrastructure planning requires a system-wide view of transport–economy interdependencies illuminated by this approach.

- Spatial and sectoral aggregation: Economic data are at national level and mapped to ports; lack of subnational economic resolution may bias results where hinterland–port relations vary within countries. Sector resolution is limited to 11 sectors (EORA), potentially masking finer commodity heterogeneity. - Data coverage and harmonization: MRIO covers 176 of 207 transport-network countries; analysis is restricted to overlapping flows between MRIO and BACI, and discrepancies between sources may affect precision. Base year is 2015, which may limit applicability to periods with structural changes post-2015. - Methodological assumptions: Mode of transport is defined by dominant (longest-distance) leg, which may misclassify multimodal nuances. The PLOC relies on input–output linearity and hypothetical extraction (Leontief/Ghosh models), which abstract from price adjustments, capacity constraints, and behavioral responses. Network capacities and route choices are inferred from AIS-based observations and sectoral constraints, which, while detailed, may not capture all operator strategies or temporal fluctuations.