RESEARCH DOSSIER
Marine War Risk Navigation Zones and Accumulation
Abstract. Marine war-risk insurance converts geopolitical danger into contractually specified exposures attached to vessels, cargo, liability interests, voyages, and time. Its geographic architecture is frequently misunderstood. A listed war-risk area is a market signal and a potentially important contractual input; it is neither an internationally binding navigational prohibition nor a universally applicable coverage exclusion. A vessel may cross the same geographical boundary under several contracts with different attachment conditions, limits, notification duties, and claims consequences. At portfolio level, war-risk accumulation arises when many insured interests encounter a common hostile event, a shared transit bottleneck, a threatened port, or a prolonged campaign of correlated incidents. This monograph examines those processes through original legal texts, clause wordings, official shipping-security records, sector statistics, and the quantitative literature on dependence. It analyzes Joint War Committee geography and the September 2026 JWLA-035 revision, International Maritime Organization navigation obligations and incident reporting, Institute War Clauses (Cargo) duration provisions, English marine insurance law, P&I reinsurance arrangements, and the documented Red Sea and Black Sea disruptions. An illustrative, fully synthetic portfolio demonstrates that geographic concentration, temporal overlap, dependent severity, contractual recoverability, and reinsurance definitions jointly determine gross and net loss. Its principal analytical result is that a geographical zone is an observation layer, while insured accumulation is a time-indexed network of contractual interests and common-cause loss mechanisms. The monograph sets out auditable data requirements, valuation sensitivities, source limitations, and research priorities without proposing a new named doctrine or claiming unverified empirical novelty.
Keywords: marine war risk; Joint War Committee; listed areas; navigation restrictions; Institute War Clauses; hull insurance; marine cargo; protection and indemnity; Red Sea; Black Sea; route concentration; accumulation risk; common-cause loss; reinsurance; treaty aggregation; geospatial exposure; maritime law; dependence; catastrophe modeling; insurance evidence.
Introduction
Navigation connects the physical movement of ships with a sequence of insurance decisions. A ship can retain a known insured value while its hazard environment changes rapidly at the entrance to a strait, near a conflict-affected port, or along an alternative route. Cargo can change vessels, be discharged temporarily, or enter a terminal accumulation. A P&I club may insure particular liabilities under rules separate from the hull policy. The reinsurer may see an aggregation of claims defined by contract language rather than by the maritime geography alone. Consequently, the scientifically relevant object is the joint history of position, interest, policy, peril, and loss.
The international framework already separates these dimensions. The Marine Insurance Act 1906, section 3 identifies maritime adventures and maritime perils, including war and captures, while the IMO ships' routeing framework governs maritime navigation safety through a different institutional process. The Joint War Committee's listed-area process serves participating marine hull war underwriters, using risk assessment and notification rather than maritime legislative authority. These categories interact during a voyage without becoming interchangeable.
The research questions are: how are marine war-risk geographical designations made, revised, and incorporated into contracts; how do route choices alter both individual hazard exposure and cross-policy dependence; how can gross and net accumulations be measured without conflating hull, cargo, P&I, and shipping delay; and which propositions survive comparison with the available historical and contemporary evidence? The study focuses on commercial maritime insurance. Aviation appears only where clause interfaces reveal a meaningful difference. The analysis encompasses international operating conditions, with English law used as an explicitly identified contractual reference system rather than as a worldwide legal default.
The documentary cutoff is November 17, 2026. Official observations are distinguished from market commentaries, from examples made for this study, and from contractual interpretations requiring the individual policy schedule and endorsement. The selected source set includes enacted statutes, an international treaty text, official IMO and UN documents, industry-issued or identifiable clause material, market association circulars, global trade and insurance statistics, and an original academic publisher's methodological bibliography. The factual sources establish observed events and institutional practices; all numerical portfolio experiments below are hypothetical and demonstrate arithmetic relationships only.
Conceptual and Historical Foundations
Marine adventure, peril, interest, and insured loss
Marine insurance describes the allocation of financial consequences arising from a marine adventure. The law's description of insurable property, earnings, and third-party liability in section 3 of the 1906 Act supplies three distinct objects. A hull policy may address damage to the vessel and specifically agreed associated interests; cargo insurance addresses goods belonging to parties with insurable interests; and P&I responds to selected third-party liabilities, subject to club rules and special cover. These interests can coexist on the same ship without having the same insured, limit, deductible, attachment time, or reinsurer.
An incident does not determine its own coverage outcome. Marine insurance law connects covered loss to its proximate cause: section 55 of the 1906 Act states the default connection between a peril insured against and loss, subject to the policy. A missile strike may create physical hull damage and cargo loss; pollution, wreck removal, and crew liabilities can raise other policy questions. A delayed arrival after a successful diversion produces a commercially measurable cost, but delay is not automatically an insured physical-damage claim. Section 55 also contains a default exclusion concerning loss proximately caused by delay unless the policy provides otherwise. This is especially important when a public report uses the language of economic disruption, while the insurer's experience is presented as physical claims.
The operative unit of analysis must therefore be explicit. A vessel is identified by a stable IMO number, a voyage by a dated itinerary, cargo by shipment or bill-of-lading interest, a policy by insurer and wording, an event by causal boundaries, and a claim by the insured's recoverable amount. Vessel identity is a valuable joining key, yet it cannot identify every cargo owner or all coverage layers. Gross insured value, economic value at risk, probable damage, paid claims, outstanding reserves, and reinsurer recoverables are separate quantities.
War, strikes, terrorism, piracy, and navigational danger
War-risk wordings allocate specified consequences of organized violence, capture, and related hazards; they may be paired with separate strikes, terrorism, or piracy wordings. The widely used Institute War Clauses (Cargo), CL385, dated January 1, 2009, identify war, civil war, rebellion and certain hostile acts, specified capture and detention arising from those perils, and derelict weapons of war. Each category has its own causal and temporal implications. A floating mine may threaten a ship that is geographically distant from the site where the mine was laid, while a detention can occur without explosive damage. These possibilities make fixed distance to a battlefield a poor sole predictor of insured loss.
The market has long distinguished war-related underwriting from ordinary maritime navigational safety. IMO's explanation of ships' routeing systems traces the practice to late nineteenth-century North Atlantic routing and places modern routeing under SOLAS chapter V and associated international navigation instruments. Market area notices are another kind of map: they classify potential underwriting concern. Their coexistence is rational because a traffic separation scheme answers where ships should navigate for safety, while a war-risk listed area warns the contracting parties where a premium, notification, or permission decision may be required. The historical continuity is the use of geography as a coordinating device; the legal authority and consequence differ.
There is a further distinction between contemporary strikes or missiles and derelict weapons left in the sea. In its Black Sea mine circular of May 2022, IMO directed stakeholders to security reporting and specifically addressed floating mines as a threat beyond an immediate engagement area. The mechanism creates spatial persistence: the insured exposure can continue after a particular combat incident and move with currents, monitoring limitations, and route choices. For analysis, the origin of a weapon, its present position, the insured vessel's track, and the applicable war wording must be recorded separately.
Insurance markets and observable scale
The magnitude of the marine sector supplies perspective but is not itself a measure of insured war exposure. According to the IUMI 2026 statistics release, global marine insurance premium income reached USD 42.6 billion in 2025; cargo accounted for USD 24.2 billion. The same release discusses war cover's growing prominence amid geopolitical stress. Premiums represent aggregated insurance income across several marine lines, regions, and perils, and cannot be converted directly into sums insured at a threatened strait. War risk may be separately priced, selectively bought back, excluded, or covered by particular facilities. Inferring maximum loss from total sector premium would equate annual revenue with simultaneous values at risk.
The statistical lesson is one of denominator discipline. Reported hull premium divided by the global fleet is not an unbiased probability of a missile casualty; cargo premium divided by trade value is not a war-event severity measure. The observations concern different populations, valuation systems, and coverage designs. The IUMI 2026 Facts and Figures further identify distinct cargo and hull market compositions, demonstrating why a multi-line accumulation inventory must preserve line-specific information before cross-line integration.
Causation across connected interests
A common initiating act can produce heterogeneous consequences. Suppose a single projectile disables a container vessel near a terminal. Hull damage may be adjusted by repair or total-loss valuation, some cargo may be physically destroyed, some remains sound but delayed, and third parties might allege pollution or property damage. Salvor expenditure and general-average contribution may arise. Marine Insurance Act section 66 establishes the legal concept of general-average sacrifice and expenditure and makes insurance recovery dependent on policy terms and the insured peril. These categories cannot be consolidated into the vessel's insured hull value and added again when reinsurance is calculated: the same underlying expense could be shared or recovered through different interests.
It follows that casualty reconstruction should begin with a causal ledger. Each event record should identify the initiating peril, physical sequence, time and location, affected interests, expense categories, contractual adjustments, and recoveries. This avoids the common analytical error of treating a ship casualty, a cargo declaration, a P&I liability reserve, and a salvage claim as four independent war events merely because they appear in four datasets.
Zone Designation
Institutional provenance of a listed area
The London market's Joint War Committee comprises representatives of Lloyd's and International Underwriting Association hull-war business. Its committee statement and explanations describe independent security advice, the issuance of Listed Areas, and the expectation that ships entering designated regions notify underwriters. It expressly distinguishes committee listing from negotiated rating decisions. An individual insurer and insured or broker determine the contractual consequences. This institutional structure makes the listing an evidentiary indicator of heightened perceived exposure rather than a universal price schedule or an automatic policy exclusion.
The association's Listed Areas page identifies the current listing as JWLA-035 and records a September 2026 review. The IUA circular index associates JWLA-035 with the Black Sea update dated September 17, 2026 and lists preceding circulars. The circular itself is dated September 16. The difference between the date printed on an issued document and the date displayed in an association index illustrates a manageable but consequential provenance distinction: a historical reconstruction should retain both document date and distribution/index date rather than silently select one.
The JWLA-035 circular text defines the Black Sea area excluding territorial waters of adjoining countries other than Russia and Ukraine. It also defines other waters and states that application to individual contracts remains a matter for specific negotiation. This hosted PDF reproduces an LMA/IUA circular but is a third-party-hosted copy; the issuance and currency of the circular were independently cross-checked against the official committee and IUA index. The distinction is recorded in the evidence ledger. Its geographical definition should be stored with version and text, not replaced by an imprecise annotation such as 'Black Sea dangerous.'
The legal status of a polygon
Four geographical layers should be distinguished in a georeferenced maritime record. First, a coastal state's territorial sea, exclusive economic zone, internal waters, and straits are legal categories under applicable public international law. The United Nations Convention on the Law of the Sea governs, among other matters, territorial-sea navigation and innocent passage, subject to its provisions and applicable state obligations. Second, IMO routeing measures and national maritime notices influence navigation and safety. Third, a market-listed area describes a perceived underwriting risk. Fourth, the binding contract may contain a schedule, navigation warranty, excluded region, additional-premium mechanism, or specific endorsement with its own definitions.
The boundaries may coincide spatially without coinciding legally. A vessel crossing a line on a commercial war-risk chart does not thereby lose its rights under UNCLOS; conversely, a right of passage does not assure the availability of a particular insurance policy. No universal inference can be drawn from political recognition or government warnings to a private cover attachment decision. Jurisdiction, contractual wording, and actual route must be traced together.
A geospatial analyst must retain precision at the edge. The JWLA-035 text uses exceptions for particular territorial waters, a coordinate-defined part of the southern Red Sea and adjoining waters, and special definitions for named countries and ports. A boundary crossing within uncertainty of the vessel position estimate can be misclassified if the analyst overlays a simplified coastline polygon. A vessel's arrival inside a port authority area may have different contractual relevance from passage near the coast. Methodologically, records should preserve coordinate reference system, geometry version, issue timestamp, intended versus actual route, intersection tolerance, and the source responsible for each feature.
Notice, permission, additional premium, and contractual attachment
For a listed-area voyage, underwriters may require prompt declaration before entry and may quote an additional war premium, adjust limits, or agree conditions. The JWC's market explanation makes clear that the committee does not determine individual rating; consequently an academic study should not label a listed area as a fixed tariff zone. One shipowner's contractual permission or navigation condition cannot be inferred from another shipowner's premium. Nor should a later list automatically be applied retrospectively to a loss predating its effective contractual adoption.
English law provides useful examples of why the wording matters. The Insurance Act 2015, section 10 modifies the effect of breach of warranty in relevant insurance contracts, while section 11 on loss-relevant terms addresses certain terms that tend to reduce particular loss risks. The Act preserves distinctions between contract terms defining the risk as a whole and those operating as warranties or risk-mitigation conditions. It is therefore analytically unsound to announce that every failure to report an entry into a listed area automatically cancels the entire policy, or that section 11 always restores cover for a voyage outside an agreed geographical scope. Exact policy terms, applicable law, timing, and facts govern.
A chronological contract reconstruction should record at least: original schedule and navigation limits; current listed-area circular; broker or insured declaration; receipt time; insurer response and quote; insured acceptance; endorsement issue time and any retroactivity; actual boundary crossing; casualty time; and any agreed exclusion, deductible, or cancellation term. The aim is evidentiary clarity. The information also reveals operational risk: a claim may be disputed because of inconsistent timestamps even where the physical causal facts are well established.
Distinct cargo attachment and termination
Hull-war and cargo-war navigation cannot be modeled with one generic voyage cover flag. Under the 2009 Institute War Clauses (Cargo), ordinary attachment occurs as the insured cargo is loaded on an overseas vessel, with termination and reattachment defined around discharge and specific onward carriage. The document provides special provisions for intermediate ports, ports of refuge, changes of destination, mines, and deviations. Its exclusions expressly include delay and frustration of the voyage, with specified exceptions for general-average and salvage charges. This is a contractual boundary of major aggregation significance.
Consider a ship passing through a listed area while cargo remains insured under a warehouse-to-warehouse all-risks policy. It would be misleading to infer that the cargo's war coverage necessarily attaches throughout the warehouse-to-warehouse period on identical terms. The cargo war wording may attach only at overseas loading, include a limited intermediate-port period, and terminate at discharge. A port attack during transshipment can therefore require a close comparison of the exact war clause, endorsements, on-carriage provisions, and cargo location; a warehouse fire before loading may involve a different peril and coverage. The physical cargo's presence in a polygon does not by itself decide coverage.
The clause also illustrates why voyage deviations require contractual attention. Under the cited wording, a change of destination after attachment is subject to prompt notification and agreement of rates and terms, with conditional treatment of a loss before agreement. A model treating all diversion miles as unchanged war insurance exposure misses both the route hazard effect and changes in the covered time interval. The analyst must calculate the intersection of the new route with the war-risk zone and revise the insured-period indicator.
Time-stamped designation as a research object
Version control is essential because one public list can represent a moving political and operational assessment. A rigorous longitudinal database should store document ID, issuer, issue date, publication date, supersession links, language, exact geographic scope, exceptions and ports, coordinates, data source, and verification tier. The analyst should build a daily time-dependent lookup Z(x,t) rather than classify all historical voyages under the newest boundary. A classification for an incident on date t can be constructed from the list actually applicable and a separately documented contractual endorsement on that date.
A revision is evidence that the market's stated geographic assessment changed; it is not automatically proof that an individual vessel's physical probability of attack increased by a measurable factor. Announcements and revised boundaries may be responses to incidents, which creates endogeneity when testing whether listed status predicts attacks. Evaluation should therefore compare lagged hazard measures, incident chronology, control routes and changes in reporting coverage rather than simply regress war losses on listing status.
Route Concentration
The route as a time-varying exposure
A voyage is a space-time trajectory, not a point-to-point line. Let vessel i follow x_i(t) with draught, speed, ship type, insured value and route decisions changing over time. For a dated risk boundary Z_t, a basic measured dwell time is T_i(Z) = integral 1{x_i(t) in Z_t} dt; the total insured value simultaneously present is V_Z(t) = sum_i v_i(t) 1{x_i(t) in Z_t} c_i(t), where c_i(t) indicates relevant cover attachment. These are exposure accounting identities rather than a new stochastic theory. They become usable only when location observations, times, valuations and contracts are reconciled.
A port or strait transforms flows into concentrations. Ships queue near pilot stations, wait for convoy windows, load at terminals, and transit narrow channels at similar times. A common hostile act may affect several vessels in close proximity, while a series of attacks may alter the route of many ships and concentrate them elsewhere. These mechanisms imply that a count of ships transiting annually is insufficient for a simultaneous catastrophe footprint. The relevant object is the maximum and distribution of contemporaneous insured values, with attention to which policies cover each location and stage.
The IMO routeing discussion explains navigational systems as safety and efficiency instruments. Modern hydrographic systems have a further layer: the IHO standards in force provide versioned navigation data specifications, including the S-100 hydrographic data model. Neither standardized chart geometry nor an AIS track alone proves the legal or contractual status of a risk zone. It supplies spatial coordinates that can be joined, with documented precision, to legal and insurance records. The join is a research step requiring data validation.
Rerouting can redistribute rather than eliminate exposure
A ship avoiding a threatened strait may reduce exposure to targeted attacks in that region while extending voyage duration, increasing fuel consumption, altering schedule, and introducing different hazards. The UNCTAD Review of Maritime Transport 2025 documents prolonged Red Sea disruptions and major diversion around the Cape of Good Hope. UNCTAD estimates that route changes drove an increase in ton-miles in 2024, while by May 2025 Suez Canal tonnage remained far below its 2023 level. The statement is about observed trade routing and transport intensity, not a direct count of insured war claims.
An insurer's exposure can respond in opposing directions. One portfolio may experience fewer covered transits through the listed area and lower direct war severity; another may experience additional vessel-days and a rising total value of concurrent cargo in alternative hubs. A marine cargo insurer whose business is concentrated on a particular port may see a local reduction in reported accumulation; a logistics group may face a new concentration at its substitute transshipment port. Reinsurance net exposure can move again if one treaty includes the new geographic arrangement and another excludes it. Total expected loss cannot be inferred from the avoidance of one polygon.
There is also an operational selection effect. Insured ships retaining passage through a threatened route may differ systematically in cargo type, contract, nationality, value, financial constraints, risk tolerance, and protective procedures from ships rerouted. Post-diversion casualty rates for the remaining vessels cannot be treated as if they describe the entire pre-diversion fleet. Any empirical evaluation must model exposure by route choice and document why ships entered, remained, or avoided an area.
Navigation obligations, practical risk, and contractual action
Under the IMO's SOLAS framework, safe navigation is governed by international safety obligations implemented through states and maritime practice. Voyage planning under SOLAS chapter V considers appropriate charts, routeing systems, hazards, and environmental measures. The existence of conflict risk requires operational judgment by master, operator, flag state and other competent authorities; the insurance listing is additional input to their deliberation. A published risk map does not authorize an unsafe route, and a quoted war premium does not waive navigation law.
Operational records should include voyage plan and amendments, notices to mariners, security assessments, operator advisories, communications with insurers, port authority notices, logbook data, and independent incident reporting. Automatic Identification System (AIS) observations can contribute to reconstruction but may contain gaps, deliberate deactivation, identity errors, or positional anomalies; their completeness is a data-quality question. Independent corroboration by port calls, satellite imagery, flag-state reports or casualty records is especially important where war conditions impair normal reporting.
A quantitative claim about concentration needs a denominator representing the population genuinely at risk. For hull, possible denominators are insured vessel-days or insured vessel-value-days inside the zone. For cargo, they are covered shipment-value-days at sea or in eligible port extensions. For P&I, they may be insured ship-days adjusted for type and liability limit. None is interchangeable with raw AIS transits. Reporting an attack rate per ship-week while claiming a rate per dollar of cargo value creates avoidable distortion.
Operational case: Red Sea and Bab el-Mandeb
The Red Sea crisis permits observation of attacks, international response and route adjustment across different source types. UN Security Council Resolution 2722, adopted on January 10, 2024, condemned attacks on merchant and commercial shipping beginning in November 2023 and addressed the freedom of navigation. The IMO's Red Sea incident record identifies reported and confirmed incidents in its defined reporting periods and describes its role in supplying verified inputs to UN reporting. Those counts are governed by incident definition and verification status; they are not a comprehensive insurer loss triangle.
The reporting chronology continued through Security Council Resolution 2826 (2026), adopted on July 14, 2026, which extended the reporting request under Resolution 2722 through January 15, 2027. The September 2026 Secretary-General report transmitted an IMO-compiled update covering incidents reported through September 9, 2026. These documents establish the formal international reporting sequence as of this study’s cutoff. Their incident records remain operational and diplomatic evidence rather than insured-claim bordereaux.
The IMO July 2025 incident statement documents attacks on the bulk carriers Magic Seas and Eternity C and describes casualties and rescue information. A contemporaneous IMO incident update supplies vessel identities and incident-specific descriptions. The case is instructive because it combines identifiable vessels, physical attack mechanisms, life-safety effects and potential cargo and liability consequences. The statements alone do not establish what sums were insured, whether clauses attached, what recoveries were paid, or whether claims were aggregated by a particular treaty.
Comparing these incident records with UNCTAD routing data supports the existence of simultaneous security and trade-routing disruption. It does not permit the inference that all diverted freight expenditure represents a war-insured claim. For a loss study, each incident would require linked survey evidence, policy sections, cargo manifest interests, claims development, and ceded recoveries. This difference between operational and financial evidence explains why publicly visible attack counts cannot serve as direct calibrations of industry-wide insured war catastrophe losses.
Operational case: Black Sea and the September 2026 boundary
The IMO Black Sea and Sea of Azov notices document wartime security and shipping concerns beginning in 2022. Their chronology includes reports concerning stranded ships and floating mines. The IMO's May 24, 2022 circular on free-floating mines signals a mechanism extending beyond the zone of an immediate strike. In the insurance market, the September 2026 JWLA-035 update altered the stated Black Sea definition. These two types of documents answer different questions: the former records safety warnings and an operational hazard, and the latter records a change in an underwriting geographic list.
The exact 2026 circular excludes from the designated Black Sea area territorial waters of adjoining countries other than Russia and Ukraine. In Alandia’s September 18, 2026 notice, an individual leading underwriter announced a cancellation-and-reinstatement mechanism and a September 24 effective time for incorporating the revised listed areas into specified policies. That dated insurer notice demonstrates why the market circular date, insurer communication date, and effective contractual date must be recorded separately; the stated handling applies to the insurer and wordings described in that notice. This creates an important geospatial difference between the name of a water body and the legally described area of notification interest. It also exposes an empirical problem: vessels crossing different national territorial seas may be recorded as using 'the Black Sea' in trade databases but occupy different listed-area subregions. The effect on war policy response depends on the contract's adoption of the list and any separately agreed territorial definitions.
Comparative evidence from the Red Sea and Black Sea indicates that route concentration is conditioned by both chokepoints and incident mobility. In the Red Sea, narrow passage and coordinated rerouting give prominence to transit flows; in the Black Sea, ports, territorial sea exceptions, floating mines and persistent regional conflict complicate fixed zone indicators. This is an analytical synthesis of documented mechanisms, not a claim that one region has a quantitatively larger insured war loss distribution than the other. Such a ranking would require comparable exposure and claims data absent from the public source set.
War Loss Aggregation
Units, layers, and common causes
An accumulation is a collection of insured losses with dependence sufficient to threaten underwriting limits, reinsurance layers, liquidity, or solvency. Its size depends on the selected measure. Sum insured is the maximum specified value of covered interests before loss adjustment; gross modeled loss is the insurer's modeled obligation before reinsurance; net modeled loss reflects actually applicable ceded recoveries and other adjustments. Multiple contracts may respond to one incident, and several incidents may constitute one reinsurance occurrence only if the applicable contract supports that treatment.
Three causal patterns are especially important. A single casualty can damage a hull and many cargo interests. A spatial attack can affect multiple vessels, terminals, or storage sites. A campaign can create repeated attacks and persistent diversion, causing temporally correlated losses whose legal aggregation may differ by treaty. Because the same security process can cause loss at different times and locations, the empirical analyst should distinguish physical event clustering, claim co-movement and contractual aggregation definitions. A statistical cluster is evidence of dependence, not automatically a legal 'one event.'
The 2009 cargo war clause underscores how coverage gates narrow the insurable set before modeling severity. The contract's defined insured goods, loading, discharge, intermediate port conditions, excluded expenses and general-average provisions matter. On the liability side, the Gard International Group reinsurance circular describes a separate special excess war-risks P&I architecture for the 2025 policy year, including a USD 500 million per-event, per-ship reference and specified Russia/Ukraine/Belarus territorial treatment. These are institutional examples of differentiation across coverage, geography, year and layer, not universal marine war limits.
Gross-loss accounting and a common-cause expression
For insured interest j and event e, a transparent gross-loss accounting identity is L_e^gross = sum_j C_{j,e} min{(D_{j,e}-d_j)_+, M_j}, where D denotes otherwise eligible loss after policy-specific exclusions and valuation rules, d_j an applicable deductible, M_j an applicable remaining limit, and C is 1 only when the contract responds. This simplified expression suppresses coinsurance, contribution, multiple limits, adjustments, and special expense provisions; those must be added for an actual book. Its role here is to prevent the common mistake of aggregating every asset located in a zone at full value.
Let H_i be hull loss, G_{ik} cargo loss for shipment k on vessel i, P_i liability loss, and A properly allocated common expenditure. A causal loss ledger can display H_i, each G_{ik}, P_i, and A separately, then apply each policy and contribution rule before summing distinct insurer obligations. Salvage reimbursement, general-average contributions, and subrogation require accounting reconciliation; double counting them across hull and cargo can materially exaggerate the modeled event loss. The general-average rule in section 66 provides the legal reason to keep contribution and initial expenditure separate.
A loss record should also separate occurrence date, policy attachment date, notification date, claim reporting date, and settlement date. A war campaign may generate delayed reporting, disputed causation, and changing case reserves. Incident counts are typically indexed by attack date; actuarial triangles may be indexed by underwriting or occurrence period. A direct merge without these dates produces apparent discontinuities that are artifacts of recording rather than changes in hazard.
Dependence, tail risk, and aggregation
The difference between independent and common-cause losses can be described without inventing a specialized model. For losses L_1,...,L_n, the variance of total loss is Var(sum L_i) = sum Var(L_i) + 2 sum_{i<j} Cov(L_i,L_j). Positive covariance increases aggregate variance when marginal distributions are fixed. Tail dependence and extreme outcomes are more complex than variance, particularly when attacks, port shutdowns and rerouting are jointly driven by a persistent conflict regime. McNeil, Frey, and Embrechts' Quantitative Risk Management discusses dependence structures, aggregation and extreme-value methods; this monograph uses those established concepts to organize marine exposure rather than proposing a new formal risk-measure theory.
Correlation alone is inadequate for a severe multi-vessel loss. Two portfolios can have similar average claim correlations but different probabilities of concurrent total losses if one is exposed to shared terminal damage or coordinated attacks. Stress scenarios should therefore model at least a ship-specific event, a multi-ship terminal event, and a campaign affecting both passage probability and severity. Reinsurance stress tests should be recomputed under alternative contractual occurrence interpretations, because an econometric common factor does not determine treaty grouping.
A geographical hotspot also creates exposure correlation without requiring direct physical damage to every vessel. When many ships alter courses together, their arrival times and port stays may become more synchronized, and cargo value accumulates at a substitute hub. A treaty whose coverage includes port war risks may respond differently from a hull policy covering only a vessel at sea. Thus risk concentration can be transferred between places and insured interests even where the direct incident rate in the original zone falls.
Reproducible synthetic accumulation experiment
To make the arithmetic auditable, consider a wholly hypothetical portfolio with four vessels in the same defined area during a short event window. Each vessel is valued at USD 40 million. Each carries USD 30 million in separately insured cargo interests. Assume, only for this example, that applicable hull-war and cargo-war covers are in force, that all interests are insured by the same group on a 100% share, that relevant losses fall within policy limits, and that there are no deductibles, salvage recoveries, contribution offsets or third-party liabilities. Total declared insured values are USD 160 million hull plus USD 120 million cargo, or USD 280 million. None of these figures is drawn from a real insurer or market loss database.
In a single-vessel incident, assume the affected hull incurs 50% damage and its cargo incurs 40% loss. Modeled gross loss is 0.50 × 40 + 0.40 × 30 = USD 32 million. If two vessels are affected under identical severity assumptions, the loss is USD 64 million; if all four are affected, USD 128 million. The extreme result is four times the single-vessel scenario because the illustrative severities and values are fixed. That multiplication reflects exposure coincidence, not an empirically estimated frequency or a realistic upper policy limit.
| Synthetic outcome | Hull loss (USD m) | Cargo loss (USD m) | Total gross loss (USD m) |
|---|---|---|---|
| One vessel affected | 20 | 12 | 32 |
| Two vessels affected | 40 | 24 | 64 |
| Four vessels affected | 80 | 48 | 128 |
To demonstrate the separate effect of reinsurance terms, suppose the hypothetical insurer has per-occurrence excess-of-loss protection of USD 80 million above USD 20 million and an additional annual aggregate ceiling of USD 100 million for that cover. With only one occurrence in the year, the single-vessel USD 32 million loss produces min(max(32-20,0),80) = USD 12 million ceded and USD 20 million retained; the four-vessel USD 128 million loss produces USD 80 million ceded and USD 48 million retained. The result is conditional on all four losses meeting the contract's occurrence definition and the assumed coverage actually attaching. These terms are invented for illustration, not represented as market terms.
| Synthetic event loss | Ceded by illustrative layer (USD m) | Net after that layer (USD m) |
|---|---|---|
| 32 | 12 | 20 |
| 64 | 44 | 20 |
| 128 | 80 | 48 |
If the four affected vessels suffer separate attacks that the treaty treats as four separate occurrences, each USD 32 million claim would yield USD 12 million before aggregate restrictions, for a nominal total ceded USD 48 million and net USD 80 million. Under a hypothetical one-occurrence grouping the same physical aggregate USD 128 million loss would yield USD 80 million ceded and USD 48 million net. This USD 32 million difference is entirely an artifact of the example's contractual aggregation conditions. The example establishes why an 'event' identifier must be kept separate from an analyst's incident cluster. The annual aggregate ceiling is not binding in either variant; if multiple larger events occurred, it could become decisive.
A route-diversion exposure sensitivity
Now suppose the four vessels would ordinarily spend two days simultaneously inside the designated area. A diversion decision leaves only one vessel using that route, with the other three avoiding it. At the stated time, concentrated declared value falls from USD 280 million to USD 70 million. That 75% reduction in simultaneous listed-area value is an accounting result under the assumed positions and unchanged values; it is not a 75% reduction in expected insured loss. The alternative route may add voyage days and may cause those other vessels to concentrate at a different port. A model that records only the original area would miss the newly formed exposures.
A further sensitivity makes the coverage gate explicit. If the USD 30 million cargo interests on two of the original four vessels do not have war cover at the relevant time, the site's physically co-located declared asset value remains USD 280 million, while the war-insured amount eligible under the example falls by USD 60 million to USD 220 million. The correct figure depends on attachment, exclusion and interest definitions. Geography and insured exposure are separately measurable and jointly necessary.
The experiment demonstrates three monotonicities that hold only under its fixed assumptions: increasing the number of equally valued affected insured vessels increases gross loss; reducing the number present reduces the coincident value in that specific area; and more generous ceded recovery reduces the insurer's modeled net loss for a fixed gross outcome. Outside those assumptions, routing, selection, coverage and dependence can change in competing directions. These arithmetic observations are not new actuarial laws.
The P&I and reinsurance interface
A war casualty may create pollution, wreck removal, collision, passenger or crew liabilities. Whether any of those liabilities are covered, and by whom, requires the club's rules, the specific ship entry, special war cover, territorial exclusions and reinsurance arrangements. The 2025 Gard International Group circular discloses a war-risk excess program structure and regional buyback. The UK P&I Club's 2025 war-risk resolution similarly reports defined limits and geographical exclusions in that policy year. Their terms are dated illustrations of layered market arrangements, not a substitute for checking a 2026 certificate or a future wording.
The aggregation unit for P&I can be a ship and an event, while hull and cargo insurance may have other insureds and ceded treaty structures. It is therefore invalid to add a headline P&I 'per ship' limit to a hull agreed value and call that sum the vessel's probable physical war loss. P&I is contingent liability, often requiring a separate legal and factual determination, and there may be overlapping recoveries or contractual priorities. Scenario analysis should show separate gross liability development, legal allocation, and the relevant reinsurance attachment rather than absorb P&I into an all-purpose hull severity percentage.
Pricing feedback and the insurance capacity cycle
Underwriting decisions respond to both perceived probability and available capacity. After an incident, underwriters may raise additional premiums, reduce available limits, alter listed-area notification requirements, or exclude a specific exposure. These responses change which vessels enter the covered portfolio. Historical claims ratios observed after a repricing event may improve because weaker risks leave the book rather than because the underlying threat diminishes. The JWC's rating statement indicates why premiums must be reconstructed from individual negotiations instead of inferred from the area list. The 2026 IUMI market report describes broader market capacity and competition alongside geopolitical stress; these are different explanatory variables from individual vessel loss hazards.
An economically coherent retrospective study tracks offered terms, accepted quotes, route choice, declared values, ceded limits, and actual losses. It also records non-renewals and uninsured voyages when possible. Without that denominator, the selected insured sample may incorrectly appear safer after a dangerous event because its composition changed. This mechanism is a general insurance selection problem applied to a maritime spatial setting.
Evidence Appraisal and Sensitivity
Source hierarchy and documentary reliability
Primary law establishes legal text within its proper jurisdiction and temporal version. Clause documents establish what a standard wording says but not whether an individual insured purchased that wording unchanged. The JWC circular establishes an underwriting committee's listed geography, not the causal probability of a particular future attack. IMO incident records establish incidents meeting an official reporting and confirmation protocol, not private insurance paid claims. UNCTAD reports estimate shipping flows using specified source datasets. IUMI market totals concern premiums and selected loss-ratio reporting, not the national or insurer-level war sum insured in a given polygon. The 2026 IUMI statistical release and 2025 UNCTAD transport review should therefore be interpreted on their own units and reference periods.
The third-party-hosted JWLA-035 PDF receives a specific provenance qualification: the original issuer name, date, and geography are visible in the document, while the current official association indexes were used to corroborate existence and currency. This is stronger than relying on an unlabeled excerpt; it is weaker than independent inspection of a digitally signed issuer original with immutable accession metadata. The IUA record of listed circulars is the official independent check. A subsequent archival update should replace or supplement the mirror only after verifying document identity and version.
Commercial AIS histories, claims bordereaux, premium submissions, insurance endorsements, and reinsurance treaties are typically not available in the open institutional source set. This creates a strict evidentiary ceiling. The study supports mechanics and documented cases, and it supports synthetic quantitative demonstrations. It does not claim an externally validated estimate of expected annual aggregate war losses, a fitted regional hazard model, or insurer-specific maximum probable loss.
Population, observation, and selection bias
Attack reporting is event-driven. Port throughput is movement-driven. Insurance claims are contract-driven. The selection rules do not match. IMO's confirmed Red Sea incident reporting can omit events outside its definition or confirmation window, while trade-route estimates may capture a vessel's decision to divert without observing any insured loss. If a dataset combines the two, an identifiable denominator must be the vessel population exposed during the same period, by location, ship class and reporting status. It should not silently treat vessels that ceased operating in a region as non-casualties with the same exposure as vessels that remained.
There is a serious post-treatment issue when list revisions follow attacks. A regression using today's listed areas to explain yesterday's loss pattern leaks future information into the predictor. Proper design uses the circular valid before an observed voyage or incident, with lagged conflict indicators and explicit measurement of policy adoption. Equally, a comparison between ships transiting and avoiding a risk area is affected by self-selection, charter obligations and route suitability. Identification might require matched ship type and value, fixed origin-destination pairs, vessel-level histories and documented changes in advice; even then causal claims about the pricing or protective effect of listing require careful justification.
Alternative explanations for observed trends
A fall in reported incidents after a major route diversion can arise from lower traffic rather than lower hazard conditional on transit. An apparent rise in aggregate insurance claims may reflect higher declared vessel and cargo values, more policies purchasing the cover, a shift toward high-value goods, an expansion of reporting, or a truly more severe attack distribution. A decline in net loss can arise from improved reinsurance attachment, favorable development, and commutations rather than improved operational security. Separating these explanations is central to scientifically meaningful attribution.
A useful sensitivity matrix varies independently: listed-area polygon and revision date; AIS data availability and interpolation threshold; insured value inflation; cargo cover attachment and termination; hull casualty severity; liability development; one-event versus multiple-event classification; deductible and occurrence limit; reinsurance reinstatement; and missing-claims reporting lag. The base case should never conceal a binary assumption about war-cover eligibility. That assumption can change modeled recoverability from a positive amount to zero without altering physical loss.
The published quantitative literature also cautions against treating correlation as a complete description of co-extreme behavior. The McNeil–Frey–Embrechts treatment establishes the mathematical background for dependence and aggregate-risk analysis, but it does not provide a ready-calibrated marine war model. For this subject, model risk is driven at least as much by contractual definition, exposure data and sparse extreme events as by the selection of a dependence family.
Legal and temporal sensitivity
The Marine Insurance Act 1906 and Insurance Act 2015 are analyzed as United Kingdom statutory texts applicable in the relevant contexts and subject to applicable contractual and conflict-of-laws rules. Other jurisdictions may differ in causation, warranty, disclosure, sanctions, limits and claims handling. For an international study, local law and treaty obligations must be determined for each legal question. The Marine Insurance Act's proximate-cause provision is a precise statement about a legal default, and the 2015 warranty change is a precise change within its own scope; neither is an international rule enacted by every coastal or flag state.
The same discipline applies to policy years and the meaning of 'current.' The 2025 P&I examples are not presented as the applicable wording for November 2026. The JWLA-035 boundary is dated September 2026, and the war conditions can evolve after the research cutoff. Security Council Resolution 2812 (2026) extended the reporting request through July 15, 2026, while Resolution 2826 (2026) subsequently extended it through January 15, 2027. The exact sequence matters: citing the January extension alone as the mandate applicable in November would be an outdated description. Scientific statements therefore specify what a source proves and when.
Quality-control results and limitations
Internal scholarly review for this manuscript checks that the zone authority, contract, public law, operational data and insurer accounting layers remain distinct; that numerical statements are traced to dated official sources or clearly labeled synthetic assumptions; that the reinsurance calculations are reproducible; and that no externally unverified war-loss frequency or actual market premium quote is asserted. Editorial review checks one identical H1 and title, American English, short contextual source links, source-specific attributions, coherent chapter logic, and complete references. These are internal authorial and editorial checks, not an independent external peer review.
Material limitations are identifiable: direct complete copies of all bespoke hull war policies and market-specific endorsements were not inspected; the illustrative Institute War Clauses (Cargo) represent a common wording rather than universal coverage; exact premium transaction data and granular claims settlements were unavailable; the JWC circular's detailed geography was inspected in a third-party-hosted copy cross-checked to official listings; and the UNCTAD/IUMI published aggregate statistics cannot provide vessel-level insured amounts. These limitations constrain inference but do not prevent a definitive documentary synthesis or correctness checks on the disclosed synthetic calculations.
Scientific Synthesis and Research Questions
Integrated findings
The reviewed records support seven connected findings. The first is that marine war-risk geographical listing originates in underwriting judgment and gains contractual effect through policy-specific incorporation, notification and negotiation. The second is that the legal right of passage, IMO navigation requirements, commercial listed areas and policy navigation conditions are separate source families and therefore must be separately cited. The third is that hull, cargo and P&I interests attached to the same voyage create heterogeneous insured consequences with distinct periods, parties and limits. The fourth is that route diversion can lower exposure at one chokepoint while increasing exposure duration or concentration elsewhere. The fifth is that common-cause war losses require synchronized records of vessels, cargo, locations and claims, not a simple count of listed-zone ships. The sixth is that the definition of reinsurance occurrence can materially alter net loss even where physical loss is held fixed. The seventh is that published attack and shipping-flow statistics support mechanism identification but cannot alone calibrate a private insurance loss distribution.
These findings follow from the JWC institutional and contract distinction, the cargo war duration provisions, and the combined IMO Red Sea case materials. Their application does not require a new doctrine. The research contribution is the explicit cross-document reconciliation of geographical, temporal, legal, actuarial and financial evidence, with illustrative arithmetic that exposes the direction of several otherwise hidden measurement errors.
An auditable exposure reconstruction
A scientifically reproducible marine war portfolio can be organized as linked observation tables. The zone table holds issuer, document version, geometry and applicable dates. The vessel table holds IMO number, vessel characteristics and track confidence. The voyage table holds actual route, course changes, port visits, draught and relevant exposure interval. The cargo table holds shipment values, interests and the mode/time of attachment. The policy table holds assured, insurer share, wording, limits, deductibles, endorsements and governing law. The incident table holds verified date, location, type, causal sequence and source. The claim table holds coverage determination and development. The reinsurance table holds contract year, layer, geographic restrictions, occurrence definition and recoveries. These are ordinary research data structures; their validity is assessed by their ability to answer specific questions consistently.
Linking should be accomplished with stable identifiers and explicit uncertainty rather than forced one-to-one relationships. One vessel may have several cargo interests, several hull interests, different underwriting participants and successive claims. One incident may affect multiple vessels and one vessel may be damaged more than once. Records should permit many-to-many mappings and retain the original documentary source. The scientific passport for a modeled portfolio should include a test that reconstructs every gross loss from underlying interests and every net loss from treaty arithmetic.
Research agenda and evidentiary tests
The most valuable next empirical work would assemble multi-year confidential or suitably anonymized vessel-level hull war declarations, cargo values, route histories, clause versions and claim records alongside dated JWC boundaries. It would then estimate conditional exposure and loss under explicit selection adjustment. A second project would study incident clusters against insured-value-weighted contemporaneous traffic rather than vessel counts. A third would examine how reinsurance event definitions and treaty exclusions affect catastrophe recoverables. A fourth would reconstruct port and transshipment cargo concentration during large-scale route diversions. Each requires data access, documentary permissions and robust governance.
A falsifiable empirical question is whether a preannounced listed-area expansion predicts a larger subsequent reduction in covered insured-value-days than in all-vessel transit-days, after controlling for fixed route and ship characteristics. Another is whether cross-line hull/cargo war claims exhibit greater conditional tail dependence during common attack campaigns than do geographically similar peacetime marine casualties. A third is whether contractual event grouping changes the ranking of ceded losses across plausible simultaneous attack scenarios. No result is assumed here; those are test designs whose conclusions depend on future observations and identified assumptions.
A research program can also compare official incident records to insurer claims emergence by development month. The expected gaps should be characterized by definitional differences and claim reporting delay, not retrospectively filled with invented paid-loss observations. The IMO incident register and the UNCTAD route report provide credible external benchmarks for security and trade movement, but the causal translation into insurance recoveries must be established through contracts and financial data.
Operational meaning for risk science
For underwriters, the principal implication is to distinguish declared zone entry from the concurrent accumulation of actually insured interests. For risk managers, it is to document route and cargo changes that affect coverage periods. For actuaries, it is to calibrate loss only with exposure units commensurate with the outcome and to stress dependence beyond independent vessel assumptions. For reinsurers, it is to audit aggregation under the actual event definition. For researchers, it is to preserve the genealogy of each geographical revision, the chain of evidence and the limitations of public incident statistics. These are practical consequences of the analyzed instruments, not personalized insurance advice.
Conclusion
Marine war risk is governed by the joint evolution of navigation geography, physical threat, contractual coverage and interdependent values at risk. A listed area provides a dated underwriting signal; a voyage generates a changing exposure path; a clause defines the covered interest and attachment; and a hostile event can produce losses across hull, cargo, liability and reinsurance relationships. The published documentary record establishes these distinctions, and the Red Sea and Black Sea cases show why they matter in concrete operating environments. The synthetic calculations demonstrate how the same asset distribution can generate different gross and net results when co-location, coverage attachment and occurrence wording are changed. A credible accumulation assessment therefore begins with verified contractual and geospatial data, then models common causes and ceded recoveries under explicit uncertainty. The defensible outcome of this monograph is a traceable analytical synthesis and research agenda, with empirical limits disclosed and no claim to a newly established theory.
References
Gard. (2024). Reinsurance arrangements for the 2025 policy year arranged through the International Group of P&I Clubs, Member Circular 13/2024. https://gard.no/en/circulars/13-2024-reinsurance-arrangements-for-the-2025-policy-year-arranged-through/
International Hydrographic Organization. (2026). S-100 standards in force (standards table updated October 29, 2026). https://iho.int/en/s-100-standards-in-force-0
International Maritime Organization. (n.d.). Ships' routeing. https://www.imo.org/en/ourwork/safety/pages/shipsrouteing.aspx
International Maritime Organization. (n.d.). International Convention for the Safety of Life at Sea (SOLAS), 1974. https://www.imo.org/en/About/Conventions/Pages/International-Convention-for-the-Safety-of-Life-at-Sea-%28SOLAS%29%2C-1974.aspx
International Maritime Organization. (2022–2026). Black Sea and Sea of Azov: Advice for IMO Member States, seafarers and shipping (circular index). https://www.imo.org/en/mediacentre/hottopics/pages/cls4524.aspx
International Maritime Organization. (2025). Secretary-General urges constructive dialogue amid renewed Red Sea attacks (July 8, 2025). https://www.imo.org/en/mediacentre/pages/whatsnew-2285.aspx
International Maritime Organization. (2025). Report on Houthi attacks on merchant and commercial vessels in the Red Sea, 8 June–9 July 2025 (incident update). https://wwwcdn.imo.org/localresources/en/MediaCentre/Documents/Red%20Sea%20Incidents%20update%209%20July%202025.pdf
International Maritime Organization. (2023–2026). Red Sea area: Statements and confirmed incidents (continuously updated official page; inspected November 17, 2026). https://www.imo.org/en/mediacentre/hottopics/pages/red-sea.aspx
International Union of Marine Insurance. (2026). IUMI Stats Report 2026 (October 23, 2026). https://iumi.com/statistics/iumi-stats-report-2026/
International Union of Marine Insurance. (2026). Facts & Figures Press Release 2026 (September 22, 2026). https://iumi.com/statistics/facts-figures-press-release-2026/
International Underwriting Association of London. (2009). Institute War Clauses (Cargo), CL385, 1 January 2009 (clause issuer index). https://www.iua.co.uk/IUA_Member/ClauseDownload?id=2&ordinal=1051
International Underwriting Association of London. (2026). Joint War Committee Risk List: Listed Areas (including JWLA-035 index). https://www.iua.co.uk/IUA_Member/IUA/UnderwritingItems/Committees/Joint_War_Committee_Risk_List.aspx
Joint War Committee (LMA/IUA). (2026). JWC Listed Areas: Hull War, Piracy, Terrorism and Related Perils, JWLA-035 (circular dated September 16, 2026; third-party hosted copy checked against issuer indexes). https://mayan.co.il/wp-content/uploads/2026/11/jwc-listed-areas-jwla-035-2026-09-16.pdf
Lloyd's Market Association. (2026). Joint War Committee (committee remit, market guidance and listing). https://lmalloyds.com/committee/joint-war-committee/
Lloyd's Market Association. (2026). Listed Areas (current JWLA-035 reference). https://lmalloyds.com/specialist-areas/underwriting/listed-areas/
Lloyd's Market Association and International Underwriting Association of London. (2009). Institute War Clauses (Cargo), CL385, 1/1/09 (available full wording, hosted industry copy). https://cee-specialty.eu/images/Download/ENG/institute-war-clauses-cargo-2009.pdf
McNeil, A. J., Frey, R., & Embrechts, P. (2015). Quantitative Risk Management: Concepts, Techniques and Tools (revised ed.). Princeton University Press. https://press.princeton.edu/books/hardcover/9780691166278/quantitative-risk-management
Alandia. (2026). JWLA-035: Notice of revised listed areas and policy incorporation (September 18, 2026). https://alandia.com/news/jwla-035/
UK P&I Club. (2025). Board resolutions in respect of war risk, supplementary bio-chem and oil pollution cover (February 13, 2025). https://www.ukpandi.com/news-and-resources/circulars/article/circular-05/25-uk-club-board-resolutions-in-respect-of-war-risk-supplementary-bio-chem-and-oil-pollution-cover/
United Kingdom. (1906). Marine Insurance Act 1906, section 3 (marine adventure and maritime perils). https://www.legislation.gov.uk/ukpga/Edw7/6/41/section/3
United Kingdom. (1906). Marine Insurance Act 1906, section 55 (included and excluded losses). https://www.legislation.gov.uk/ukpga/Edw7/6/41/section/55
United Kingdom. (1906). Marine Insurance Act 1906, section 66 (general average loss). https://www.legislation.gov.uk/ukpga/Edw7/6/41/section/66
United Kingdom. (2015). Insurance Act 2015, section 10 (breach of warranty). https://www.legislation.gov.uk/ukpga/2015/4/section/10
United Kingdom. (2015). Insurance Act 2015, section 11 (terms not relevant to the actual loss). https://www.legislation.gov.uk/ukpga/2015/4/section/11
United Nations. (1982). United Nations Convention on the Law of the Sea, Part II (territorial sea and contiguous zone). https://www.un.org/depts/los/convention_agreements/texts/unclos/part2.htm
United Nations Security Council. (2024). Resolution 2722 (2024), S/RES/2722 (January 10, 2024). https://mandates.un.org/mandate/S/RES/2722%20%282024%29
United Nations Security Council. (2026). Resolution 2812 (2026), S/RES/2812 (January 14, 2026). https://mandates.un.org/mandate/S/RES/2812%20%282026%29
United Nations Security Council. (2026). Resolution 2826 (2026), S/RES/2826 (July 14, 2026). https://digitallibrary.un.org/record/4121238
United Nations Secretary-General. (2026). Letter dated 10 September 2026 transmitting the IMO update on attacks against merchant vessels in the Red Sea, S/2026/721. https://digitallibrary.un.org/record/4127881
UN Trade and Development (UNCTAD). (2025). Review of Maritime Transport 2025: Staying the Course in Turbulent Waters (September 24, 2025). https://unctad.org/publication/review-maritime-transport-2025