Pith. sign in

REVIEW 3 major objections 6 minor 36 references

Closing a maritime chokepoint costs far more than the trade that passes through it, because complementary intermediate inputs break downstream production.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-14 14:21 UTC pith:DCYIRCDQ

load-bearing objection Solid structural counterfactual: losses ≠ transit value, heavy-tailed incidence, exporter skew from geography under a neutral map, and joint sub/super-additivity from series vs parallel corridors—with levels as unbuffered upper envelopes. the 3 major comments →

arxiv 2607.09951 v1 pith:DCYIRCDQ submitted 2026-07-10 econ.GN q-fin.EC

Macroeconomic Risks from Maritime Trade Disruptions

Vipin P. Veetil , Fathimath S. Vemmarath This is my paper
classification econ.GN q-fin.EC
keywords Maritime chokepointsProduction networksTrade disruptionsSupply-chain riskInput-outputComplementarityJoint closuresExporter-importer asymmetry
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper argues that the economic cost of shutting a canal or strait is not measured, and is not even bounded, by the value of cargo that normally transits it. Losses arise because ships carry intermediate goods that are complementary in production: missing crude, feedstocks, or components leave downstream plants with incomplete input bundles that re-matching cannot fully restore. Buyers and sellers can reassign displaced orders within sectors under capacity caps and a reallocation friction, which limits damage but cannot eliminate it. The resulting country losses are heavy-tailed; they hit commodity-concentrated exporters hardest, reach economies that never ship through the passage, and make the two ends of a corridor lose unequally. Joint closures can cost more or less than the sum of their parts depending on whether the gates duplicate the same corridor or starve common buyers of distinct complementary inputs.

Core claim

Interrupting a shipping passage produces macroeconomic losses that stem from complementary intermediate-input disruptions, not from the transit value alone. Re-matching displaced trade on both sides of the market limits the damage but cannot eliminate it. Incidence is heavy-tailed and reaches non-transiting economies; exporters on commodity corridors lose several times more than importers; and joint closures are sub-additive when gates share corridors and super-additive when they starve common buyers of distinct complementary inputs.

What carries the argument

A short-run production-network counterfactual: route geography severs links (net of detours and leakage), two-sided capacity-capped RAS re-matching rewrites the order book under a reallocation friction, and CES nests penalize incomplete input bundles. The post-closure benchmark-selected equilibrium yields an exact four-channel identity that splits every country’s loss into direct severance, network propagation, and bundle penalties on autonomous and intermediate output.

Load-bearing premise

The model treats closures as unbuffered short-run shocks: no inventories, strategic reserves, price-driven demand substitution, or capacity expansion, with only a fixed residual leak through the closed passage and partial within-sector re-matching.

What would settle it

Measure country-sector output after a sustained multi-gate episode that severs distinct complementary inputs at common buyers (for example Malacca plus the Taiwan corridor): if world and buyer losses stay at or below the sum of the single-gate losses after buffers are accounted for, the super-additivity claim fails; if exporters on commodity corridors do not lose several times more than their diversified buyers, the asymmetry claim fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper builds a short-run world production-network model of maritime chokepoint closures. Closures sever intermediate-input links according to route geography and a leakage residual; displaced trade is re-matched by capacity-capped two-sided RAS balancing; and incomplete bundles are penalized by nested CES technology. Losses are computed from a benchmark-selected triangular fixed point and an exact four-channel identity. On an extended 87-economy, 50-sector 2019 ICIO table the paper reports: (i) heavy-tailed country exposure (Hormuz ~0.6% of world VA, UAE ~17% of GDP); (ii) exporter–importer asymmetry of several times on commodity corridors, inherited from bipartite geography under a neutral balancing map; (iii) joint-closure sub-additivity for the Middle East scenario and super-additivity for East Asia and Russia–Europe, disciplined by linear Bonferroni-style sub-additivity and a CES condition for super-additivity. Validation against the 2023–24 Red Sea episode checks a parameter-free network footprint and treats scale as an unbuffered envelope.

Significance. If the results hold, the paper supplies a usable structural alternative to transit-value gauges of chokepoint risk and a clean series-versus-parallel account of joint-closure additivity. The technical package is a genuine contribution: support-preserving biproportional balancing with neutrality (Lemma 2), a triangular fulfillment fixed point with maximality and interior uniqueness under a cushion condition, an exact four-channel loss identity, detour-equivalence classes that prevent double-counting, and Propositions 5–6 that separate linear sub-additivity from CES-driven super-additivity. Public code, Zenodo inputs, and a pip-installable interactive package raise the bar for replication and policy use. The Red Sea footprint test is a serious, parameter-free check of transmission shape. The main quantitative magnitudes should be read as unbuffered upper envelopes, which the authors largely acknowledge; the topological claims are more secure than the point estimates of tail depths and loss multiples.

major comments (3)
  1. §4.1, Assumptions 1–2, eqs. (5)–(6) and (10)–(12): The headline quantitative objects (λmax tails, Hormuz exporter multiples ~17×, S(QME)=0.8 vs S(QEA)=1.3) are produced under the unbuffered short-run envelope—no inventories, strategic reserves, price substitution, or capacity expansion, with baseline δ≡0.10 and τ=0.3. OA.4 shows ranking and sign robustness of S across the (τ,ρ) rectangle, but does not report how the exporter–importer ratios or Zipf tail coefficients compress under δ_energy=0.45, higher recovery, or a buffered τ. Because §6 already shows scale over-prediction, the abstract and §4 should either (a) report these magnitudes under the commodity-differentiated and higher-recovery corners as primary results, or (b) reframe the abstract numbers explicitly as upper envelopes rather than as the paper’s central quantitative claims.
  2. §6.1–6.2: The Red Sea validation cleanly separates a parameter-free footprint (shape) from an external scale envelope, and the orthogonal placebos (Panama, Korea) are informative. It does not, however, validate the two load-bearing cross-sectional claims of §4.4–4.6: relative country loss ratios and the exporter–importer asymmetry. The panel is limited to OECD majors at quarterly frequency and deliberately absorbs country-time aggregates. Either extend the validation to relative losses among the data-rich majors (e.g., Japan/Korea/India vs US/Germany under the Red Sea footprint) or state clearly that the asymmetry and heavy-tail magnitudes remain unvalidated against realized multi-country outcomes and rest on the structural geography of Z*.
  3. §5.2–5.3, Proposition 6 and Table 7: Super-additivity is an existence result for sufficiently complementary inputs (ρ<ρ⋆) under the distinct-sectoral-input hypothesis; at baseline ρ=−1 the East Asia and Russia–Europe joints are only mildly super-additive (S=1.3 and 1.1) while Middle East is sub-additive (0.8). The pairwise Table 9 is helpful, but the paper should report, for each joint scenario, the share of the departure from additivity attributable to Bonferroni overlap versus the CES cascade (e.g., by comparing LlinW to full LW). Without that decomposition it is hard to judge how much of the reported S indices is the nonlinear mechanism the theory emphasizes versus residual geographic overlap after the detour-class correction (39).
minor comments (6)
  1. Abstract and §1: “three to five times” understates the corridor-wide and leading-exporter ratios reported in §4.6 (up to ~17× at Hormuz on the leading-exporter basis; corridor-wide ratios are milder). Align the abstract wording with the figure that is actually shown.
  2. Figure 2 and Figure 10: The common color scale with γ=0.5 (or square-root) transform is necessary but should be stated in the figure notes more prominently; several panels look nearly empty without it.
  3. §3.4–3.5: The incidence rule (36) is clear, but a short table of the share of seaborne intermediate value assigned to each of the twelve gates (and the overlap mass on multi-gate corridors) would help readers gauge sparsity and double-counting risk before §4.
  4. Notation: λc is used both for country proportional GDP loss (30) and, nearby, for node-level objects; eλj for the output multiplier is easy to confuse with λc. A one-line notation reminder at the start of §4 would help.
  5. §1.2 and the maritime-choke package: Excellent replication posture. Please pin the exact package version / commit hash that regenerates the tables in the paper so that future bit-rot does not break the claim.
  6. Typos / polish: “offthe”, “anda”, “reading offthe” and similar missing spaces appear in several places (e.g., around the Red Sea and joint-scenario discussion). A copy-edit pass is needed.

Circularity Check

1 steps flagged

No material circularity: parameters and incidence come from external evidence and geography; companion-model self-citation is methodological, not a load-bearing uniqueness chain that forces the empirical claims.

specific steps
  1. self citation load bearing [§1 Introduction; §2 The Model (opening); related literature §1.1]
    "The model is the short-run production-network specification of Bhattathiripad and Veetil (2026), developed there for trade sanctions and adapted here to maritime closure. ... In this it is a companion to Bhattathiripad and Veetil (2026), whose short-run production-network model we adapt, and which turns the same two-sided balancing to the measurement of national power under trade sanctions."

    The CES nests, capacity-capped two-sided RAS balancing, fulfillment fixed point, and benchmark-selected equilibrium are imported from the authors’ own companion paper rather than re-derived from external first principles. This is methodological self-citation of the apparatus, not a uniqueness theorem that forbids alternatives or a fit that renames the chokepoint loss targets as predictions. Incidence Γ, reroutability E, calibration from OA.5, and the Red Sea parameter-free footprint remain independent of that citation, so the step is minor and not load-bearing for the empirical claims.

full rationale

The paper’s central objects—country losses λc, heavy-tailed worst-case exposure, exporter–importer ratios, and joint super-additivity indices S(Q)—are computed by solving a structural counterfactual on an incidence matrix built from route geography, seaborne-sector restrictions, and observed ICIO/GTAP flows (Sections 3–5), not by fitting those targets. Structural parameters (τ, ρ, ϱ, δ) are taken from external recovery-rate, production-function, and historical transit evidence (Table 3; Online Appendix OA.5), with the baseline described as the center of those independently estimated intervals rather than a fit to chokepoint GDP losses. The Red Sea exercise separates shape from scale: the footprint IND is built from Z*, Γ, and E with no free parameter and is tested via a continuous-treatment event study; scale is checked against external OECD/ECB/UNCTAD/Suez-receipts anchors and is acknowledged to over-predict as an unbuffered envelope (Section 6). Lemma 2 makes the RAS balancing neutral so asymmetry is inherited from the bipartite geometry of Z*, not imposed by the map; Props. 5–6 give linear sub-additivity and a CES condition for super-additivity that are then evaluated on data. The only circularity-adjacent element is adaptation of the short-run production-network apparatus from the authors’ companion sanctions paper; that is ordinary methodological self-citation and does not reduce the maritime incidence results or the Red Sea shape test to inputs by construction. Score 2 reflects that minor self-citation without elevating it to a forced derivation.

Axiom & Free-Parameter Ledger

5 free parameters · 8 axioms · 3 invented entities

The central quantitative claims rest on a short-run production-network counterfactual whose primitives are mostly standard IO and CES objects, plus a small set of calibrated frictions and a geographic incidence construction. Free parameters set the level of losses; domain assumptions (fixed capacity, complementary inputs, least-cost routing, leakage) determine the qualitative pattern. No new physical entity is postulated—the invented objects are modeling constructs (incidence matrix, detour classes, benchmark-selected equilibrium).

free parameters (5)
  • reallocation friction τ = 0.3 (range ~0.2–0.5 in robustness)
    Share of displaced same-sector trade not re-matched within the horizon; baseline 0.3 from firm recovery literature, not estimated on chokepoint losses.
  • inner CES exponent ρ = −1 (sensitivity across more negative values)
    Complementarity across sectoral intermediate inputs; baseline −1 at the mild end of short-run network-macro range.
  • outer-nest curvature ϱ = −1
    Complementarity between intermediate composite and non-intermediate block; baseline −1.
  • closure leakage δ (and δ_energy) = δ≡0.10 baseline; δ_energy=0.45 robustness
    Residual share of a fully exposed flow still transiting a closed passage; uniform baseline 0.10, energy robustness 0.45 from historical transit episodes.
  • reroutability entries e_k,cc' / detour distances Δ_k = geographically calibrated matrices (OA.2)
    Geographic shares of loss after detours; calibrated from route distances rather than free economic fit, but still model inputs that grade severity by corridor.
axioms (8)
  • domain assumption Short-run capacity: no country-sector expands activity above benchmark; ports, relationships, and scale are fixed on the disruption horizon.
    Imposed via capped RAS row/column targets and activity box [0,x] (§2.2–2.4).
  • domain assumption Intermediate inputs across sectors are complements (ρ,ϱ<0); within-sector country sources are substitutes subject to friction τ.
    CES nests and block-diagonal reallocation (§2.2–2.3).
  • domain assumption Benchmark seaborne flows follow least-cost maritime routes; incidence is route × seaborne-seller-sector × positive flow.
    Routing rules and Γ construction (§3).
  • ad hoc to paper Closures retain positive leakage δ_s>0 so balancing and fixed point remain well-posed; full severance is a singular limit.
    Assumption 1 and Proposition 1; economically motivated by war-risk residuals but also a mathematical regularizer.
  • domain assumption Two-sided RAS/Sinkhorn balancing is the post-closure order book: least-revision feasible reorganization under row and column caps, not market-clearing prices.
    Definition 1; neutrality Lemma 2 underpins asymmetry as measurement.
  • ad hoc to paper Benchmark-selected (maximal) equilibrium is the reported counterfactual among possibly multiple fixed points.
    Definition 3 and Proposition 3; conservative least-loss selection.
  • standard math Standard biproportional balancing / Sinkhorn existence on positive support and dissipative IO propagation.
    Used throughout §2 and OA.3.
  • domain assumption Value-added losses are unbuffered gross VA declines, not welfare; no inventory or price channel.
    §2.5 footnote and §4.1 calibration discussion.
invented entities (3)
  • Chokepoint incidence matrix Γ and detour-equivalence classes D independent evidence
    purpose: Map geography into which intermediate links a closure severs and how joint closures share detours without double-counting.
    Constructed objects from routes and ports, not new physical forces; falsifiable against shipping-lane data but idealized at regional resolution.
  • Benchmark-selected post-closure equilibrium (ex,eh) with fulfillment vector h no independent evidence
    purpose: Define the unique interior least-loss fixed point used for all reported losses.
    Selection device among fixed points of the delivery loop; standard in spirit but paper-specific.
  • Four-channel loss identity (direct severance, network propagation, ζ, ξ) no independent evidence
    purpose: Apportion country losses into economically distinct mechanisms at equilibrium.
    Accounting identity at the model’s fixed point (eq. 28), not an external measurement.

pith-pipeline@v1.1.0-grok45 · 56125 in / 4642 out tokens · 47247 ms · 2026-07-14T14:21:29.149123+00:00 · methodology

0 comments
read the original abstract

This paper develops a model of maritime chokepoint closures in which interrupting a shipping passage produces losses that are not measured, or even bounded, by the value of the trade that transits it. They stem from disruptions in the flow of intermediate goods that are complementary in downstream production. Re-matching displaced trade on the buyer and seller sides of the market limits the damage but cannot eliminate it. Across countries, the incidence is heavy-tailed, and it reaches economies whose cargo never crosses the passage, not only those that route their trade through it. The two ends of a severed corridor lose unequally, the exporting side by three to five times as economic geography funnels commodity-concentrated sellers through a single passage while their buyers re-source. Joint closures depart from the sum of their parts: the Middle East scenario is sub-additive, while the East Asia and Russia-Europe scenarios are super-additive.

Figures

Figures reproduced from arXiv: 2607.09951 by Fathimath S. Vemmarath, Vipin P. Veetil.

Figure 1
Figure 1. Figure 1: The twelve maritime chokepoints studied in the paper. A [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Country-level GDP loss λc ({k}) on a world map for each of the twelve single-chokepoint closures, common color scale across panels (power transform γ = 0.5 so the smaller closures retain visible structure). ROW-block aggregates are expanded onto their constituent territories listed in Table OA.1.1. Gray territories are not represented in the extended ICIO universe. The heatmap sorts economies into three ti… view at source ↗
Figure 3
Figure 3. Figure 3: Rank-size (Zipf) plots of worst-case exposure at the baseline calibration ( [PITH_FULL_IMAGE:figures/full_fig_p041_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Mean GDP loss versus tail GDP loss. The worst case [PITH_FULL_IMAGE:figures/full_fig_p044_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Cross-chokepoint exposure profile of major economies. [PITH_FULL_IMAGE:figures/full_fig_p046_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Exporter–importer loss asymmetry, by chokepoint. For each single closure the bars give [PITH_FULL_IMAGE:figures/full_fig_p048_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Channel decomposition of country-level value-added loss under closure of the Strait of [PITH_FULL_IMAGE:figures/full_fig_p051_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Network-propagation share of country-level value-added loss for the top-twenty [PITH_FULL_IMAGE:figures/full_fig_p053_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Direct-impact share of country-level value-added loss for thirteen African economies, [PITH_FULL_IMAGE:figures/full_fig_p055_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: World-map choropleth of country-level value-added loss [PITH_FULL_IMAGE:figures/full_fig_p065_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Incidence and timing: event study of the 2023–24 Red Sea disruption. Each point is the coefficient bk on model-predicted network-propagated Red Sea exposure interacted with event time k (Appendix OA.6.1, equation (OA.6.1)), in percentage points of value-added growth per one-standard-deviation of exposure, with 95% confidence intervals clustered by country; the base period is k = −1 and the dashed line mar… view at source ↗
Figure 12
Figure 12. Figure 12: Specificity: the horse-race against orthogonal gate footprints. The realized decline is regressed jointly on the Red Sea footprint and on the footprints of the gates uncorrelated with it (Panama, Korea), inside the full fixed effects. Squares are single-footprint estimates, circles the joint estimates, with 95% confidence intervals clustered by country. The Red Sea coefficient survives while the orthogona… view at source ↗
Figure 13
Figure 13. Figure 13: The parameters whose unbuffered envelopes are consistent with the observed falls coincide with the independently estimated box. World value-added loss (a) and Egypt’s goods￾network loss component (b) from the Red Sea closure across the calibration rectangle τ ∈ [0.2,0.5], ρ ∈ [−1,−0.25] at the container-corridor leakage δ = 0.10 (make coherence figure.py). Green rectangles mark the parameter box estimated… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

36 extracted references

  1. [1]

    The Network Origins of Aggregate Fluctuations

    Acemoglu, Daron, Vasco M. Carvalho, Asuman Ozdaglar, and Alireza Tahbaz-Salehi.2012. “The Network Origins of Aggregate Fluctuations.”Econometrica, 80(5): 1977–2016. Acharya, Viral V., Lasse H. Pedersen, Thomas Philippon, and Matthew Richardson.2017. “Measuring Systemic Risk.”The Review of Financial Studies, 30(1): 2–47. Adrian, Tobias, and Markus K. Brunn...

    2012

  2. [2]

    The Pareto Distribution and the Cobb–Douglas Production Function in Activity Analysis

    Houthakker, H. S.1955. “The Pareto Distribution and the Cobb–Douglas Production Function in Activity Analysis.”Review of Economic Studies, 23(1): 27–31. Hugot, Jules, and Camilo Umana-Dajud.2017. “Trade Costs and the Suez and Panama Canals.” Vniversitas Econ´omica, 0(0): 1–33. Hummels, David.2007. “Transportation Costs and International Trade in the Secon...

    1955

  3. [3]

    Red Sea Attacks Disrupt Global Trade

    Kamali, Parisa, Robin Koepke, Alessandra Sozzi, and Jasper Verschuur.2024. “Red Sea Attacks Disrupt Global Trade.” IMF Blog. K¨anzig, Diego R., and Ramya Raghavan.2026. “The Macroeconomic Effects of Supply Chain Shocks: Evidence from Global Shipping Disruptions.” Working paper. Keohane, Robert O., and Joseph S. Nye.1977.Power and Interdependence: World Po...

    2024

  4. [4]

    Disruption of Maritime Trade Chokepoints and the Global LNG Trade: An Agent-Based Modeling Approach

    Mahan, Alfred Thayer.1890.The Influence of Sea Power upon History, 1660–1783.Boston:Little, Brown. Meza, Abel, Ibrahim Ari, Mohammed Al-Sada, and Muammer Koc ¸.2022. “Disruption of Maritime Trade Chokepoints and the Global LNG Trade: An Agent-Based Modeling Approach.”Maritime Transport Research, 3: 100071. Notteboom, Theo, and Jean-Paul Rodrigue.2021. “Bl...

    2022

  5. [5]

    The State of Commodity Depen- dence 2023

    United Nations Conference on Trade and Development.2023. “The State of Commodity Depen- dence 2023.” United Nations, Geneva. United Nations Conference on Trade and Development.2024. “Navigating Troubled Waters: Impact to Global Trade of Disruption of Shipping Routes in the Red Sea, Black Sea and Panama Canal.” UNCTAD Rapid Assessment, UNCTAD/OSG/INF/2024/...

    2023

  6. [6]

    Fix a disrupted sector-s block, writeB= [ bzij ]σ(i)=s for its (flow) kernel,r for the row targets (6), andcfor the column targets (5)

    A.1 Proofs of the Section 2 Results Proof of Proposition 1(i). Fix a disrupted sector-s block, writeB= [ bzij ]σ(i)=s for its (flow) kernel,r for the row targets (6), andcfor the column targets (5). The block totals agree, P i:σ(i)=s ri = P j csj, by the footnote to (5). By the classical theory of biproportional scaling (Sinkhorn, 1964; Bacharach, 1970), ...

    1964

  7. [7]

    Fromg ≥t h, monotonicity and degree-one homogeneity of eκ(ω) give eκ(ω) i (g) ≥t eκ(ω) i (h), and eκ(ω) i (h) > 0 for every i becauseh > 0and the leakage preserves the benchmark support. Strict subhomogeneity of the outer aggregator then yields, for everyi, gi =G i eκ(ω) i (g) ≥G i teκ(ω) i (h) > t Gi eκ(ω) i (h) =t h i, contradicting attainment of the mi...

    2026

  8. [8]

    Z∗ Extended intermediate-flow matrix atG ∗ ×Nresolution

    Trade flows and benchmark network Symbol Meaning Z, z ij Benchmark intermediate-flow matrix and its entries (OECD ICIO 2025). Z∗ Extended intermediate-flow matrix atG ∗ ×Nresolution. 97 Symbol Meaning µj Buyerj’s total intermediate-input purchases, P i zij . A, a ij Input-share matrix,a ij =z ij /µ j; column-stochastic. A∗ Extended input-share matrix. Zω,...

    2025

  9. [9]

    We anchor on 2019 even though the 2025 edition carries reference years through 2022, because every later year is contaminated for our purpose

    provides the natural anchor: its 2025 edition covers 81 economies and 50 ISIC Revision 4 sectors for reference year 2019, giving a 4,050×4,050 flow matrix with well-documented accounting consistency. We anchor on 2019 even though the 2025 edition carries reference years through 2022, because every later year is contaminated for our purpose. The benchmark ...

    2025

  10. [10]

    The OECD table supplies the correct levels, and GTAP supplies the within-ROW shares used to disaggregate each ROW cell

    at a geographic resolution that resolves the chokepoint-relevant parts of ROW. The OECD table supplies the correct levels, and GTAP supplies the within-ROW shares used to disaggregate each ROW cell. This hybrid approach inherits the accounting quality of OECD ICIO while exploiting GTAP’s superior geographic coverage. OA.1.1 The OECD ICIO 2025 baseline Let...

    2025

  11. [11]

    102 The matrix has theG×Gblock structure: Z=   Z11 Z12 · · ·Z 1G Z21 Z22 · · ·Z 2G ... ... . . . ... ZG1 ZG2 · · ·Z GG   (OA.1.1) where each blockZ cc′ ∈R N×N + records the N 2 = 2,500 sector flows from country c to country c′. Approximately 2.8 million of the 16.4 million cells are non-zero. Sector codes follow ISIC ...

    2025

  12. [12]

    Its geographic scope individually resolves all the Gulf states, the main African economies, and most of the rest-of-world regions that OECD ICIO aggregates into ROW. The construction draws on three arrays from the GTAPbasedata.harfile: • VIWS (i, r, c): the value of imports of commodity i from source region r to destination region c, valued at world price...

    2014

  13. [13]

    A formal sensitivity check across GTAP releases is outside the scope of this paper

    and 12 (base year 2017, refined) update the levels but leave the within-block sourcing patterns close to the GTAP 10 values for the maritime-relevant economies in the seven ROW sub-aggregates. A formal sensitivity check across GTAP releases is outside the scope of this paper. We report the GTAP 10 results. The qualitative ranking of chokepoint exposures a...

    2017

  14. [14]

    We use GTAP only for within-ROW shares, not for levels, so the critical assumption is that the relative bilateral distribution of flows across ROW blocks is sufficiently stable between 2014 and

    2014

  15. [15]

    The resulting extended matrixZ ∗ has dimensions G∗N×G ∗N = 4,350 ×4,350, with G∗ = 87 economies at N = 50 sectors

    This is 107 standard practice in the IOT extension literature (Tukker et al., 2013): production network shares within a geographic sub-aggregate typically change slowly relative to the aggregate level. The resulting extended matrixZ ∗ has dimensions G∗N×G ∗N = 4,350 ×4,350, with G∗ = 87 economies at N = 50 sectors. This matrix, together with the derived i...

    2013

  16. [16]

    OA.2.1 What the reroutability matrix encodes For each maritime chokepoint k, the reroutability matrixE k = [ek,cc ′] is a G×G matrix over the G = 81 economies of the OECD ICIO 2025 database (80 economies plus a Rest-of-the-World aggregate). The entry ek,cc ′ ∈ [0,1] is the share of a benchmark intermediate-input flow from country c to country c′ that is l...

    2025

  17. [17]

    A short route loses more to a given detour than a long one, since e falls as D rises, and any flow trapped behind a terminal gate loses everything

    The rule delivers the heterogeneity the 108 paper requires. A short route loses more to a given detour than a long one, since e falls as D rises, and any flow trapped behind a terminal gate loses everything. OA.2.3 From distance to lost share A voyage detour increases the cost of moving cargo through several channels. Fuel and crew time scale with distanc...

    2009

  18. [18]

    On a block whose support carries zeros the projective diameter is infinite and the Birkhoffbound is uninformative. There, convergence follows from the strict Gale–Hoffman feasibility established in the proof of Proposition 1(i) together with the monotone divergence-descent property of the alternating scaling (Bacharach, 1970). And the observed rate follow...

    1970

  19. [19]

    The balancing also has the decision-theoretic foundation invoked in Section 2.2: it is the minimal revision of the disrupted trade pattern consistent with the two capacity caps. Remark2 (The balancing as a minimum-divergence adjustment).On each disrupted sector block, the intended-order matrix of Definition 1 is the unique solution of min B≥0 X i,j bij lo...

    1940

  20. [20]

    It is orthogonal to the reroutability e of Section 2.1, thearound-passagesalvage a longer voyage recovers: e sends a shipment around the obstacle, δ lets it through despite the closure. The empirical counterpart of δ is accordingly a ratio: the transit that continues through a passage during a closure, divided by its pre-episode benchmark, read at the ste...

    2026

  21. [21]

    Strauss Center for Inter- national Security and Law,

    Tanker War, 1984–88 Hormuz / Gulf war-risk, sustained 0.8–1.0∼ 450 ship attacks, strait never closed, flow main- tained → LB (Robert S. Strauss Center for Inter- national Security and Law,

    1984

  22. [22]

    Malacca listing, 2005–06 Malacca named risk, no closure ≈1 JWC war-risk listing, traffic uninterrupted; defi- nitional (Ho,

    2005

  23. [23]

    Gulf of Oman attacks, 2019 Hormuz war-risk, brief 0.85–1.0 premia×10, flow maintained→LB (Tan,

    2019

  24. [24]

    Ever Given, Mar 2021 Suez Canal physical, total (6 d) ≈0 † full blockage for six days; definitional (Notte- boom and Rodrigue,

    2021

  25. [25]

    friction 0.85–0.95 ∼ 20–27 tankers queued days, then cleared → inferred (Insurance Journal,

    Bosphorus insurance rule, 2022 Turkish Straits admin. friction 0.85–0.95 ∼ 20–27 tankers queued days, then cleared → inferred (Insurance Journal,

    2022

  26. [26]

    ∼ 50 Mt/yr benchmark → volume ratio (United Nations,

    Black Sea blockade →BSGI, 2022–23 Black Sea / Turkish blockade then corridor 0→0.6 32.9 Mt moved vs. ∼ 50 Mt/yr benchmark → volume ratio (United Nations,

    2022

  27. [27]

    Panama drought, 2023–24 Panama Canal capacity rationing‡ n/a transits cut 36 → 22/day; informs the capacity ceiling, notδ(Seatrade Maritime News,

    2023

  28. [28]

    Red Sea (boxes), 2023–24 BAM / Suez war-risk closure 0.05–0.15 container transits − ∼90% at trough → AIS resid- ual (United Nations Conference on Trade and Development, 2024; Lloyd’s List Intelligence,

    2023

  29. [29]

    Energy Information Administration, 2025b) Determined Hormuz closure Hormuz counterfactual unident.S no precedent; history gives a lower bound only (U.S

    Red Sea (crude & products), 2023–24 BAM war-risk closure 0.40–0.55 oil flow −55–60%, 55% of tankers persist → vol- ume/share residual (Lloyd’s List Intelligence, 2024; U.S. Energy Information Administration, 2025b) Determined Hormuz closure Hormuz counterfactual unident.S no precedent; history gives a lower bound only (U.S. Energy Information Administrati...

    2023

  30. [30]

    The firm-level natural-experiment evidence points the same way and is sharper about the horizon

    Atalay separately estimates the elasticity between value added and the intermediate aggregate at roughly 0.2, but that is a different nest from the one the bundle factor penalizes, which is substitution among the sectoral inputs themselves, and we do not use it. The firm-level natural-experiment evidence points the same way and is sharper about the horizo...

    2019

  31. [31]

    short-run network-macro benchmark (calibration) across-sector [−4,−1] Bonadio et al. (2021); Goldberg and Reed (2023) COVID and trade-disruption assessments (calibration) across-sector [−4,−1] Read at face value the structural evidence is concentrated well below the Cobb–Douglas benchmark: the directly estimated, finely disaggregated elasticities sit near...

    2021

  32. [32]

    the closure

    It sets out the footprint regression behind Section 6.1 and reports its full battery. It backs the coherence check of Section 6.2 with the full parameter surface it summarizes and the buffer accounting that relates the model’s unbuffered loss to the realized aggregate. And it records two negative results, on widening the cross-section and on the canonical...

    2023

  33. [33]

    The results begin with incidence and the specification ladder

    computed for the main acute-window estimate leave the conclusion unchanged (null-imposed Rademacher bootstrap, p= 0.02). The results begin with incidence and the specification ladder. We estimate equation(OA.6.1) on the OECD ten-industry value-added panel of forty-two economies over 2022Q1–2024Q4. The coefficient on propagated exposure is negative through...

    2023

  34. [34]

    0.72) against a post-onset average of −0.26, and reaches its trough about a year after onset

    has statistically flat leads, a pre-onset average of +0.79 (s.e. 0.72) against a post-onset average of −0.26, and reaches its trough about a year after onset. Estimating without the United States returns −0.68 (t = −3.3), and adding it back (recovered from the BEA value-added-by-industry series on FRED) moves the estimate only to −0.66, so the result is n...

    2023

  35. [35]

    Table OA.6.2 reports the two surfaces on the (τ, ρ) rectangle at the baseline leakageδ= 0.10. 140 (a) against all gate footprints (b) against orthogonal gates Figure OA.6.2:The full network-footprint horse-race.Each panel enters several gates’ propa- gated footprints in the same regression, inside the full fixed effects; squares are single-footprint estim...

    2024

  36. [36]

    significant

    All three are named explicitly in the external assessments of Section 6.2. Writing the realized aggregate as b LW with a buffer-and-horizon attenuationb∈ (0,1], the modest realized world fall, well under a tenth of a percent, corresponds to b≈ 0.25–0.4. This is of a piece with the roughly seven-tenths deflation of the effective crude-oil friction that nin...

    2021