Future Trade Systems

The trading system has two layers and they are usually discussed as one. The flows layer is what actually crosses borders: values, volumes, partners, the length of a supply chain. The rules layer is the machinery that says what a state may do at its border and what happens when it does something else: tariff bindings, preference schemes, rules of origin, export controls, and an adjudication system with an appellate stage. Almost every confident sentence written about the future of trade mixes the two.

Separated, they give opposite answers to the same framing. On flows, trade is at record value, is being reallocated rather than reduced, and the fragmentation that exists is confined to two hotspots and is roughly an order of magnitude smaller than the Cold War's. On rules, the enforcement layer of the multilateral system has genuinely stopped working, is measurably less used, and the replacement covers about a third of the membership and has almost no case law. The rules half of the framing is the well-evidenced half, and a brief that treats the two as one story will get the flows half wrong. Lead with the asymmetry; everything else in this brief sits on one side of it.

Established The aggregate first, because it is the fact most commentary is written against. UNCTAD's April 2026 update puts 2025 global trade in goods and services at about $35 trillion, an increase of $2.5 trillion on the year — a record, with growth continuing into early 2026. The January 2026 update adds the structure: services are now 27% of global trade and grew about 9% in 2025, far outpacing goods; South–South merchandise exports reached $6.8 trillion in 2025 against $0.5 trillion in 1995; 57% of developing-country exports now go to other developing markets; and nearly two-thirds of global trade takes place within value chains. Frontier UNCTAD also counts roughly 18,000 new discriminatory trade measures since 2020, and that number is flagged weaker than the others because it depends entirely on what counts as a measure. Established On volumes rather than values, the WTO's October 2025 outlook as reported in the trade press had goods trade volume growing 4.9% year-on-year in the first half of 2025, a 2025 forecast of +2.4%, and a 2026 forecast cut to +0.5% from an August interim projection of 1.8%. Record values and a sharply slowing volume forecast are both true and are different claims.

Established One number this brief does not print, because it could not be verified. The trade-to-GDP ratio is the single most-requested statistic for this subject and neither the World Bank indicator page nor the WTO statistical releases were reachable in the research pass behind this brief. No openness ratio is stated here. Frontier The qualitative claim it is usually deployed to make — that world trade openness plateaued after 2008 rather than collapsing — is carried instead on verified trade values and on the decomposition below, which does not need the ratio. Saying which number is missing and why is more useful than an unverified one.

Frontier The sharpest empirical result in the file is a decomposition, and it goes the opposite way to the commentary. Sébastien Jean's December 2024 paper measures the fragmentation signature and then removes the two hotspots. Including them, West-to-East flows lagged 2017 by 12% and East-to-West by 23%. Excluding Russia and the US–China pair, East-to-West flows outperformed the rest of world trade by 6% and West-to-East by 10%. His summary: outside hotspots, no sign left of geoeconomic fragmentation. Frontier He also finds that the average distance travelled by one dollar of manufactured trade increased — chains got longer, not shorter — and concludes that no widespread nearshoring trend is visible. Established The two hotspots are real and large: Russia's exports to sanctioning countries divided by 10 and imports by 7; Chinese imports from the US lagging third-country trade by 32% and US imports from China by 43%. Fragmentation is not absent. It is local.

Frontier Jean's alternative explanation for what the data actually show is a different story entirely and it is not about geopolitics. China's manufactured-goods surplus rose from about 8% to 11% of the world total since early 2023 — his words are massive, sudden and across the board — with Chinese exports outperforming world trade by 15% since 2019 while Chinese imports underperformed by 6%. Frontier On this reading the dominant trend in the trade data is a policy-driven manufacturing surge in one country, and bloc formation is a small residual on top of it.

Frontier The best-identified result on the other side comes from the institution most associated with warning about fragmentation, and it quantifies how small the effect is. Gopinath and co-authors, using gravity models with PPML and fully saturated fixed effects, find that after the first quarter of 2022 trade between geopolitically distant blocs declined 12% more than within-bloc trade and FDI projects between blocs declined 20% more. Established For scale, in their own paper: during the Cold War, trade between rival blocs fell about 67% relative to within-bloc trade and trade with nonaligned economies about 37%, and the authors state that current fragmentation is an order of magnitude smaller. Import reallocation rose 15% post-pandemic overall and 40% for advanced economies against a 2003–2019 baseline.

Frontier Their connector-country result is the one to carry, because it is a sign reversal rather than a magnitude. Across nonaligned economies, a 1% increase in US import share is associated with a 1.63% higher Chinese export share comparing 2013–17 with 2018–23, and with a 0.72 percentage-point increase in Chinese FDI share per percentage point of US import gain. In the Cold War the equivalent slope was −0.026 and statistically insignificant. Established In the Cold War, a nonaligned country that traded more with one bloc did not thereby trade more with the other; today it does. Whatever this is, it is not a Cold War, and the arithmetic saying so is published by the IMF. Interest running against the finding raises its weight.

Frontier Three independent studies looked for reshoring in the trade data and none of them found it. Alfaro and Chor: China's share of US imports peaked at 21.6% in 2017 and fell to 16.5% in 2022 — while the dollar value rose, from $505.1bn to $531.3bn. Vietnam gained nearly 2 percentage points; Chinese FDI reached about 7% of Vietnam's inbound investment by 2021; Chinese manufacturing FDI in Mexico rose fivefold from $31.6m in 2017 to $151.5m in 2022, concentrated in computer equipment and motor-vehicle parts; US semiconductor employment rose 1.9% over 2017–22. Established The price of the exercise is in the same paper: a 5-percentage-point fall in China's import share is associated with roughly 9.8% higher import prices from Vietnam and 3.2% from Mexico.

Established Freund, Mattoo, Mulabdic and Ruta make the same finding sharper and add the result that kills the de-risking story. US imports of tariffed Chinese goods grew 40% slower than imports from other partners; tariffed-goods imports from China were 14% lower in 2022 than 2017 while imports of the same products from the rest of the world rose 48%. Gainers: Vietnam 1.9pp, Taiwan 1.0, Canada 0.75, Mexico 0.64, India 0.57, Korea 0.53. Frontier The gainers are the countries most integrated with China: moving from the 25th to the 75th percentile of intra-industry trade with China is associated with about 3pp higher export growth to the US across tariffed products and 4.5pp for strategic goods, and at HS6 level a 10-percentage-point increase in imports from China is associated with 0.4pp higher export growth to the United States. Frontier And concentration rose rather than fell: for over 70% of products, at least 75% of China's lost market share went to a single supplier; China remained the top supplier of US imported goods in 2022 overall and a top-three supplier in over 60% of the products where its share fell most sharply. The policy aimed at diversification produced concentration.

Established Now the rules layer, where the framing is right. Appellate Body appointments were blocked from 2017; the body lost quorum when two members' terms expired on 10 December 2019 and has been completely non-functional since 30 November 2020. Hopewell supplies the measured consequences: 24 panel rulings appealed into the void as of November 2024, 64% of panel reports issued 2020–2023 appealed into the void, the US filing 9 of the 24 and India 5, with developing countries now accounting for over half; and new disputes falling from an average of 19 a year before the collapse to 7 a year after — roughly a third of the previous rate. Named non-compliance cases include Indonesia's nickel export ban and India's SEZ subsidies. Established The replacement, the Multi-Party Interim Appeal Arbitration Arrangement, began with 19 founding members in April 2020 and reached 57 members when the UK acceded on 25 June 2025, covering about a third of the WTO membership but 18 of the top 30 dispute users. It has issued two awards in six years — Colombia–Frozen Fries in December 2022 and China–IP Enforcement in July 2025. That is the number that tells the story: a mechanism with growing membership and almost no case law.

Frontier Why the United States blocked it is usually skipped and the brief is worse without it. The substantive American objections, as recorded by a former WTO deputy director-general who advocates restoring the system, are doctrinal: the Appellate Body's prohibition of zeroing in dumping-margin calculation; its narrowing of the public-body definition, which constrains countervailing duties against state-owned enterprises; its limits on safeguards; and its constraints on non-market-economy instruments. The American position is that the Appellate Body removed trade remedies Congress had required in order to approve the agreements at all. Established The same source records that over 500 cases were heard in the WTO's first twenty years and that two-thirds of members now lack binding dispute settlement. Frontier The paralysis is a doctrinal dispute about the proper scope of adjudication, not simple vandalism, and any account of restoring it that does not engage the objections is describing a negotiation that will not happen.

Frontier The organising question is which layer the next decade is actually about, and there are two coherent readings of the flows evidence that nobody has separated. (a) The world is fragmenting into blocs — the IMF research framing and most policy commentary; evidence is the −12% bloc-to-bloc trade and −20% FDI effect after 2022Q1; settled by whether the gap widens or stabilises through 2030. (b) Fragmentation is confined to two hotspots and is otherwise absent — Jean's decomposition, where removing Russia and the US–China pair leaves East–West flows outperforming world trade by 6–10%; settled by extending the decomposition through 2026 with the newer tariff regime included. Established A third position needs no further evidence at all: quantitatively this is not a Cold War. The bloc coefficient is 12% against 67%, and the connector slope is +1.63 today against −0.026 and insignificant then. That comparison is arithmetic on published estimates and is treated here as established.

Frontier Fourth, and gaining ground: the real story is Chinese manufacturing overcapacity rather than geopolitics. The surplus moving from 8% to 11% of the world total in about two years, exports outperforming world trade by 15% since 2019 and imports underperforming by 6%, is a larger movement in the data than the measured bloc effect. Speculative Whether it persists depends on Chinese domestic demand policy, which is not a trade variable, and this is the position most likely to look obvious in retrospect and is currently a minority reading.

Frontier Fifth, friend-shoring is rerouting rather than reshoring — well supported, with one honest weakness. Chinese dollar exports to the US rose while the share fell; the gainers are China's closest supply-chain partners; the average distance per dollar of manufactured trade rose. Speculative Sixth, the sharper version — that rerouting is largely transshipment or thin processing — is inference, not measurement. It rests on a product-level correlation: at HS6 level, 10 percentage points more imports from China is associated with 0.4pp higher export growth to the United States. That is consistent with transshipment and equally consistent with genuine assembly, and nobody in the published record has firm-level customs data linking a specific Chinese input to a specific US-bound export.

Frontier On the rules side the live disagreement is whether the system is being rebuilt or abandoned, and the same facts support both. Rebuilt around the WTO rather than within it: the interim appeal arrangement has grown to 57 members including most heavy dispute users. Abandoned: new disputes fell from 19 a year to 7, 64% of recent panel reports were appealed into a void, and the interim mechanism has produced two awards in six years. Speculative These are compatible if the arrangement is a standby facility that members join for insurance and do not use, which is a different institution from the one its advocates describe. Whether its caseload rises materially by 2028 is the discriminating observation and it is a few years away.

Speculative A widely repeated claim fails outright when the denominator is fixed, and it is worth stating as a live position because it is still asserted. “Preference utilisation rates are startlingly low” is not supported once utilisation is measured against goods with a positive most-favoured-nation tariff — the goods where a preference is worth anything. Established On Japan's import side, average utilisation is above 70%, with Thailand at 92%, Vietnam, Malaysia, Indonesia, Brunei and the Philippines in a 72–92% band and Singapore at 72%. Myanmar, Cambodia and Laos sit at 5–17% and the explanation is that general preferences already deliver the benefit, so the agreement route is redundant. Frontier The position that survives is that rules of origin, not tariffs, are the binding constraint on regional agreements: larger preference margins raise utilisation at 1% significance, and restrictive origin rules lower it, with combined tariff-classification-plus-value-added rules and wholly-obtained rules doing the most damage and change-of-chapter more restrictive than change-of-heading.

Frontier Export controls are the instrument with the best-identified cost estimate and the worst-identified benefit estimate. On the imposing side, affected US suppliers showed a −2.5% cumulative abnormal return within 20 days of announcement, an average market-capitalisation loss of $857 million per affected firm and $130 billion in aggregate, revenue down 8.6% and employment down 6.6%, with cash flow falling by 20% and EBIT by 25% of average treated firm value — and no significant capital-expenditure reduction, so firms cut labour rather than investment. On the target side, Chinese firms substituted toward domestic and non-US suppliers and suppliers in allied countries gained revenue. Speculative The benefit — measurable delay in the target's capability convergence — has never been estimated by anyone. The policy is being run on a cost estimate with no matching benefit estimate, which is an unusual position for an instrument this consequential.

Speculative Two minority framings deserve stating rather than dropping. One holds that services and digital trade are the real growth story and the fragmentation debate is an argument about the shrinking half: services are 27% of global trade and grew 9% in 2025 against goods volume forecasts near zero. Its weakness is that services trade measurement is far poorer than goods measurement, so the comparison is between a well-measured quantity and a badly measured one. Frontier The other holds that the measurement itself is the problem — gross bilateral flows cannot answer the question being asked of them when nearly two-thirds of trade occurs within value chains — and would be settled only by universal adoption of value-added trade accounts at usable frequency, which does not exist and is not close.

Established Every reallocation number in this subject is computed on gross bilateral flows, and gross flows double-count intermediates. A chain that gains a stage looks like more trade even when value added is unchanged, and UNCTAD's own statement that nearly two-thirds of global trade takes place within value chains is the size of the problem rather than a decoration on it. Frontier That is not a reason to discard the numbers; it is a reason to state what they measure. The direction is established — chains lengthened, connectors intermediate, the China link persists — and the magnitude of true value-added reallocation is not.

Frontier The transshipment inference is the specific place this bites. The 0.4-percentage-point HS6 correlation is the entire empirical basis for the claim that rerouting is thin processing, and it is a correlation between two aggregates. Speculative Rules-of-origin audits and customs-level value-added data would separate transshipment from genuine assembly, and both exist inside national customs administrations. Until someone uses them, rerouting is a well-supported inference and not a measurement, and this brief says inferred wherever it says it.

Frontier The count of discriminatory measures is the weakest widely quoted number in the subject. Roughly 18,000 since 2020 depends on a definition broad enough to include instruments of very different economic weight, and no published series separates them by trade coverage. Speculative A tariff line and a subsidy notification and a licensing requirement all count as one, so the series measures activity in trade policy rather than restriction of trade.

Established And one measurement was simply unavailable. The trade-to-GDP openness ratio, which almost every account of this subject leads with, could not be verified in the research pass behind this brief and is therefore absent from it. Frontier The argument is made on trade values and on the hotspot decomposition instead, and nothing in this brief depends on an openness ratio. An unverified headline number is worse than a stated gap.

Frontier The first dependency is data rather than theory: firm-level customs records linking a specific Chinese input to a specific US-bound export. Every claim about transshipment, thin processing and the true depth of rerouting waits on it, and it is held by national customs authorities in the connector countries themselves, which have the least incentive to publish it. Speculative Nothing about the analysis is hard once the data exist; the constraint is entirely one of access.

Frontier Second, value-added trade accounts at annual frequency and full country coverage. The existing inter-country input-output databases run several years behind and are constructed from national tables of uneven quality, which is why every current estimate is made on gross flows and then caveated. Speculative A two-year publication lag on value-added accounts would change what can be claimed about reallocation more than any new econometric technique.

Established Third, and unusually for this map, a legal outcome is a dependency. Whether the appellate stage is restored, replaced or abandoned determines whether the non-compliance cases now on record — a nickel export ban maintained after an adverse ruling, subsidy findings appealed into a void — are anomalies or the new baseline. Frontier That is a negotiation among member states rather than a research result, and the American objections are doctrinal and specific enough that restoration means conceding on zeroing, public-body scope, safeguards and non-market-economy instruments, or persuading the United States to drop four positions it has held across administrations of both parties.

Frontier Fourth, comparable preference-utilisation studies on the correct denominator for the European Union, the United States and the regional Asian agreement. One high-quality import-side study on one country is the entire basis for the correction offered here, and a claim about the world resting on Japan is thin. Speculative The data are held by the customs authorities administering the preferences.

Frontier The cheapest high-value study is a repeat of an existing calculation on newer data: recompute the supplier-concentration result for 2023–26. The original finding — that for over 70% of affected products a single supplier took at least three quarters of China's lost market share — is the sharpest available refutation of the de-risking story, and it is a straightforward calculation on published trade data. Speculative If concentration has since fallen, the diversification claim revives; if it has risen, the policy has an outcome nobody intended and the evidence would be unambiguous.

Frontier Second, the 2025–26 tariff round is a replication of the 2018–19 natural experiment and nobody has pre-registered the analysis. The earlier episode delivered complete pass-through, a foreign export supply elasticity indistinguishable from perfectly elastic, and relative employment declines in the industries the tariffs protected. Speculative A pre-specified replication on the newer round, with the same estimators and the same outcome set, would either confirm one of the most robust results in trade empirics on fresh data or identify the conditions under which it fails. Announcing the design before the data arrive is the whole value.

Frontier Third, extend the hotspot decomposition through 2026. Jean's result is the strongest contrarian finding in the file and it ends before the newest tariff regime. Re-running it with the later data and the same two exclusions is a small piece of work with a large amount riding on it. Speculative If the outperformance outside hotspots survives, general fragmentation is not happening; if it disappears, the hotspots have spread and the bloc reading gains its first real support.

Frontier Fourth, rules-of-origin liberalisation is a natural experiment already running. Regional cumulation provisions change the origin arithmetic for identifiable product lines on identifiable dates, and utilisation is administratively recorded on both sides. Speculative That is a clean before-and-after on the variable the utilisation literature identifies as binding, and it would test the determinants result outside the one country it was estimated on.

Speculative Fifth, and the hardest: estimate what export controls buy. The cost side is well identified down to firm-level abnormal returns. The benefit is a counterfactual capability path for the target, which requires a measure of technological convergence and a credible control group. Handwave Every public defence of the instrument asserts a delay in years without naming the measurement that would produce one, and that assertion is currently doing the work of an estimate.

Established The 2018–19 tariff episode is the cleanest natural experiment in modern trade policy and three independent teams agree on what it did. On incidence, pass-through to duty-inclusive import prices was complete, with before-duty prices showing no statistically significant decline and an estimated foreign export supply elasticity of −0.002 with a standard error of 0.05 — indistinguishable from perfectly elastic. Foreign exporters did not absorb the tariff. Established The welfare arithmetic: consumer and deadweight loss of $51bn, 0.27% of GDP, decomposed as a 15% import share times 13% of imports targeted times a 14% average tariff increase; tariff revenue of +$34.3bn in the model against $29.1bn actual; and an aggregate short-run real income loss of −$7.2bn, 0.04% of GDP.

Established The terms-of-trade line deserves separating out, because it is the one gain in the ledger and it is an artefact. The measured terms-of-trade gain was +$9.4bn, 0.05% of GDP — and only $0.5bn without retaliation. Almost the entire apparent gain comes from the export side being hit by other countries' retaliation, which depressed the prices Americans received rather than the prices they paid. Frontier A policy defended by a terms-of-trade argument was in fact collecting almost all of that gain from being retaliated against.

Established The political geography is measured and is the most quoted part of the paper for good reason. Protection was concentrated in politically competitive counties, an inverted U in Republican vote share peaking at 40–60%; retaliation fell disproportionately on heavily Republican agricultural counties; and real tradeable wages fell 1.0% on average with a standard deviation of 0.5%, with Republican counties taking 32% larger losses. Established On employment, industries more exposed to tariff increases show relative reductions in employment — the protective effect outweighed by input-cost increases and retaliation — alongside relative increases in producer prices and, at county level, higher unemployment and lower labour-force participation. Protection reduced employment in the protected sector's own economy.

Established Consumer incidence was small in aggregate and large in specific categories, and both halves matter. Prices paid by importers rose one-to-one with tariffs; aggregate consumer-price effects were small because the affected consumption share was small; and specific categories accumulated close to 6% by early 2020 — sports vehicles, motor parts, household supplies, furnishings. Chinese imports fell roughly 40%, with tariffs covering $376bn of Chinese exports, about 50% of US imports from China.

Established On the preference machinery, the determinants are the engineering, not the headline rate. Utilisation measured against goods with a positive most-favoured-nation tariff averages above 70% on Japan's import side, and the two things that move it are the size of the preference margin, which raises it at 1% significance, and the restrictiveness of origin rules, which lowers it. A low headline utilisation rate is usually a statement about most-favoured-nation tariffs already being near zero, not about a failing agreement. Established The genuinely low case is the African continental agreement and it is low for structural reasons rather than for margin reasons: 54 of 55 African Union members signed and 47 ratified, with actual trading under a guided initiative launched 7 October 2022 covering 9 countries and 96 products, described as limited to start, the first recorded transaction being Kenyan tea and car batteries to Ghana. Frontier The independent source declines to state an intra-African trade share, and the bodies that do publish one are the agreement's own promoters.

Within this map: Autonomous Supply Chains, which owns the operational layer this brief measures in aggregate; Future Ports and Shipping, where a longer chain becomes a physical cost; Global Cooperation Models, which owns the general problem of an agreement without an enforcement stage; Future Legal Systems, where adjudication without appellate review is a recurring structure; and Infrastructure Resilience, which owns the question of what a lengthened chain does to failure modes.

The producer-side instruments — export bans, processing mandates, quotas with strategic reservations — are treated as an economic policy in this category's resource slot, and appear here only as the object of the enforcement question. The materials-recovery answer to import dependence belongs to Industrial Ecology.

Outside it: gravity-model trade econometrics, which supplies every reallocation estimate here; international economic law, which supplies the doctrinal dispute behind the appellate collapse; input-output economics, which supplies the value-added accounting the whole subject is short of; and the political economy of protection, which supplies the county-level result.

Established The institutional finding is that a rules system can lose its enforcement stage while keeping everything else, and the effects are measurable within four years. Panel reports still issue. Members still litigate. And 64% of panel reports issued 2020–2023 were appealed into a void, 24 rulings sat unenforced by November 2024, and new disputes fell from 19 a year to 7. Frontier Litigation collapsed because winning stopped being worth the cost of proving it, which is what an unenforceable judgment does to a docket.

Established The replacement is an unusually clean case of an institution that exists on paper. The interim appeal arrangement grew from 19 members in April 2020 to 57 in June 2025, covering about a third of the membership but 18 of the top 30 dispute users. It has issued two awards in six years. Holdouts include the United States, India, Indonesia, Russia, South Korea and Taiwan. Frontier Membership growth and caseload are the two observable variables and they have moved in opposite directions, which is the signature of insurance rather than of use.

Established Two-thirds of WTO members now lack binding dispute settlement, on the account of a former deputy director-general who is arguing for restoration — an interested party whose interest runs toward making the loss sound large, and whose factual claims about the American objections are the more valuable half of the source. Frontier Those objections are the institutional obstacle and they are specific: zeroing, the scope of the public-body definition, the limits placed on safeguards, and the treatment of non-market economies. A restoration proposal that does not name all four is not a proposal.

Frontier Interested parties in this brief run in identifiable directions and it is load-bearing to say which. UNCTAD's institutional position is that fragmentation harms developing countries and the rules need reform in their favour, and its framing shows it while its trade values are national-accounts derived. An arbitration firm supplies the interim-mechanism record and has a professional stake in arbitration as the alternative. The IMF's institutional interest is in warning about fragmentation, which is exactly why its own “order of magnitude smaller” comparison carries extra weight. Established A finding published against the publisher's interest is the most valuable object in a file like this one, and there are two of them here.

Established The distributional facts of protection are measured and they are not the ones the policy was sold on. Real tradeable wages fell 1.0% on average, retaliation concentrated on heavily Republican agricultural counties, and Republican counties took 32% larger losses. Frontier A policy targeted at politically competitive places imposed its heaviest measured costs on the places most committed to it, which is a general feature of retaliation rather than a partisan point: the retaliating country chooses where to aim, and it aims politically.

Established Consumers paid, and the incidence was concentrated. Aggregate consumer price effects were small because the affected consumption share was small, and specific categories — motor parts, household supplies, furnishings — accumulated close to 6% by early 2020. Frontier Small in aggregate and 6% on furniture are the same finding described at two levels, and which one is the honest headline depends on whose budget is being described.

Frontier Export controls impose costs on identifiable third parties who had no part in the dispute. Affected suppliers lost 8.6% of revenue and 6.6% of employment, cutting labour rather than capital expenditure, while suppliers in allied countries gained revenue — a transfer from workers in the imposing country to firms in third countries, with the strategic benefit unmeasured. Speculative Whether that is a price worth paying is a judgement nobody can currently inform, because only one side of the ledger has ever been estimated.

Frontier And the enforcement collapse falls hardest on the members least able to retaliate without it. Developing countries now account for over half the rulings appealed into the void, and a small economy's only leverage against a large one was a binding ruling. Established That is why the interim arrangement's membership includes most heavy users while the largest economies stay out: a mechanism that binds only those who opt into it binds the weak.

Frontier The largest thing this subject is about is whether a global division of labour can persist without a binding rule that a large state can be held to. The evidence so far is that it can, for at least half a decade, at record volume: trade at about $35 trillion, two-thirds of it organised within value chains, while the enforcement layer does not function. Speculative Whether that is durable or is running on the accumulated habit of a functioning system is the question the next decade answers, and there is no precedent at this scale.

Established The Cold War comparison is the only precedent available and the numbers say it does not apply. Bloc-to-bloc trade then fell about 67% against about 12% now, and the connector relationship has flipped sign from −0.026 and insignificant to +1.63. Frontier A world in which trading more with one great power means trading more with the other is structurally different from a world of two economies, and it puts a set of medium-sized states in a position — profitable intermediation between rivals — that nobody has a theory for.

Speculative The deflationary reading deserves to be stated at full strength: the reorganisation may be almost entirely a Chinese manufacturing surge, and the geopolitical vocabulary a story told about it after the fact. A surplus moving from 8% to 11% of the world total in two years is a larger movement than any bloc effect measured, and it has a domestic industrial-policy explanation that requires no blocs at all. Handwave Choosing between that account and the fragmentation account on the current data is a matter of which residual you find more interesting, not of which fits better.

These horizons track litigation dockets, tariff schedules and statistical publication lags rather than technology:

• 10 yr: Established The observable variables are already defined: whether the interim appeal arrangement's caseload rises above two awards, whether new disputes recover from 7 a year toward the pre-collapse 19, and whether the supplier-concentration result — one supplier taking at least 75% of China's lost share in over 70% of products — holds on 2023–26 data. Frontier Expect the hotspot decomposition to be redone and to be the most consequential single number published in this subject. Speculative Expect no firm-level customs evidence on transshipment, because the countries holding it have no reason to publish it.
• 25 yr: Speculative Either value-added trade accounting becomes routine and the whole reallocation literature is rewritten on a better measure, or it does not and the field continues arguing about gross flows with the caveat attached. Speculative On rules, the plausible split is between a restored appellate stage with a narrowed standard of review and a permanent two-track system in which a third of members litigate bindingly and the rest do not. Handwave Which happens is a negotiating outcome, and the doctrinal objections have survived administrations of both American parties.
• 50 yr: Speculative If the connector-country structure persists, the durable winners of this period are medium-sized economies that intermediate between rivals, and their trade policy becomes a matter of remaining acceptable to both — a position with no modern precedent and no literature. Speculative If the hotspots spread instead, the Cold War arithmetic becomes the right comparison for the first time, and the relevant coefficients would have to move by roughly a factor of five to get there.
• 100 / 250+ yr: Handwave Beyond useful forecasting. The one observation with that reach is that trading systems have repeatedly outlasted the rule systems built to govern them and have repeatedly been reorganised by whichever economy industrialised fastest — which is a pattern read off a handful of episodes, not a base rate.

• Depends on Nothing on this map. This brief waits on no result another brief produces. Its open questions are questions about customs records, input-output accounts and a legal negotiation, all of which could be resolved without any new scientific result. No typed depends-on edge is claimed.
• Requires (not on this map) Three things this subject waits on, none of them a research result another brief produces. First, an appellate stage the United States will rejoin. The Appellate Body lost quorum on 10 December 2019 and has been non-functional since 30 November 2020; 64% of panel reports issued 2020–2023 were appealed into the void, 24 rulings sat unenforced by November 2024, new disputes fell from about 19 a year to 7, and two-thirds of members now lack binding dispute settlement. The interim appeal arrangement reached 57 members in June 2025 and has issued two awards in six years. The American objections are doctrinal and specific — the prohibition of zeroing, the narrowed public-body definition, the limits on safeguards, and the constraints on non-market-economy instruments — and any restoration requires conceding on them or persuading the United States to abandon four positions it has held across administrations of both parties. Second, value-added trade accounts published at annual frequency: nearly two-thirds of global trade takes place within value chains, every reallocation figure in this brief is computed on gross bilateral flows that double-count intermediates, and the inference that rerouting is thin processing rests entirely on a product-level correlation of 0.4 percentage points at HS6. Third, a substitute supplier that is not itself sourcing from the incumbent: for over 70% of affected products a single supplier took at least three quarters of China's lost US market share, the largest gainers are the countries most integrated with China in intra-industry trade, and China remained a top-three supplier in over 60% of the products where its share fell most sharply. Diversification policy has so far produced a longer chain through a more concentrated intermediary at a measured price premium of about 9.8% from Vietnam and 3.2% from Mexico. All three are things governments and statistical agencies could choose to supply and have not.
• Enables In principle every brief on this map that assumes a component arrives from somewhere else inherits the flows structure recorded here, and every industrial build-out assumes an import channel that the export-control record shows can be closed by one announcement. No typed enabling edge is claimed, because the relationship has never been measured: nobody has estimated what a trade-rule change does to a specific technology programme, and the transshipment inference means even the current chain structure is known only in direction.
• Adjacent Gravity-model trade econometrics, which supplies every reallocation estimate here; international economic law, which supplies the doctrinal dispute behind the appellate collapse; input-output economics, which supplies the value-added accounting the subject lacks; the political economy of protection; and within this map Autonomous Supply Chains, Global Cooperation Models and Future Ports and Shipping.

Established “World trade is shrinking.” It is at record value — about $35 trillion in 2025, up $2.5 trillion on the year — with services at 27% of the total and growing about 9%, South–South merchandise exports at $6.8 trillion against $0.5 trillion in 1995, and nearly two-thirds of trade organised within value chains. Frontier Volume growth is forecast to slow sharply, to +0.5% in 2026 from an interim projection of 1.8%. That is a different claim and it should be made in volumes, not values.

Frontier “Near-shoring and reshoring are visible in the data.” Three independent studies find rerouting with longer chains and higher prices. The average distance travelled by a dollar of manufactured trade rose. Chinese dollar exports to the United States rose while the share fell, from $505.1bn to $531.3bn. Established US semiconductor employment rose 1.9% over five years — that is the reshoring signal, and it is small.

Established “This is a new Cold War.” Quantitatively it is not, on the arithmetic of the paper most often cited for the fragmentation claim. Bloc-to-bloc trade is down about 12% relative to within-bloc trade against about 67% during the Cold War, and the connector-country slope is +1.63 today against −0.026 and insignificant then. Frontier The relationship has changed sign, not just magnitude, and a country that trades more with one great power now trades more with the other.

Speculative “Preference utilisation rates are startlingly low.” Not against the right denominator. Measured on goods carrying a positive most-favoured-nation tariff, Japan's import-side utilisation averages above 70% and reaches 92% for Thailand; the 5–17% cases are countries where general preferences already deliver the benefit. Established A low headline rate usually means a small preference margin or expensive rules of origin. Do not repeat the claim without naming the denominator. The African continental agreement is genuinely low, for its own reasons, and is the exception rather than the pattern.

Established “Tariffs are paid by the exporting country.” Pass-through to duty-inclusive import prices was complete, with a foreign export supply elasticity of −0.002 (se 0.05) — indistinguishable from perfectly elastic. Established This is as settled as trade empirics gets, and three independent teams reach it. Established “Protection protects” fails on the same episode: industries more exposed to the tariff increases showed relative employment declines, with higher county-level unemployment and lower participation.

Established “At least the tariffs improved the terms of trade.” The measured gain was $9.4bn with retaliation and $0.5bn without. Almost the whole of it came from other countries retaliating against American exporters, which lowered the prices Americans received. Frontier A benefit that exists only because you were retaliated against is not a benefit of the policy in any sense the argument intends.

Frontier “Transshipment through connector countries has been measured.” It has been inferred, from a correlation: 10 percentage points more imports from China at HS6 is associated with 0.4 percentage points higher export growth to the United States. Speculative That is consistent with transshipment and equally consistent with genuine assembly, and the data cannot separate them. Say inferred.

Established And one number is deliberately absent, so that its absence does not read as an oversight. No trade-to-GDP openness ratio appears anywhere in this brief. The World Bank indicator page and the WTO statistical releases were unreachable in the research pass behind it, and the figure most people would expect in the first paragraph is therefore missing on purpose. Handwave Anyone quoting an openness ratio in this debate should be able to say which vintage and which denominator, because the ratio moves several points on definitional choices alone.