The gap
The Asian Development Bank measures a metric called the trade finance gap: the dollar volume of trade-finance applications submitted to banks each year and rejected. In 2022, that gap was $2.5 trillion. Up from $1.7 trillion in 2020. Up from $1.5 trillion in 2018. It is growing, not closing.
For context: total global trade in 2024 was $33 trillion (UNCTAD). The gap is roughly 7.5% of all trade by dollar value, somewhere between Germany’s annual GDP and India’s. And 80 to 90% of world trade depends on trade finance in some form (WTO), so the gap doesn’t represent edge cases. It is the financing the system cannot provide for what would otherwise be routine commerce.
This is not a market clearing at a higher price. It is a market failure. The demand is documented. The capital exists. What doesn’t exist is infrastructure that lets a bank underwrite a $400,000 trade between a Vietnamese textile factory and a Spanish distributor at a cost that makes the deal economic for everyone involved. Paper-based diligence on a low-margin SME trade doesn’t pencil out, so the bank says no. The trade either dies, or the SME runs it on open-account terms with no recourse and absorbs the entire counterparty risk themselves.
Who gets locked out
The system doesn’t reject randomly. According to the WTO’s 2024 World Trade Report, multinational corporations have their trade finance applications approved 93% of the time. Small and medium enterprises have theirs approved 50% of the time.
That is not friction. That is a wall. The World Bank estimates 65 million MSMEs are credit-constrained globally. These are businesses with documented demand, documented orders, documented operating histories, who cannot access the financial instruments the system was built to provide. The ICC Banking Commission survey puts a sharper edge on it: 53% of all rejected trade-finance transactions worldwide are SME applications. The cost of bank diligence on a $400,000 trade is roughly the same as diligence on a $40 million one, so banks concentrate where the unit economics work.
For frontier-market SMEs the asymmetry compounds. In Côte d’Ivoire and Senegal, smaller businesses pay 7 to 9 percent premium over refinancing rates to access trade finance, against 4 to 5 percent for large companies in the same markets (WTO 2024).
This is the trap EDMA Group lived in. We were a glove distributor sourcing from Asia and selling to the US and Europe, exactly the SME profile the data describes. The 50% rejection rate wasn’t a statistic to us; it was the daily reality of trying to finance trades that would have been routine for a Fortune-500 buyer with the same counterparties. The protocol thesis didn’t come from reading the WTO report. It came from being in the rejected half.
The first-pass break
For the trades that do get an LC issued, the discrepancy problem starts. Across global LC traffic, 60 to 75 percent of document presentations are refused by banks on first review (ICC Banking Commission; some recent studies cite 65 to 80%). A port code that doesn’t match the LC text, a certificate of origin dated one day off, a packing list with a typo in the carton count. Any of those triggers a refusal under UCP 600’s strict-compliance standard.
The Uniform Customs and Practice for Documentary Credits, version 600, was published by the ICC in 2007 to harmonise and modernise LC examination practice. The 2014 ICC Trade Survey reported 31% of banks were seeing an increase in the rate of spurious discrepancies, refusals over technicalities defensible under the rules but commercially irrelevant. The 2015 survey reported 34% saw an increase in first-pass refusal rates overall.
18 years after UCP 600, the discrepancy rate has not measurably improved.
The breakdown of why documents fail reads exactly like a system built around inconsistent paper artifacts: inconsistent dates (~25% of discrepancies), description mismatch (~22%, a comma in the goods description that doesn’t match the LC), late shipment (~18%), late presentation (~15%), incorrect document version (~12%).
None of these are fraud signals. They are typographical and clerical errors that the system treats as commercially equivalent to fraud, because that is how strict compliance works. The seller’s $400K cargo has shipped, the goods are at the port, and the bank refuses payment because the bill of lading says Hamburg while the LC says Port of Hamburg.
The paper jam
Every day, 4 billion pieces of paper circulate between the physical and financial supply chains of global trade (ICC + MonetaGo). Bills of lading, invoices, certificates of origin, packing lists, insurance certificates, inspection reports, dock receipts, warehouse receipts, customs forms.
A single cross-border transaction averages 36 different documents and 240 copies of those documents, couriered, scanned, faxed, re-printed, re-signed, re-stamped, archived in physical files.
The carbon math is grim. Roughly 4 kilograms of CO₂ per paper-based transaction from production, transport, and logistics. Multiply by hundreds of millions of cross-border transactions per year and trade finance is one of the highest per-transaction carbon footprints in the financial system, and the only major one that is almost entirely structural rather than productive. Paper that exists because the law requires paper, not because the paper does any work that data couldn’t do.
The compliance cost is worse. The ICC estimated in 2017 that eliminating paper from trade transactions would reduce compliance costs by up to 30%. That is a 30% structural overhead that exists only because every party in the chain has to manually re-verify what every previous party already verified, because there is no shared evidence layer they all trust.
The legal infrastructure barely exists. As of 2024, only three major jurisdictions have laws that recognise digital trade documents: the UK (under the Electronic Trade Documents Act 2023), Singapore, and the UAE as legally equivalent to paper. Everywhere else, an electronic bill of lading is legally nullable in a dispute. Trade finance has reached the cloud era running paper protocols because the law that governs the documents was written for ships that took weeks to cross oceans.
- ~36 documents per transaction on average (BL, COO, packing list, customs forms, invoices, inspection reports, insurance certs, dock and warehouse receipts)
- ~240 copies of those documents per transaction (couriered, scanned, faxed, re-printed, re-signed, archived in physical files)
- Sources: ICC + MonetaGo joint research; ICC Banking Commission
- ~30% structural compliance cost overhead from paper (ICC 2017 estimate of savings from full digitisation)
- 60–75% of LC document presentations refused on first review (ICC Banking Commission, UCP 600 strict-compliance regime)
- ~4 kg CO₂ per paper-based transaction from production, transport, logistics. Trade finance is one of the highest per-transaction carbon footprints in the financial system
- United Kingdom · Electronic Trade Documents Act 2023
- Singapore · Electronic Transactions Act amendments
- United Arab Emirates · recent dedicated framework
- Everywhere else: an electronic bill of lading is legally nullable in a dispute. Trade finance has reached the cloud era still running paper protocols because the law that governs the documents was written for ships that took weeks to cross oceans
The fraud bill
The paper system enables fraud that a verified-evidence system would catch.
The ICC and MonetaGo estimated in a 2022 joint report that $2.5 billion per year in trade-finance fraud losses are realised by financiers globally. This is the annual write-down banks take because forged documents got through their LC checks before the fraud was discovered. That figure is the realised losses number, not exposure or fraud volume. Actual fraud volume runs higher; recoveries and corrected positions absorb some of it.
The marquee cases of the last 15 years all show the same structural pattern. 2009, Saad Group (Saudi Arabia): multi-billion collapse, some banks lost ~$500 million each on forged trade documents across multiple bank LCs. 2014, Qingdao metals scandal (China): the same metals pledged as collateral to multiple banks at once. Duplicate-financing fraud at scale, possible only because no shared registry existed for warehouse receipts. 2017, LME warehouse duplicates: same pattern with metals at London Metal Exchange-listed warehouses. 2020, Hin Leong Trading (Singapore): $3.5 billion combined bank exposure. Founder OK Lim sentenced to 17 years in 2024. PwC’s interim judicial-manager report documented systematic forgery and duplicate financing: fake sales, circular trades between affiliated companies, bills of lading signed by Hin Leong employees impersonating third parties. Banks involved: HSBC, ABN Amro, Société Générale, Standard Chartered, OCBC, Crédit Agricole. 2020, Agritrade International (Singapore): 20+ banks, hundreds of millions in losses, similar forgery pattern. 2021, Greensill Capital: collapsed on questionable receivables, severely damaging trust in supply-chain finance products and taking down a substantial portion of Credit Suisse’s exposure.
The common pattern in every case is paper documents being trusted because no party can verify them in real time against the underlying physical event. The bill of lading says cargo X loaded on vessel Y on date Z. The bank that funds the LC has no way to verify cargo X is the same cargo X another bank funded last week, or that the bill of lading wasn’t issued by the same employee who signed three other competing bills.
This is what document-based settlement actually means when there is no verification layer: the documents are the only evidence, and the documents can be lied about.
The de-risking cascade
Banks didn’t sit with the fraud risk. They de-risked.
Between 2011 and 2024, correspondent banking relationships in the Pacific Islands dropped 60% (World Bank), double the world average for the period. After Hin Leong, ABN Amro withdrew from trade and commodity finance entirely. Société Générale and BNP Paribas consolidated their trade-finance activities and stopped onboarding new clients in higher-risk corridors. The Bank for International Settlements has documented that rising compliance costs and regulatory uncertainty are the primary reasons banks shed correspondent relationships.
A 2024 CEPR study traced what this means on the ground. When a local respondent bank in an emerging market loses its correspondent-banking access, the firms that bank with it start exporting less. Their domestic sales rise to partially offset, but only partially. Total revenues shrink. Employment shrinks. The capital is theoretically still available somewhere in the global financial system; the connection to it disappears.
This is the second-order effect of the LC fraud problem. The frauds were paper-based. The banks couldn’t detect them. The banks de-risked the corridors. The SMEs in those corridors got cut out. The trade finance gap widened. The system is not just bad at financing small trades. Its failure mode actively makes the gap worse over time.
- Fraud1Paper-based fraud realised by banksForged documents get through LC checks before the fraud is discovered. ~$2.5B per year in trade-finance fraud losses are realised globally (ICC + MonetaGo 2022). The marquee cases (Saad, Qingdao, LME warehouse duplicates, Hin Leong, Agritrade, Greensill) all share one pattern: paper documents trusted because no party can verify them in real time against the underlying physical event.Bank exposure >$14B across just the six marquee cases of the last 15 years.
- Regulation2Compliance costs surge, regulators tightenBasel III & IV raise capital requirements on trade-finance positions. Sanctions regimes expand, AML/KYC documentation requirements multiply, regulatory uncertainty grows. The Bank for International Settlements documents that rising compliance costs and regulatory uncertainty are the primary reasons banks shed correspondent relationships.BIS Quarterly Review repeatedly flags this as the dominant factor in correspondent-banking contraction.
- De-risk3Correspondent relationships cutBanks exit the corridors that produced the fraud. Pacific Islands correspondent banking dropped 60% between 2011 and 2024 (World Bank), double the world average. ABN Amro withdrew from trade and commodity finance entirely after Hin Leong. Société Générale and BNP Paribas consolidated their trade-finance activities and stopped onboarding new clients in higher-risk corridors.World Bank data Pacific Islands −60% (2011–2024); broader emerging-market drops 20–40% by region.
- Cascade4Local banks lose cross-border accessWhen a local respondent bank in an emerging market loses its correspondent-banking access, its clients (SME importers and exporters) lose access to USD/EUR clearing, LC issuance, trade finance, and FX hedging. The capital is theoretically still available somewhere in the global financial system. The connection to it disappears.CEPR 2024 study traced this empirically across emerging-market respondent banks and their corporate clients.
- Gap5SMEs export less, gap widensFirms that bank with cut-off respondents start exporting less. Domestic sales rise to partially offset, but only partially. Total revenues shrink. Employment shrinks. The $2.5T trade finance gap widens. The system's failure mode actively makes the gap worse over time, rather than self-correcting.Net effect SME exports contract; the gap that's been stuck since 2023 grows structurally rather than from a single shock.
The opportunity
The upside has been quantified by the system’s own institutions.
McKinsey estimates that full adoption of electronic bills of lading alone would unlock $40 billion in additional global trade volume. That is trade that doesn’t happen today because the paper friction makes it uneconomic. The container shipping lines alone would save $6.5 billion per year in direct costs from the same shift.
The WTO estimates that implementing its Trade Facilitation Agreement (paperless customs, electronic data submission, digital signatures) would increase global trade by $750 billion to $1 trillion per year.
The UK Electronic Trade Documents Act of 2023 projected 75% reduction in transaction costs and deals processing in 25% of the time for businesses that adopt digital trade documents. Singapore and the UAE have passed similar legislation. The ICC’s Digital Standards Initiative is targeting 60 to 80% of world trade digitised by 2026.
The numbers are there. The legal frameworks are slowly arriving. What is missing is the infrastructure that makes the data verifiable end-to-end: operational ground both sides of a trade agree on, a marketplace where financiers can price risk on real signals, and a settlement rail that releases payment on verified milestones instead of paper presentations.
That is the three-layer alternative. TradeOS runs the operation. Global Trade Marketplace matches deals with capital. Settlement releases money on cryptographic proof of milestones. We didn’t design this to capture the $40 billion McKinsey identified. We designed it because the system that produces those numbers was actively destroying our business. The disruption is what happens when something better exists.
