By: Christoph Gugelmann
Trade finance is one of the strangest contradictions in global finance. It is short-term. It self-liquidates. It is tied to real goods moving through real ports. On the numbers, it is among the safest lending banks do: the ICC Trade Register covers more than $25 trillion of exposures and shows default rates persistently below 0.3 per cent.
And yet the Asian Development Bank puts unmet demand at roughly $2.5 trillion a year. That figure is a survey-based upper bound, but the pattern behind it is not in doubt. Some 41 per cent of SME applications get rejected, and the pain lands on small firms and emerging-market exporters. Meanwhile, the asset class is almost invisible in the securitisation markets that fund mortgages, auto loans, and credit cards at scale.
So why doesn’t capital flood in?
Start with an honest admission: the loss data cannot answer that question. The Trade Register’s figures only cover approved transactions. If banks are good at screening, low losses on the approved book are exactly what you would see even if the rejected 41 per cent were genuinely bad credit. The public data fit two opposite stories.
Story one is adverse selection. The rejected borrowers really are riskier, and cheaper origination would simply approve bad credit faster.
Story two is a unit-cost problem. Every compliant trade finance deal requires document checks, sanctions screening, counterparty due diligence, collateral perfection and ongoing monitoring. Those costs are largely fixed per transaction. When they exceed the fee income on a $75,000 shipment, the rational lender declines, whatever the credit quality. On this reading, the gap is a minimum viable ticket size. The SMEs beneath it are excluded by arithmetic, not by riskiness.
My working paper, Liquid Supply Chains, argues the second story deserves to be taken seriously. Full disclosure before I go further: I am the founder of Rhofin, where our team is building AI-enabled trade finance origination. I have a direct commercial interest in the answer. Read what follows with that in mind.
Three shifts are now attacking the unit-cost arithmetic.
The first is agentic AI. Trade finance underwriting is document-heavy, repetitive work: reading invoices, bills of lading, and packing lists, cross-checking them, screening counterparties, flagging anomalies, building an audit trail. Machines are increasingly good at this kind of structured verification work. But one qualification matters more than the headline. The binding constraint banks report is not analyst hours. It is the liability attached to a compliance failure, and a sanctions penalty does not shrink because the screening was cheap. Automation only helps if error rates fall alongside costs. That is a conditional claim. It has to be proven in live transactions, not assumed in theory.
The second shift is legal. Digital trade documents are becoming enforceable collateral. The UK’s Electronic Trade Documents Act, Singapore’s adoption of the UNCITRAL model law and the control rules for electronic documents of title in the US Uniform Commercial Code all point the same way. In the right structure, a financier can hold an enforceable interest in goods while they move, with release linked to repayment or acceptance. Digitisation stops being a convenience and becomes collateral infrastructure.
The third shift is data. Logistics systems now make the trade itself observable. Bookings, vessel movements, milestone events, proof of delivery: together they form a third-party record that an invoice can be anchored to. A fake invoice is easy to manufacture. A fake shipment moving through independent logistics systems is much harder.
Put the three together and something important happens. The unpriceable tail, the fraud and document failure investors cannot see, converts into ordinary, measurable credit risk. Trade finance does not need to become risk-free. It needs to become verifiable, standardised, and transparent enough for institutional money to fund with confidence.
This is where private credit comes in. Banks remain essential to trade, but they are boxed in by compliance liability, balance-sheet usage, and operational capacity. Private credit grew precisely by funding what banks cannot hold at scale, yet most of it sits in sponsor-backed direct lending. Shipment-level trade finance is the opposite animal: short duration, self-liquidating, tied directly to real-economy flows, and diversified across thousands of buyers, sellers and routes. The funding sequence is clear enough: warehouse lines first, then forward-flow purchases, eventually rated securitisation.
The mortgage precedent is tempting, and it is only honest with a caveat attached. Yes, mortgages became a global asset class once origination and performance data were standardised enough for investors to underwrite pools. But mortgage liquidity was also built on agency guarantees that absorbed credit risk at public expense. No such guarantor stands behind trade receivables. Private substitutes have to do that work: retained first loss, decision-level transparency to funders, and the discipline of short tenor, which means an originator faces the consequences of bad screening within months rather than decades.
Greensill is the whole architecture stated in the negative. Prospective receivables tied to no completed trade. Concentration hidden inside a supply chain finance label. End investors with no way to verify the claims they were funding. Insurance standing in for diligence. Every design requirement above is meant to negate a specific Greensill failure mode.
The stakes go well beyond portfolio construction. SMEs account for roughly 90 per cent of the world’s businesses and more than half of global employment. In emerging economies, they contribute up to 40 per cent of GDP. When good firms are short of working capital, new credit does not sit idle. It turns into more orders, more production and more growth. That is what a binding constraint looks like. A typical small exporter can be out of pocket for 100 to 135 days on every order, so a fixed credit line caps how many orders it can run at once, whatever the demand. Growth is capped by cash, not by customers.
There is a fairness point too. Large corporates already enjoy transaction-level financing through captive finance arms and supply chain finance programmes. Their suppliers and smaller rivals do not. Extending shipment-level finance down-market creates no new privilege; it generalises an existing one. And the effects go deeper than volume: credit constraints shape who gets to compete, not just how much incumbents ship. As supply chains diversify away from established hubs, many of the new suppliers will be firms without deep credit histories. If finance does not move with the trade, diversification will be throttled by working capital rather than demand.
The good news is that this is measurable. Trade exposures mature in 45 to 90 days, so the central claim can be tested within quarters: do newly bankable SME segments perform like the incumbent trade-finance book, or do losses rise in the way adverse selection would predict?
For private credit, that is the opportunity: not another leveraged-loan substitute, but a granular, short-duration asset class tied to the movement of goods.
Trade finance has never lacked demand. It has lacked a scalable way to make small transactions investable. That may finally be changing.
