Trade Settlement Infrastructure
Why the recourse question has a third answer, and what that implies about what we’re building
Ask how a commodity trade finance facility handles recourse and you’ll usually get a binary. Either the facility has recourse to the trader, or it doesn’t. Pick one, price accordingly.
Both answers are bad at mid-market ticket sizes, and working through why with Caroline Hirasaka — a trader who went through our financing scheme in detail — is what moved us off the binary entirely.
She set out the two options cleanly. Option A: non-recourse on buyer credit risk, where the financier absorbs the buyer’s insolvency or default. Option B: recourse against the trader, where the trader remains liable if the buyer fails to pay. Her view was that Option A was the one worth structuring properly, if the economics supported it — because you cannot credibly tell a trader that repayment risk sits with you and then hand them a guarantee that says otherwise.
We agreed with the direction and ended up somewhere slightly different: not choosing between the two parties, but splitting between the two causes.
Recourse split by cause
Full recourse makes a facility unattractive to exactly the traders worth having — the ones with alternatives. Full non-recourse means underwriting risks nobody can price from outside the commercial chain. Both failures come from the same mistake: treating “the deal went wrong” as one event with one owner, when it’s several different events with different owners.
So we split by cause.
Credit risk on a conforming presentation sits with us, with no recourse to the trader. The reasoning is narrow and deliberate: where the credit is confirmed by an acceptable bank, this is bank risk, not buyer risk. Known counterparty, known instrument, priceable. It is the risk we are equipped to carry, and carrying it is most of what the trader is actually paying for.
Performance and documentary risk sits with the trader, with full recourse. Discrepancies, short shipment, quality and quantity failures, misrepresentation, fraud. A financier standing outside the commercial chain cannot meaningfully assess these — they are consequences of how the trader performs. Pricing them into the facility means charging every trader for the behaviour of the worst one.
The instrument this produces is a performance guarantee, not a repayment guarantee, and our documents now say exactly that, in those words. The distinction is the whole design: the trader is not standing behind the buyer’s ability to pay, and we are not standing behind the trader’s ability to perform.
What this is not is unconditional non-recourse. We take credit risk where it is bank risk or where it is insured. Where it is neither, we either decline the transaction or take it with recourse and write that into the termsheet. Country and sanctions risk we don’t take at all — we decline the deal. Price risk gets fixed or hedged. Cargo risk gets insured, with the financing structure as beneficiary.
The leverage problem
The sharpest thing Caroline raised wasn’t about recourse at all.
The bundle we offer — capital, settlement control, documentary control, counterparty assessment, risk allocation — gives a trader real operational leverage. Her observation was that this cuts both ways: the greater the leverage offered, the more critical the trader’s quality becomes. Every improvement to the offer makes counterparty selection more load-bearing.
That’s the central tension of the product, and it doesn’t have a clever answer. What it has is a tiered limit ladder. A first transaction runs at the tight end — smaller limit, tighter structure, higher rate within the range, inspection at both ends, full document pack. As cycles settle and the trader demonstrates performance, limits rise, pricing improves, advance rates increase, and decision time shortens.
Get CARTHO’s stories in your inbox
Join Medium for free to get updates from this writer.
What doesn’t loosen at any tier: requirements on the credit and the issuing bank, KYC/AML and sanctions screening, the performance guarantee, and inspection of non-exchange-traded goods. Those are structural rather than commercial, and they don’t get traded away for volume.
The ladder isn’t a loyalty programme. It’s the only honest way we know to price a risk that can’t be assessed in advance — let the counterparty reveal itself at a size where being wrong is survivable.
Why “Trade Settlement Infrastructure”
Caroline’s closing point was that defining this as a trade finance company understates it: Cartho might not be a single structure so much as an engine for structures, and the broader frame is trade settlement infrastructure.
We think that’s right, and the reason is architectural rather than aspirational.
What we’ve built is one control loop: settlement control, documentary control, counterparty assessment, and risk allocated by cause. That loop is instrument-agnostic. It currently runs through a documentary credit structure, because that’s what our first transactions use. The same loop supports receivables purchase, factoring and forfaiting, warehouse receipt structures, and prepayment against title or trust receipt. Different instruments, same control layer underneath.
That’s what makes the infrastructure framing meaningful instead of promotional. The claim isn’t that we do more things. It’s that the hard part: controlling settlement and documents, assessing counterparties, allocating risk to whoever can actually manage it is the same problem regardless of which instrument sits on top. Build it once and the instrument becomes configuration.
We also don’t put this frame in front of traders. A trader reading our documents needs one question answered “should I send this deal?” and an infrastructure thesis gets in the way of answering it. The frame matters for people deciding whether to build alongside us. That’s a different conversation, and it’s this one.
Where this stands
Two transactions are in structuring now: one in specialty coffee, one in refined metals. One is currently gated on a single condition — written bank acknowledgement that credit proceeds route to the controlled account. Until that acknowledgement exists, we don’t fund.
What we’re looking for
Banks and institutional lenders. These transactions sit in the range where a full facility rarely justifies a bank’s origination and underwriting cost, but where the underlying exposure is often bank risk against a conforming presentation. If your institution has appetite for that exposure without the origination overhead, there’s a conversation to have about co-financing and corridor partnership — we do the documentary and counterparty work, you take a defined slice of a risk you already know how to price.
Insurers and credit risk carriers, on the same logic. The structure is designed to be insurable rather than to substitute for insurance.
Traders, if you’ve read this far into something not written for you: the shortest path is to send a live transaction and let the structure be assessed against it. Terms are transaction-specific and live in a termsheet, not in an article.
If you’re a bank, an insurer, or a trading house looking at the same gap, I’d like to hear how you see it, including where you think we have it wrong. That conversation has improved this product more than any quarter of internal work.
Thanks to Caroline Hirasaka, whose read of our financing scheme prompted the restructuring described here.
Sergei Goriachev — Founder, Cartho cartho.trade
Cartho builds the control layer for mid-market commodity trade settlement — the cargo transactions that sit below the threshold where banks originate directly.
