Ref CP-013 · 18 September 2026 · Joshua M. · Reading time 9 minutes

Trade-based money laundering: reading an invoice for red flags

Most trade-based money laundering training stops at a typology list: over-invoicing, under-invoicing, phantom shipments, multiple invoicing. Useful as vocabulary, useless when there is an actual invoice on your screen and a payment sitting in a queue. This note is about what those four patterns look like against real documentation, and why the unit price field earns more of your attention than anything else on the page.

What is actually in front of you

When people picture trade finance compliance they picture documentary credits: a letter of credit, a full set of documents presented to a bank, a trained checker comparing them line by line against the terms. That work still exists, but it is now the minority of what an analyst sees. Most trade-based money laundering questions reach me in a much scrappier form. A payment of two hundred and forty thousand dollars has hit a monitoring rule. The relationship manager has been asked for supporting documentation. Back comes a PDF invoice, sometimes a packing list, occasionally a bill of lading, and a short email saying the customer imports machinery parts.

So the practical skill is not documentary examination in the classical sense. It is the ability to look at one or two pieces of paper of uncertain provenance and decide whether the commercial story they tell is plausible enough to close, or implausible enough to escalate. That is a narrower skill than it sounds, and it is learnable.

Start by naming what you have. A commercial invoice states who is selling, who is buying, what, how much of it, at what price per unit, on what delivery and payment terms. A packing list tells you how the goods are physically made up: cartons, weights, dimensions. A bill of lading or air waybill evidences that a carrier took possession. A certificate of origin says where the goods were made. Each document is a claim by somebody. None of them is proof. The bill of lading is the closest thing to independent evidence in the pile, because a third party carrier issued it, and that is precisely why the absence of one is worth a question.

Unit price is the field that does the work

If I could keep one field on a commercial invoice and delete the rest, I would keep unit price.

The reason is arithmetic. An invoice total is a single number with no external reference point. Two hundred and forty thousand dollars is neither large nor small until you know what it bought. But a unit price sits at the intersection of three things you can test: it must multiply out correctly against the stated quantity, it must be consistent with the physical description on the packing list, and it must bear some relationship to what that commodity costs in the world. Value manipulation, which is the engine of most trade-based laundering, has to show up in the unit price or in the quantity. There is nowhere else for it to hide.

So the first thing I do with any trade document is rebuild the arithmetic by hand. Quantity times unit price, does it equal the line total. Do the line totals equal the invoice total. Does the invoice total equal the payment being made, and if not, why not — is this a part shipment, a deposit, a consolidated settlement of several invoices? I have lost count of the number of files where the sums simply did not work and nobody had checked, because the total at the bottom looked like a normal number and the eye slides to normal numbers.

Then cross-read against the packing list. Three hundred cartons with a gross weight of four thousand kilos means each carton weighs a little over thirteen kilos. If the invoice says those cartons contain industrial diamonds, something is wrong with one of the two documents. If it says they contain ceramic tiles, the weight is unremarkable but the stated unit price of eight hundred dollars a carton is not. Physical plausibility and price plausibility are two separate tests and you should run both.

Checking a price when you are not a commodities expert

Nobody expects you to know the spot price of refined palm oil. I have run trade-facing teams for years and I do not know it either. What I do expect, from myself and from anyone I have trained, is the ability to notice when a stated price is wrong by an order of magnitude.

The distinction matters because it sets a sensible standard. A fifteen per cent variance from some benchmark is not a red flag. Prices vary by grade, by season, by contract term, by incoterm, by whether the buyer is taking one pallet or a full container load, by currency movement between contract and shipment. If you flag every fifteen per cent variance you will produce noise and your credibility will erode within a quarter. A factor of ten is a different animal. Laundering value through trade requires meaningful distortion, because the point of the exercise is to move value, and small distortions move small value.

For the benchmark itself, use whatever is free and public. Exchange-traded commodities have published prices. Many governments publish customs trade statistics showing average declared values per unit for a given commodity code between given countries, which is often the single most useful reference you can get. For finished goods, online retail is a ceiling: nobody is buying wholesale at more than the consumer pays, so a wholesale invoice priced above retail is telling you something. And for genuinely obscure industrial items, a supplier catalogue found in three minutes of searching is better than nothing, which is what you have otherwise.

Write down the benchmark you used and where it came from. An analyst who says "the price looked high" has an opinion. An analyst who says "declared at USD 412 per metric tonne against a published national customs average of USD 39 per metric tonne for the same commodity code on the same trade route in the same quarter" has a finding, and it will survive review by quality assurance, by the MLRO, and by anyone who reads it two years later. The contents of an enhanced due diligence file should include those references as screenshots or saved pages, not as a recollection.

The four patterns, as they appear on paper

Over- and under-invoicing

Same mechanism, opposite directions. Over-invoicing moves value to the exporter: the buyer pays more than the goods are worth, and the surplus has been transferred across a border with a commercial explanation attached. Under-invoicing moves value to the importer, who receives goods worth more than the payment made. On paper, both look like one thing — a unit price that does not match the world. What tells you the direction is which side of the transaction your customer sits on.

The version that catches people out is the one where the price is defensible in isolation but the pattern is not. I have seen relationships where every individual invoice sat within a plausible band, but the same customer bought identical goods from the same supplier at prices that moved by a factor of four across eight months with no market event to explain it. One invoice tells you very little. Twelve invoices tell you a great deal, which is an argument for always pulling the payment history before forming a view.

Phantom shipments

Here there are no goods at all. The invoice is real in the sense that it exists as a document; the trade it describes never happened. The tells are evidentiary rather than arithmetical. There is no transport document, or the transport document has been produced by an entity you cannot find, or the container number does not conform to the standard format, or the named vessel was not on that route in that window. Public vessel tracking is available and free at a basic level, and I would rather an analyst spend ten minutes there than an hour rewriting a narrative.

Watch, too, for round numbers. Genuine commerce produces untidy figures: 17,438 units at 3.27 each. An invoice for exactly one hundred thousand dollars, for exactly one thousand units at exactly one hundred dollars, is not proof of anything. It is just unusual, and unusual things deserve a question.

Multiple invoicing

One shipment, financed or paid for several times, sometimes through several institutions so that no single one sees the duplication. Within your own institution this is findable: search on invoice number, on amount, on the counterparty name, on the bill of lading number if you have it. Across institutions it is much harder, which is one reason the Wolfsberg Group has pushed so consistently on information sharing in trade finance.

The variant I see more often in payments businesses is the same invoice presented twice with cosmetic changes — a date shifted by a week, an invoice number incremented, a line item reworded, the total identical to the cent. Identical totals across supposedly separate transactions are worth a closer look every time.

Misdescribed commodities

The goods exist and the price is plausible for what is described, but the description is not what is in the box. This is the hardest of the four for an analyst sitting at a desk, because you cannot open the container. What you can do is test for internal contradiction: goods described as high-value electronics with a gross weight suggesting textiles, a certificate of origin from a country with no production capacity in that commodity, an unusual incoterm, or a commodity code that does not match the plain-English description. FATF has published extensively on trade-based laundering typologies, and the misdescription cases in that material are consistently the ones that required a physical inspection to resolve. That is not a failure on your part. Escalate what you can see.

What to do with what you find

Three things, in decreasing order of how often they are skipped.

Ask the customer. A price ten times the benchmark can have a legitimate explanation — bespoke manufacture, a bundled services element, an urgent air-freight surcharge rolled into the unit price, a long-term contract struck before a market collapse. You are not accusing anybody of anything by asking; you are filling a gap in the record. Several of the most implausible-looking files I have worked resolved cleanly within a day of somebody picking up the phone. Guidance for the UK sector from JMLSG is clear that reasonable enquiry is part of the job, not an imposition on the relationship.

Write the benchmark into the narrative. If the matter goes to a disclosure, the arithmetic is your strongest material, because it is checkable by someone who has never seen the file. The worked example of a SAR narrative in a sibling note shows how to lay out a quantified comparison so the reader does not have to reconstruct it.

Feed it back to monitoring. If the trade concern only surfaced because a velocity rule fired, that is worth saying out loud. Trade documentation is rarely structured data, so the detection usually has to come from payment behaviour, which means the thresholds on those behavioural rules matter more in a trade-exposed portfolio than elsewhere. That is a conversation with whoever owns the rules, and it benefits from being framed the way any monitoring rule change should be framed: with the population data, not the anecdote.

The last thing I would say to a junior analyst is about confidence. You will never know as much about steel grades as the customer's procurement director does, and you are not meant to. Your expertise is different. It is knowing that value must be conserved, that documents must agree with each other, and that a price which is wrong by a factor of ten is worth asking about no matter who is on the other end of the call.

What matters: An invoice you cannot multiply out, or a unit price ten times its benchmark, is a question you are entitled to ask — and the benchmark you used belongs in the file, not just in your head.