Ref CP-013 · 11 September 2026 · Joshua M. · Reading time 8 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, multiple invoicing, phantom shipments. That list tells you what the schemes are called but not what they look like when a commercial invoice, a bill of lading and a packing list are sitting in front of you at half past four on a Thursday. This note is about reading the documents.

Why trade documents are hard in a way that payments are not

A domestic payment is a small, well-behaved object. Payer, payee, amount, date, reference. Your monitoring system can count them, compare them, and shout when the arithmetic looks unusual. Trade documentation is not like that. A single shipment might generate a commercial invoice, a proforma, a packing list, a bill of lading or air waybill, a certificate of origin, an inspection certificate, and an insurance document — each produced by a different party, each with its own conventions, none of them under your control and most of them scanned slightly crooked.

That is the whole problem. Value moves inside the description of the goods rather than inside the payment instruction. The payment leg looks entirely ordinary: an importer in one country pays an exporter in another for a shipment of goods, through correspondent channels that have done this a thousand times. Nothing in the payment message is wrong. The lie, if there is one, is in the price.

FATF has been writing about this for close to two decades, and the reason the typologies have barely changed is that the underlying mechanism is very hard to close off. You can read their trade-based money laundering material at FATF, and the Wolfsberg Group trade finance principles are the more practical companion piece, written by people who have had to operationalise this inside banks.

The four shapes, as they appear on paper

I want to describe these the way they actually turn up in a file, not as definitions.

Over- and under-invoicing

Over-invoicing is a seller charging more than the goods are worth, which moves value from the buyer's jurisdiction to the seller's. Under-invoicing runs the other way. On the page, neither of these is exotic. You have an invoice for 12,000 units of something at a stated unit price, a total that multiplies out correctly, payment terms that make sense, and a bill of lading that matches the quantity. Every field is internally consistent. Internal consistency is exactly what you would expect, because the parties wrote all the documents.

The only external reference point is the market. If cotton bed linen is invoiced at four hundred dollars a set, the arithmetic on the invoice is perfect and the commerce is nonsense.

Phantom shipments

Here the goods do not exist at all. The paperwork exists, the payment exists, and nothing was loaded onto anything. What gives it away is usually not the invoice but the transport documents around it: a bill of lading with a container number that does not conform to the standard format, a named vessel that was not in that port on that date, a port pair that no carrier serves, or a weight that the stated container could not physically hold. I have seen a packing list for eighteen tonnes of goods travelling in a twenty-foot container rated well below that.

Multiple invoicing

One shipment, financed several times, sometimes across several institutions who cannot see each other's files. This one is close to invisible from inside a single bank, which is why it survives. What you are looking for is repetition: the same invoice number with a slightly different date, the same bill of lading lading number appearing in two facilities, the same quantity and description recurring at intervals that do not match any plausible production cycle.

Misdescribed commodities

The goods exist and the price is plausible for what is written down, but what is written down is not what is in the box. Low-value goods described as high-value, or the reverse. This is the hardest of the four to see from documents alone, because the documents are consistent with each other and with the market. Usually it surfaces through customs data, an inspection, or a mismatch between the commodity described and everything else you know about the customer — a company whose stated business is agricultural machinery suddenly shipping pharmaceutical intermediates.

Unit price is the single most useful field on the page

If I could only look at one field, it would be unit price. Not the total, not the counterparty, not the payment terms. Unit price.

The reason is that the total is negotiable and the unit price is not. A large total might be a large legitimate order. A small total might be a sample shipment. But the unit price of a physical commodity is anchored to a world outside the customer relationship, and that anchor is the only thing in the file the parties did not write themselves.

Unit price is also where fraud has to show up, arithmetically. If someone wants to move three million dollars of unexplained value through a shipment of ceramic tiles, they can either invent a vast quantity of tiles — which creates a logistics problem, because vast quantities of tiles need vast numbers of containers, and containers leave traces — or they can inflate the price per square metre. Inflating the price is cheaper and quieter. So that is usually what happens, and it is visible in a field that appears on almost every commercial invoice you will ever open.

A practical habit: whenever the invoice gives you a total and a quantity but no unit price, divide. It takes four seconds and it is frequently the moment something goes wrong. Invoices that omit the unit price are not automatically suspicious — plenty of legitimate ones do — but the omission means nobody downstream has done that division, and you may be the first.

Sanity-checking a price without becoming a commodities expert

This is the part juniors worry about most, and they worry about the wrong thing. Nobody expects you to know the spot price of refined palm oil. What you are expected to notice is when a number is wrong by an order of magnitude, and that is a much lower bar than it sounds.

The workable method is coarse. Take the unit price. Find two or three public reference points — a trade statistics database, a commodity index, an industry association price sheet, or failing all that, several retail listings for a comparable product, which at least give you an upper bound because retail sits above wholesale. Write down what you found and where. If your invoice price is within a factor of two or three of your references, you have no finding worth escalating on price alone. If it is twenty times higher, you have something.

Three things I would ask you to hold onto while doing this.

• Incoterms change the number. A CIF price includes freight and insurance; an EXW price does not. Comparing a CIF invoice to an ex-works benchmark will produce a gap that means nothing.
• Grade and specification matter enormously — surgical-grade stainless steel and construction rebar are both "steel", and the price per tonne is not remotely comparable, so record which grade the invoice actually specifies before you pick a benchmark.
• Small orders cost more per unit. A five-hundred-unit order will not carry container-load pricing, and a modest premium on a small quantity is commerce, not concealment.

And record your uncertainty honestly in the file. "Benchmark located for a similar but not identical specification; invoice price is approximately eighteen times that benchmark" is a defensible sentence. Pretending to a precision you do not have is how good findings get taken apart at review.

Turning a price observation into a file

A price gap on its own is thin. It is arithmetic, and arithmetic has explanations — an urgent air-freighted replacement order, a bundled service contract, a related-party transfer price, a specification you misread. Your job is not to conclude that the customer has done something wrong. Your job is to establish whether the commercial story holds together, and to record what you found either way.

So the price finding goes looking for company. Does the corporate structure put the buyer and seller under common control? That is the single most valuable corroborant in trade work, and it is why finding the ultimate beneficial owner matters more here than in almost any other alert type. Does the customer's stated business model, as captured when you built the file, accommodate this commodity at all? If the profile is thin, the gap in what an EDD file should contain is itself part of the finding. Do the transport documents survive checking? Do the payment flows match the trade flows in timing and amount, or is money arriving before goods ever move?

When you write it up, resist the pull towards conclusions the documents will not carry. Say what the invoice said, say what the benchmark said, say what you could not verify. Whether you are working to the standards of the JMLSG guidance in the UK or to expectations set elsewhere, the disclosure is judged on whether a reader who has never seen the file can follow your reasoning. The same discipline that runs through how to write a SAR narrative applies with extra force in trade, because you are asking the reader to accept a commercial judgement, not just a pattern of transfers.

One last thing about temperament. Trade files are slow. A documentary set can take an hour to read properly and most of that hour produces nothing. That is the correct ratio. If your alert triage rhythm is built around fifteen-minute dispositions, trade alerts will break it, and they should be queued and resourced differently rather than rushed to closure.

Worth remembering: The parties wrote every document in the file except the market price, so that is where you start reading.