Trade credit

Why does a supplier short of cash itself agree to let its customers pay ninety days late?

A components maker with an overdraft it can barely service ships a lorryload of goods and agrees to be paid in ninety days. It has just lent its customer the value of that lorry, at a moment when it is itself borrowing expensively to cover wages. Put like that it looks close to irrational — a poor firm financing a richer one.

And yet firms owe each other vast sums on ordinary invoices, in every industrial economy, as a matter of routine. So the question is not why this particular supplier is foolish. It is why the party holding the goods, rather than the party holding money, so often turns out to be the lender.

Start by asking what a lender actually needs, and who is best placed to supply it. A lender needs to know whether the borrower is sound, and needs a way to make default expensive. Banks buy both at some cost: accounts, covenants, monitoring, a security interest, eventually a court.

The supplier gets much of it as a by-product of trading. It already knows the customer's order pattern, and a firm sliding into trouble usually changes what it orders before it changes what its accounts say — so the supplier often sees the deterioration first, for free. It knows the industry well enough to judge whether the customer's stated plans are plausible. It knows the resale value of its own goods, because it sells them for a living, so repossession is worth more in its hands than in a bank's.

And it holds a sanction no bank has. A bank that is not paid can sue. A supplier that is not paid can simply stop the next delivery — immediate, cheap, needing no court, and devastating if it is the only source of a part the customer's line runs on. Notice this is the same logic as reputation-based lending generally, but concentrated: the hostage is not the borrower's good name in the abstract, it is tomorrow's shipment.

There is a second mechanism running the other way, and it is easy to miss because it is not about lending at all. Delay in payment is an inspection period. The buyer takes goods it cannot fully judge until it has unpacked, tested and used them; paying in thirty days means the money is still under its control if the shipment is wrong. A supplier confident in its quality can offer that cheaply, and a poor one cannot — so terms function as a warranty and a signal, not only as finance.

Then there is the margin. A bank lending to make an interest spread must cover its losses out of that spread. A supplier lending to make a sale earns the gross margin on the goods as well. If a sale carries a thirty per cent margin, a supplier can accept a default rate that would ruin a lender charging ten per cent, because the alternative to the risky sale is often no sale.

Now, is the credit free? The price is usually hidden in the discount. "Two ten, net thirty" means two per cent off if you pay within ten days, otherwise the full sum at thirty. Forgo the discount and you pay 100 instead of 98 for twenty extra days of money: 2/98, or 2.04 per cent for twenty days. There are 18.25 such periods in a year, so compounding gives 1.0204 raised to the 18.25, about 1.446 — roughly a 45 per cent annual rate. That is not cheap money. It is a rate only a firm shut out of everything better would willingly pay, which is precisely who pays it.

Which suggests the test. If trade credit were simply disguised lending, terms would be underwritten borrower by borrower and would move with interest rates. They do not: terms cluster on a handful of industry conventions and stay put for decades, largely regardless of who the buyer is — a fact I am recalling rather than deriving. That uniformity is the signature of credit bundled into the price of the goods rather than priced per borrower, and it is what would have to be false for the story here to be wrong. If we found suppliers quoting each customer its own terms, tracking that customer's credit standing, the bank-substitute account would win outright.

Two refinements, and one thing that should not be dressed up as efficiency.

First, the mechanisms are not rivals. Information, the delivery sanction, the inspection period and the margin cushion all operate at once, in proportions that differ by industry. Which of them dominates is genuinely argued over in the literature; the individual mechanisms are much better established than their weighting.

Second, this is not the general question of why a promise is accepted at all. That question is about what makes default expensive — collateral, courts, records. This one takes enforcement as given and asks a narrower thing: given that someone will lend, why is it so often the seller? The answer is specialisation. The supplier is not a better-hearted lender; it is a lender with lower costs, because it acquired the information and the leverage while doing something else.

Third, the uncomfortable part. A great deal of what is called trade credit is not chosen by the lender. A large buyer that unilaterally moves its terms from thirty days to ninety has not bought a service; it has taken an interest-free loan from a smaller firm that must now borrow at a worse rate to bridge the gap. No production has become cheaper. That is a transfer, and calling it supply-chain efficiency does not make it one — which is why late-payment law exists in many countries and why it targets exactly the strong-buyer case.

• Why trade credit contracts sharply in a downturn, exactly when firms need it most.

Every term the collection defines is gathered in the glossary.