Credit

Why will a stranger hand over goods today in exchange for nothing but a promise?

A merchant releases goods worth a year of their income to someone they have never met, in exchange for a sentence. Nothing physical has been received. The buyer could simply not pay, and both parties know it.

Our instinct is to call this trust, and to stop there. But trust is the thing to be explained, not the explanation. What would have to be true of the world for a rational stranger to accept a promise from another rational stranger?

Try the pure case first. Two people meet once, never see each other again, and no one else observes the exchange. The borrower who takes the goods and vanishes is better off in every respect. Anticipating that, the merchant hands over nothing. Credit does not exist. Notice that no dishonesty was required to destroy it — only the absence of consequence.

So the question sharpens. Credit exists wherever breaking the promise has been made expensive enough that keeping it is the better option. Not because people are good, but because default costs something. What can make it cost?

The most direct device is collateral. The borrower hands over a claim on a thing — a house, a rig, a stock of grain — that the lender may take if payment fails. This is elegant because it converts a promise into property, and property does not need to be believed. It has requirements though: the asset must be identifiable, its ownership must be checkable, and transferring it must be possible. Where land is held without clear title, it cannot be pledged, however valuable it is. Hernando de Soto built a well-known argument on this point; how much of poor countries' credit scarcity it explains is disputed, but the mechanism itself is not.

The second device is the future itself. If the relationship repeats, the borrower who defaults loses tomorrow's credit as well as today's. Reputation, in this sense, is a hostage: the value of continued access is the thing that would be forfeited. And it works only when someone remembers — which is why the third device matters. A court makes the debt enforceable against a stranger by putting the state's coercive machinery behind the sentence. A credit bureau does something subtler: it takes a reputation that was previously local and makes it portable, so that a default in one town raises the price of borrowing in every other. Both are technologies for reproducing, among strangers, the conditions of a village.

Now, given the promise is accepted, why does it cost more than the sum lent? Interest is usually explained as one thing and is in fact at least three. Some of it compensates pure waiting: the lender gives up the use of the money now, and consumption now is worth more than the same consumption later — time preference. Some of it compensates risk: a fraction of borrowers will not repay, so the rate charged to all must cover the losses on some. And some of it compensates expected inflation: money repaid later buys less, so the lender must be handed more of it just to stand still. These are genuinely separate, and separating them is diagnostic. A high rate on an unsecured consumer loan is mostly the risk component, not impatience; a high rate in a currency losing value is mostly the inflation component and may reflect no real return at all.

Two refinements keep this from becoming a story about punishment.

The first is that enforcement can be too harsh as well as too weak. Debtors' prisons made default catastrophic, which sounds like it should have expanded lending; it did not, partly because it deterred borrowing and partly because a ruined debtor repays nothing. Limited liability and bankruptcy discharge look like weakened enforcement, and they are, but they are also predictable — lenders can price a known ceiling on loss far more easily than an unbounded one, and borrowers will take risks worth taking. A credit system needs its consequences calibrated, not maximised.

The second is that these devices substitute for one another. Strong collateral makes reputation less necessary; a reliable court makes both less necessary; a dense information network can carry lending where courts are slow. Trade credit between firms often runs on reputation and repeat dealing with barely any formal enforcement at all. When we say a country lacks credit, the useful question is which of the devices has failed, because they are not interchangeable in cost.

• Why credit expands and contracts far more violently than the underlying economy does.
• What changes when the record itself becomes the asset — credit scoring, and errors that cannot be argued with.

Every term the collection defines is gathered in the glossary.