Why Settlement Is Harder Than It Looks

Settlement is the point at which a payment becomes final. Before that moment, the transaction is a promise. After it, the money has genuinely moved and cannot be pulled back by the ordinary operation of the system.

Put that way it sounds like a technical detail. It is closer to the foundation that determines how everything above it behaves.

Finality is a choice, not a fact

Different systems make finality happen at different moments, and the choice has consequences.

In a card transaction, the merchant learns almost immediately that a payment has been authorised, but settlement happens later, in batch, and can be reversed for months afterwards through the dispute process. The merchant carries that risk.

In a real-time gross settlement system of the kind central banks operate, each payment settles individually and immediately in central bank money. Once done, it is done. The cost is that every participant must hold enough liquidity to fund each payment at the moment it is made.

Neither approach is correct in the abstract. They are different trades between speed, cost, risk, and liquidity. What matters is that the choice is made deliberately, because everything downstream inherits it.

The consequences run further than expected

A system with delayed finality needs a dispute mechanism, because errors have to be correctable after the fact. A dispute mechanism needs rules about who bears loss. Those rules need an arbiter. The arbiter needs to be funded, which means fees, and the fees have to be allocated among participants who will argue about the allocation.

A system with immediate finality avoids most of that, and pays for it elsewhere. Participants need liquidity available at all times. Errors become extremely expensive, because there is no ordinary route to reverse them. Fraud prevention has to happen before the payment, not after, which raises the cost of every transaction rather than only the disputed ones.

This is why comparing payment systems on speed alone is misleading. A system is not better because it settles faster. It is different, and the difference shows up as risk somewhere else.

Where new systems get this wrong

The recurring mistake in newer payment designs is treating finality as a performance metric. Faster is presented as strictly better, and the corresponding risk transfer goes unmentioned.

But somebody always carries the risk. If settlement is instant and irreversible, the person who sent money to the wrong place has no recourse. That may be an acceptable design, and for some use cases it clearly is. It is not a free improvement, and describing it as one tends to mean the consequences have not been thought through.

The systems that have lasted are the ones where this trade was made explicitly, written down, and priced. The ones that struggle are usually the ones that assumed the question away.