Haute Lumière
Commerce · III.09 · MMXXVI · daylight
ASSESSMENT · Chapter III.09 — Digital Settlement Without Extraction
Three instruments: a ten-point quiz, eight reflection questions, five essay prompts. The quiz checks comprehension rather than recall. The reflections are private and first-person. The essays are arguable from more than one side.
THE QUIZ — ten points
Four on recall.
1. What is interchange, who pays it to whom, and who sets it?
A fee the merchant's acquiring bank pays the cardholder's issuing bank on every card sale, at a rate set by the card network rather than by either bank. One mark for the direction of the payment, one for naming the network as the price-setter — that is the part that makes it unlike an ordinary negotiated fee.
2. State the Banco Central do Brasil's charge for settling a Pix payment, and express it as a share of the average Pix ticket.
R$0.01 per ten credits, so R$0.001 per transaction. Against an average ticket of R$443.11 that is 0.000226 per cent of value. Credit any answer reaching a tenth of a centavo; the share matters more than the decimal.
3. Name the finality characteristics of Bitcoin and of Ethereum under proof-of-stake, with the time each takes.
Bitcoin: probabilistic, six confirmations at ten minutes a block — 60 minutes, and still a probability. Ethereum: deterministic, two epochs of thirty-two twelve-second slots — 768 seconds, or 12.8 minutes. Bitcoin takes 4.69 times as long for a weaker guarantee.
4. What did Pay.UK's cost base cover in 2025, what was it, and why does the March 2026 rebate matter?
£161.1 million, covering Faster Payments, Bacs, the Image Clearing System and the standards work together. The £27.5 million rebate matters because returning an over-collection — half a penny per payment across 5.5 billion — is the test that distinguishes a cost-recovering utility from a margin-earning business.
Four on application.
5. A colleague says: "The merchant discount rate is what settlement costs. You cannot get it cheaper." Answer them with the published figures.
Settlement, priced by the institution that actually performs it, is 0.000226 per cent of value. A Brazilian merchant's all-in cost on Pix is 0.22 per cent — 975 times the settlement charge, and that gap is genuine work: fraud, disputes, onboarding, support. The card merchant discount rate of 2.2 per cent is 9,748 times the settlement charge. Full marks require both halves: the 0.22 per cent is mostly real, which is exactly what makes the remaining 1.98 per cent hard to account for.
6. Why does a percentage fee survive when the underlying cost does not scale with value? Give one structural reason and one commercial one.
Structurally, a two-sided platform can charge one side to subsidise the other, and interchange recruits cardholders at the merchant's expense — a real launch argument, from Rochet and Tirole. Commercially, once the network is universal the subsidy recruits nobody and instead funds rewards, and the rewards are what stop the cardholder switching. Credit an answer that notes the argument is about launching a network and is being used about a mature one.
7. A treasurer proposes moving supplier payments to a distributed ledger for "instant settlement". You have the finality numbers. What do you tell them?
That the best available ledger finality is 12.8 minutes and Bitcoin's is 60 minutes probabilistic, while the domestic instant rail is seconds and irrevocable — so the ledger is slower, not faster. And that the float argument is weak anyway: an hour on a $10,000,000 payment at a 5 per cent funding cost is $57.08. The stronger answer names the case where the ledger does win — cross-border, or where the settlement authority itself is not trusted.
8. A regulator mandates a zero-fee instant rail and builds nothing else. Predict the outcome and name the mechanism.
Slow adoption, because a rail with no margin has no salesforce and nobody's commission depends on the merchant signing up. The United Kingdom is the control group: seventeen years to 11.1 per cent, against Brazil's 54.7 per cent in five under a participation mandate. Credit any answer naming acceptance, rather than price or technology, as the binding constraint.
Two that require the arithmetic to be done.
9. A merchant takes R$2,400,000 a year on cards at a 2.6 per cent net margin. She migrates 60 per cent of that volume from a 2.2 per cent card fee to a rail costing 0.22 per cent. What happens to net profit? Show your working.
Net profit before:
2,400,000 × 0.026 = R$62,400. Saving:2,400,000 × 0.60 × (0.022 − 0.0022) = R$28,512.00. Uplift:28,512.00 / 62,400 = 45.7 per cent of net profit. New profit: R$90,912.00. The point of the question is that a payment decision is a profit decision. A 1.98-point fee change on 60 per cent of revenue moved net profit by nearly half without selling anything additional.
10. A national rail carries 900 transactions a second. Compare a year of it run on proof-of-work against a year on conventional rails, in energy.
Transactions a year:
900 × 365 × 86,400 = 28.38 bn. At 1,100 kWh:28.38bn × 1,100 kWh = 31,221 TWh/yr, which is 104 per cent of world electricity generation of about 30,000 TWh. At 1.5 Wh:28.38bn × 1.5 Wh = 0.0426 TWh/yr. Ratio: 733,333 times. Full marks require the caveat: proof-of-work energy is per block, not per transaction, so this is an average rather than a marginal cost. The figure without that problem is the security budget — $16.43 billion a year of issuance, $130.92 per transaction, paid by holders as dilution.
REFLECTION — eight questions, for one person and a pen
These are not for a room. Write the answers by hand if you can; the slowness is the point.
- What is the last fee you accepted without asking what it was for? What made it feel like weather rather than a price?
- Think of a time you were paid instantly and a time you waited. What actually changed in you between those two — not in your accounts, in you?
- Where in your own pricing do you charge a percentage of something whose cost to you does not rise with that something? What would a flat price do to that relationship, and who would object first?
- You carry an embedded acceptance fee in almost everything you buy — roughly £113.04 a year on £600 a month of spending, whether you pay by card or not. Does knowing that change anything you intend to do? Be honest if the answer is no.
- If you hold a rewards card, write down who you think is paying for the rewards. Then write down whether you want to keep it. Both answers are allowed.
- Recall something you believed was expensive because it was hard, and later found was expensive because somebody could. What did it take for you to see the difference?
- What in your work is currently free to the user and quietly costing somebody else? Name who, and name what happens when they stop paying.
- Where have you argued for a better system on principle, when a single ratio would have done the work? Write the ratio now.
ESSAY PROMPTS — five
Each is arguable from more than one side. Each requires at least one source the chapter cites and at least one it does not.
1. Was interchange ever justified, and is it now? Rochet and Tirole's two-sided market theory gives a genuine argument for charging merchants to subsidise cardholders while a network is being built. Argue either that the argument still holds in a mature market — because cardholder attrition remains a live risk and the subsidy is what funds fraud liability — or that it became a rent the moment cards reached universality. Use Rochet and Tirole (2006), and one empirical study of interchange regulation that the chapter does not cite.
2. The distributional case. The chapter treats Schuh, Shy and Stavins' calibration as decisive: the average cash household pays $149 a year and the average card household receives $1,133. Write the strongest critique of that calibration — its assumptions about pass-through, its vintage, its treatment of the unbanked — and then decide whether the conclusion survives the critique. Use the Boston Fed paper, and at least one reassessment or replication that the chapter does not cite.
3. Mandate, subsidy, or neither. Brazil mandated participation, India subsidised the merchant side and then ended the subsidy on 15 October 2026, and the United Kingdom did neither and reached 11.1 per cent in seventeen years. Argue for the mechanism you would choose for a country of your choosing, pricing the trade-off explicitly: a mandate coerces private institutions, a subsidy spends public money that has other uses, and doing neither is slow. Use the BIS work on fast payment systems, and one source on the political economy of payment regulation that the chapter does not cite.
4. Is the ledger case dead, or only relocated? The chapter concludes that a public instant rail delivers everything a distributed ledger promises a domestic payer, faster and cheaper, and that the ledgers retain a genuine advantage only where the settlement authority is not trusted. Argue either that this concedes too little — censorship resistance, programmability and permissionless issuance are not edge cases — or too much, in that the untrusted-authority case is smaller than its advocates claim. Use the Cambridge Centre for Alternative Finance material and the Ethereum consensus specification, and one primary source on a live cross-border corridor that the chapter does not cite.
5. The mutual that becomes the thing it replaced. The chapter's instrument is a merchant acceptance mutual funded by a 0.10 per cent levy with a five-year sunset written in on day one. Argue whether such a sunset can hold, drawing on what is known about how cooperative and mutual institutions drift once they employ people whose salaries depend on the levy. Use Ostrom (1990) on institutional design, and one history of a specific mutual or cooperative that demutualised, which the chapter does not cite.