Where you are. Lesson 11 drew the four-party square: issuer, acquirer, network, merchant, an authorisation in seconds and a settlement days later, netted like the batch rails of lessons 7 and 8. You can name every party and every message. What you have not yet followed is the money the square keeps for itself: every card sale leaves a little behind, and the split of that little is the entire economics of the card industry - why rewards exist, why merchants lobby, and why two continents wrote laws about a fee most people have never heard of. This lesson takes one stylised purchase and splits it to the cent.
Follow the missing 1.50
You tap for a coffee. The terminal chirps its approval in under a second - lesson 11’s authorisation - and the price is 100. Stylised, as always in this course: round numbers so every split reads at sight. Your statement will show 100, to the cent. The shop’s side tells a different story. When settlement lands, days later and netted, its account grows by 98.50. No invoice explains the gap; there is no line item, no receipt for the missing 1.50, nothing the barista could point at. Four companies took a cut between the tap and the deposit, in fixed proportions, under a published schedule. This lesson follows every one of the missing cents until each sits in a named pocket.
The idea in one paragraph
The gap between 100.00 tapped and 98.50 received is the merchant discount - the all-in price the merchant pays for accepting the card - and it splits three ways across lesson 11’s square. The biggest slice is interchange: a fee the acquirer pays the issuer on every sale, at a rate the network publishes, which funds the cardholder side of the system - the rewards, and the fraud losses the issuer absorbs. A thin slice is scheme fees: the network’s own charge, collected from both banks on every tap. The rest is the acquirer’s margin. The cardholder pays none of it directly, and is in fact paid points to keep tapping - which is the tell for the whole design. The merchant side funds the cardholder side, merchants price that cost into everything they sell, and the one big flow, interchange, is where regulators aim when they decide the system charges too much.
The stack inside the gap
Follow the 1.50 in the order it is carved:
- The acquirer collects the full 1.50 from the merchant’s takings: the merchant discount, one number on the shop’s statement.
- It passes 1.00 of that to the issuer: interchange, at the rate the network’s schedule sets for this kind of card and this kind of shop.
- Each bank pays the network 0.10 for running the square: scheme fees, 0.20 in all, charged to both sides of every tap.
- The acquirer keeps the last 0.40: its margin, covering terminals, fraud screening and the days it waits for settlement.
Now net it out per party. The issuer takes 1.00 in and pays its 0.10 scheme fee out: 0.90. The network collects 0.10 from each side: 0.20. The acquirer keeps its 0.40. The merchant keeps 98.50. Check the conservation law: 98.50 + 0.90 + 0.20 + 0.40 is exactly 100.00. Fees reallocate the tap; they never leak a cent of it.
One mechanical detail matters: interchange never travels as its own payment. It is deducted at source, inside lesson 11’s settlement. When the netted flows land, the issuer funds 99.00 of the 100.00 sale - the purchase minus its interchange - and the acquirer credits the merchant 98.50 after taking the rest of the stack; the network invoices both members for its scheme fees separately. Nobody wires a 1.00 fee anywhere. It is lesson 7’s netting move again, this time netting a fee against the money it rides on.
Why the fee flows toward the issuer
Every other fee in this module flowed toward whoever did the work: the rail, the operator, the network. Interchange is stranger - it flows past the network, from one member bank to another. The reason is that the issuer carries the expensive side of the square. It puts the card in the wallet, funds the rewards that keep it there, extends the float between purchase and repayment on credit cards, and absorbs most of the fraud when a stolen card buys something. Interchange is what makes carrying all of that a business. Without it, issuing cards is a cost centre; with it, every tap in the economy pays the cardholder’s bank a royalty.
That analogy also answers the obvious question: why would the network set high a fee it never receives? Because interchange is the network’s recruiting budget, spent with other people’s money. A richer schedule makes issuing this network’s cards more attractive than issuing a rival’s; more cards issued means more taps, and every tap pays the network its scheme fees. The fee is set by a party that does not pay it and paid by a party that did not set it - a structure with no natural brake, which is precisely why regulators became the brake.
The lever regulators pull
The invariant is everywhere the same: interchange flows from the merchant’s bank to the cardholder’s bank, per sale, per the network’s schedule. What differs by jurisdiction is the wrapper - whether that flow is capped, and for whom. The two big regimes answer differently.
Treat the stylised tap as dollars for a moment and run the caps over it: a 100.00 debit purchase carries at most about 0.26 of interchange in the US and 0.20 in the EU, and the EU credit cap allows 0.30. Our stylised 1.00 sits above every one of those caps - a rate only possible where the fee is uncapped. Real schedules are nothing like our single constant, either: they are long published tables that vary by card type, merchant category and channel, which is why the exercise keeps one stylised row and the figure shows one bar.
Check yourself
1. The cardholder taps 100.00 and pays exactly 100.00. Trace the issuer’s 0.90: who ultimately hands it over, and along what path?
The merchant. The acquirer collects the 1.50 merchant discount from the shop’s takings, passes 1.00 of it to the issuer as interchange, and the issuer pays 0.10 of that to the network as its scheme fee, netting 0.90. No fee ever appears on the cardholder’s side of the square; the whole stack is funded from the merchant’s shortfall - which the merchant, in turn, prices into what it sells.
2. The network sets interchange but receives none of it; its own take is the 0.20 of scheme fees. Why would it set high a fee that flows straight past it?
Because interchange is the network’s recruiting budget, spent with other people’s money. Issuers earn interchange, so a richer schedule makes issuing this network’s cards more attractive than a rival’s; more cards in wallets means more taps, and every tap pays the network its scheme fees from both sides. The network tunes a transfer between two other parties to win the side it must win - the issuers - knowing merchants can rarely refuse to accept.
3. No 1.00 fee payment ever crosses a ledger. Where does interchange physically live in lesson 11’s settlement?
As a deduction at source. The issuer settles 99.00 into the netted interbank flow instead of 100.00, keeping its interchange, and the acquirer credits the merchant 98.50 after taking the scheme fee and its margin. Interchange is an adjustment to the settlement leg, netted against the purchase money it rides on - lesson 7’s netting move, applied to a fee.
4. A customer pays cash for the same 100 coffee. What part of the card stack do they still fund, and what do they never receive?
The shop sets one shelf price to cover its costs, and the merchant discount on card sales is one of those costs - so the cash payer funds a share of the interchange the shop hands over on everyone else’s taps. What they never receive is the reward: points flow only to cardholders, out of issuer revenue the whole queue financed. That asymmetry is a large part of why regulators treat interchange as their lever.
Do this
Ten minutes, from module-02-domestic-rails. Open code/fee_split.py: the constants at the top are this lesson’s stylised stack in integer cents - fee arithmetic must land exactly, and floats drift - and main already asserts the conservation law and prints the fee breakdown. split is yours. Work the TODO(you) and return each party’s net take from one tap: the merchant after all three cuts, the issuer’s interchange minus its own scheme fee, the network’s fee collected from both sides, the acquirer’s margin. The four asserts are the spec: every cent lands somewhere, the merchant funds the system, the issuer’s revenue is not its profit, and the network charges both sides.
python3 code/fee_split.pyDone right, the fee breakdown prints and the run ends with the line
the merchant keeps 98.50 of 100.00: interchange makes the merchant's side fund the cardholder's bank, and that flow is the lever regulators capIf the conservation assert fires, you have probably charged the issuer’s scheme fee to the merchant a second time: it comes out of interchange, not out of the merchant’s line. The completed version is solutions/fee_split.py; compare after you are green.
What you can now do. You can take one card purchase and split it to the cent: name the merchant discount, carve it into interchange, scheme fees and acquirer margin, and say which party nets what and why the big flow points from the merchant’s bank to the cardholder’s bank. You can say who ultimately pays for card rewards - every shopper, through prices set to cover the merchant discount, including the ones paying cash - and you can read the EU’s percentage caps and the US’s debit-only cap as two grips on the same lever. The split assumed one thing: that the sale was good. Lesson 13 is what happens when it was not - a payment disputed weeks after settlement, on a rail where nothing is ever deleted - and the machinery that decides which party in the square eats the loss.