Payment Systems · The Course Map

A learning route: a 29-chapter main line + a 2-chapter extension Main line: US domestic + cross-border. Destination: seeing exactly which square stablecoins and CPN (Circle Payments Network, the cross-border payment network built by stablecoin issuer Circle) occupy in the larger framework of payments. Extension: what is changing in 2025–2026 (agentic payments, the messaging changeover, rail interlinking, CBDC (central bank digital currency), Pay by Bank).

1. Where This Course Is Taking You¶

By the time you finish the main line, you should be able to answer these five questions without looking anything up:

Question	Which chapters solve it
When money moves from A to B, how many ledgers actually get edited along the way?	Chapters 1–3
Why does a merchant give up 2% on a card payment, and who ends up with that 2%?	Chapter 6
Why are cross-border remittances slow and expensive — slow at which step, expensive on which line?	Chapters 13–14, 18
Which piece of the old chain do stablecoins actually replace, and which piece do they not?	Chapter 27
Who is CPN competing with — SWIFT, Visa, or Wise?	Chapter 28

These five questions are the course's skeleton. Every chapter in between exists so you can derive the answers yourself rather than memorize them.

Beyond the main line, Chapters 30–31 put the changeovers under way on both sides — the payer and the rails — on the radar, judged with the same tools the main line teaches.

2. The Panorama: Four Scenarios¶

Payment has exactly one job: get money from A to B. And the method has never been hauling money anywhere — it is changing two lines in a ledger.

Once money moves between accounts, there are only two possible directions. The industry's names for them are push (credit) and pull (debit):

Dimension	Push (credit)	Pull (debit)
Who initiates	The payer: sends the instruction that delivers money into the other side's account	The payee: on authorization the payer gave in advance, takes money out of the payer's account
Typical scenes	Transfers between friends, remittances, payroll	Card purchases, subscription charges, utility auto-debits
What risk control guards against	The payer not having the balance — or being scammed into pushing the money out themselves	Unauthorized debits — which is why every pull must leave the payer a return window

Each direction comes in two scopes:

Domestic — payer and payee are in the same country, both sides answer to one rulebook, and the money finally lands on one and the same ledger.
Cross-border — the two parties are in two different countries, each with its own rules and its own ledgers, and between those ledgers there is no direct connection.

Two directions times two scopes gives four scenarios: domestic push, domestic pull, cross-border push, cross-border pull. The figure below lays them out as a 2×2 grid.

Everything in this course lands either in one of these four squares or in the foundation all four share — and the foundation is the three things Part I builds: money is an entry in a ledger, crossing institutions takes the two steps of clearing and settlement, and different kinds of money sit at different levels of a hierarchy.

"Rails," the figure's word and one of this course's most-used, means a complete system for getting money from one account to another — the rules, the participants, and the technical pipes included. Cards are a rail. Bank transfers (ACH in the US, SEPA in the euro area) are a rail. On-chain stablecoins are a rail too.

Looking at this figure, three things are worth saying out loud now:

First, pull does not mean cards. Cards are pull, but pull exists with no card in sight: direct debit — the payee, holding the payer's standing authorization, debits the payer's bank account directly; subscriptions and utility bills run on it (the US implementation is called ACH debit, the euro area's SEPA DD). The dividing line is who initiates, not what medium is used.

Second, cross-border push — bottom right in the figure — is this course's center of gravity. The domestic routes are all reasonably cheap; cross-border push costs far more — a small remittance sent through banks gives up more than 13% on average. Expensive is why most of the past decade's innovation is crammed into this one square: Wise, stablecoins, and CPN are all fighting over it.

Third, cross-border pull — bottom left in the figure — is missing half of itself. The direct debit from point one cannot cross out of a region with a single rulebook — every country or currency zone runs its own direct debit scheme, and no globally shared one exists. Which leaves cross-border pull exactly one road: the card networks.

The seven parts, mapped onto the squares:

Part	Chapters	Which square it lands in
Part I · Foundations	1–3	The base shared by all four squares: the nature of payment is editing ledgers
Part II · Cards	4–8	Pull × domestic: how the most successful payment network on earth grew
Part III · Bank Account Rails	9–12	Chapter 9 erects the push–pull axis; Chapters 10–11 fill in push × domestic (the US, then the world); Chapter 12 covers the ecosystem above the pipes and the two global channels
Part IV · Cross-Border and Regional Clearing Rails	13–19	Push × cross-border, the source of all the pain: there is no global central bank; the three chapters in the middle open one central bank ledger each (euro, renminbi, pound)
Part V · Players and Risks	20–25	Cuts across all four squares: who makes money, what kind, and how not to lose it
Part VI · New Rails	26–28	Push × cross-border again — with a different settlement asset
Main-line wrap-up	29	A full review, one question per chapter, all with answers
Extension · Frontier (optional)	30–31	Cuts across all four squares: the payer is turning into an agent, and four base layers are mid-changeover

The Part VI row deserves a second look. Stablecoins are not a fifth scenario; they are a new solution pushed into the push × cross-border square — which is why Chapter 27's conclusion reads "what gets replaced is the money-moving middle of the old chain; the FX on- and off-ramps at the two ends are not."

3. The Chapters at a Glance¶

Every chapter runs on the same fixed structure: pick up the question the previous chapter left → background and solution → a comparison table → hand off to the next chapter → self-check questions → answers. The preview below gives just three things per chapter: the question it picks up, the one-sentence solution, and the new concepts (matching each chapter's "New concepts in this chapter" line). Details defer to the chapters themselves.

Part I · Foundations (1–3)

Chapter	Question it picks up	Solution in one sentence	New concepts
1 The nature of payment	Physical delivery is too heavy, too risky, and can't cross distance	Deposit the coins with a goldsmith, and payment becomes two lines on one ledger changing together	Ledger, book transfer, atomicity, finality (planted)
2 Clearing and settlement	Two banks' ledgers don't connect	Split the job in two: clearing (work out who owes whom) and settlement (actually move the money); two modes, RTGS and DNS	Clearing, settlement, interbank obligations, netting, settlement risk, finality, RTGS and DNS
3 The hierarchy of money	Who gets to be the center, and what do they settle with	Central bank reserves — only a non-player's IOU is neutral; to judge any money, ask: who owes you? What do they redeem with, and where are the reserves?	Central bank reserves, the hierarchy of money, the liability view, segregation of customer funds, moving between layers

Part II · Cards (4–8)

Chapter	Question it picks up	Solution in one sentence	New concepts
4 The origin of cards	The merchant doesn't know you, and the banking system can't catch retail	Wedge in a middleman both sides know (the three-party model); the merchant's four-benefit table becomes the yardstick for judging every payment product	The three-party model, guaranteed payment, two-sided networks and cold start, the embryo of the merchant discount rate
5 The four-party model	One company can't do a whole country's issuing and acquiring	Outsource issuing and acquiring to thousands of banks, with the card network setting rules only; the cold-start cost gets broken apart	The four-party model, issuer, acquirer, card network, authorization and funds flowing in opposite directions
6 Interchange	Where the merchant's $2.30 goes, and by what right	Splitting $100: $1.80 to the issuer, $0.13 to the network, $0.37 to the acquiring side; money follows risk; the tier sets the rate and the name on the plastic is not the tier; US and EU rule differences produce two different card ecosystems; the bargaining chip is acceptance, not price (Amazon vs Visa)	Interchange, MDR, scheme fees, product tiers and benefit tiers, on-us vs off-us, the Durbin Amendment, the EU IFR
7 The card transaction lifecycle	When "payment successful" appears in three seconds, has money moved?	Three phases: authorization, clearing submission, settlement; authorization is not money moving; refund, void, and chargeback are three different things	Authorization, capture, void, refund, chargeback, pre-authorization, ISO 8583
8 Card not present	Online, nobody sees the card or the person — who eats the fraud?	Use liability shift to force the technology up: CVV → AVS → 3DS → tokenization; 3DS verifies the person, tokenization stops reuse	Card-not-present, liability shift, CVV, AVS, 3DS, tokenization, network tokens, DPAN

Part III · Bank Account Rails (9–12)

Chapter	Question it picks up	Solution in one sentence	New concepts
9 Push and pull	Why can't cards do payroll?	Money crosses accounts in exactly two directions: push (payer initiates) and pull (payee takes, on authorization); pull rides two media — cards and direct debit — and only cards cross borders; every pull must leave a return window	Push and pull, direct debit, debit authorization, the return window, ACH
10 The four US rails	Which scenario should take which rail?	One side-by-side table; speed, cost, and reversibility hold each other in check; the faster, the less reversible — so fraud control is forced upstream	Fedwire, RTP, FedNow, prefunded pools, fraud control moving upstream
11 Push rails around the world	The US stack has three layers — what do other countries' stacks look like?	The same three-layer recipe — batch net, high-value RTGS, instant retail — built once per currency zone; instant layers split into "real-time final" and "real-time credited"; Brazil's batch layer has already been eaten by its instant layer	TARGET2 and TIPS, HVPS and IBPS, NetsUnion, NEFT, TED and STR
12 Domestic and global	What grows on top of fragmented pipes?	The ecosystem localizes all the way up: LPMs, local gateways, payment orchestration; past the border it converges instead — only SWIFT (push) and the card networks (pull) work everywhere, each by sidestepping the job of unifying national clearing	Local payment methods (LPMs), front-end aggregation, authorization rates, data residency, the two globally universal channels

Part IV · Cross-Border and Regional Clearing Rails (13–19)

Chapter	Question it picks up	Solution in one sentence	New concepts
13 The root problem of cross-border	There is no connection between two countries' central bank ledgers	Correspondent banking: each side books locally, SWIFT only carries messages; money never crosses the border — information does	Correspondent banking, nostro and vostro, SWIFT, multi-hop fees, de-risking
14 The remittance message	What is actually inside the message?	The key fields of MT103 and pacs.008; the charge-bearer code — OUR / SHA / BEN — decides how much the payee comes up short; two routings, serial and cover; UETR and gpi; as of November 2025, MT is retired; the final section answers why cards don't take this road	MT103 and pacs.008, charge bearer, serial and cover, UETR, SWIFT gpi
15 Euro-area clearing rails	The message arrived — now how does the money move on the ledger at the other end?	Four questions fix a rail's character: what money settles it, how it nets, when it stops being reversible, who can join directly; large-value runs on T2, instant on TIPS, both in central bank money; SEPA is a rulebook, not a system; instant payments are now a legal obligation, not a commercial choice	The four questions, T2's two components, the Settlement Finality Directive, EURO1 and the single obligation structure, scheme layer vs CSM layer, the IPR, Verification of Payee
16 China's clearing rails	Same four questions — how does China answer them?	CNAPS is a set of subsystems, not one system; China has an extra layer of non-bank payment institutions, and severing direct bank links plus centralized custody of customer funds pushed that layer back from quasi-bank to pipe; renminbi cross-border runs on three coexisting roads	CNAPS, the payment system bank code, severing direct links, centralized custody of customer funds, stored-value account operation vs payment transaction processing, the clearing bank model, the CIPS zero-balance account, the cross-boundary payment link
17 UK clearing rails	One currency, one central bank, one legal system, one time zone — so why four rails?	The causal rule: count how many combinations a market cuts out of amount, urgency, predictability, and refund rights; FPS is fast to credit, CHAPS is fast to finality; the NPA is dead, RPIB now owns the design	The causal rule, the three finality points of CHAPS, RT2, the Bacs three-day cycle and the Direct Debit Guarantee, FPS prefunding and the three access tiers, the NPA termination and RPIB, CoP and mandatory APP reimbursement, omnibus accounts
18 The four cost sources	$1,000 loses 7.4% on the way — where did it go?	Four items: FX spread, intermediary bank fees, prefunded liquidity, compliance checks; the largest one is the one users never see	The mid-market rate, FX spread, the cost of prefunded liquidity, the cost of compliance checks, corridors
19 The Wise model	Can correspondent banking be skipped?	Local-in, local-out + internal netting; trades "slow and expensive" for "tied-up capital plus licenses"; one yardstick: where did the cost move from, and where to	Local-in local-out, internal netting, rebalancing

Part V · Players and Risks (20–25)

Chapter	Question it picks up	Solution in one sentence	New concepts
20 The role map	Behind every mechanism, who is charging?	The panorama: every company is a contractor on the four-party skeleton, and online differs from offline by exactly one square — the acceptance front end; three questions place any company: who holds the license, who touches the money, who eats the loss	The acceptance front end (POS and gateway), processors (issuing side and acquiring side), BIN sponsor, PSP, PayFac, ISO, BaaS, payment orchestration, neobank, licensed vs unlicensed
21 The case studies	The grid is in place — how do you read an actual company with it?	Nine companies on four-dimension cards, read in pairs — Wise vs Airwallex, Revolut vs Chime, Adyen vs Checkout.com, Bridge vs BVNK — plus Fiat24 on its own; three patterns across the lot	None — this chapter only uses tools already built
22 Neobank anatomy	With no license, where is the user's money?	FBO accounts + an internal ledger; the Synapse case: when the ledger breaks, the money might as well not exist	FBO accounts, the internal ledger, pass-through deposit insurance eligibility, float income, net interest margin
23 The five risks	When something goes wrong, which bucket does it go in first?	A five-way split: credit, fraud, settlement, liquidity, compliance; classify first, then pick the tool	Rolling reserves, delayed settlement, liquidity risk
24 The compliance skeleton	What does compliance actually consist of?	Three blocks: admission (licenses), identity (KYC), monitoring (screening); read licenses in three tiers by how deeply you touch the money, compared across six jurisdictions; compliance is a moat, and it can't be bought	The three-tier reading of licenses, MTL, MSB, PI, EMI, passporting, China's Payment Business License, the PS Act, SVF, KYC, KYB, UBO, OFAC, SAR, the Travel Rule
25 Ledgers and reconciliation	How do you keep the books forever self-consistent and matched?	Double-entry bookkeeping + daily reconciliation driving breaks to zero; three system properties: idempotency, state machines, asynchronous callbacks	Double-entry bookkeeping, journal entries, reconciliation, breaks, idempotency, state machines, asynchronous callbacks

Part VI · New Rails (26–28) and the Main-Line Wrap-Up (29)

Chapter	Question it picks up	Solution in one sentence	New concepts
26 What stablecoins are	Whose liability is it?	No innovation in the hierarchy of money (same layer as a wallet balance); the innovation is the shared ledger: clearing and settlement complete in one act; the price is a downgraded settlement asset	Stablecoins, reserve assets, the shared ledger, on-chain settlement, tokenized deposits
27 What stablecoins solve	Of the four cost items, how many do they cut?	Kills intermediary bank fees and sharply eases prefunding; can't touch the spread or compliance; what gets replaced is the middle segment, not the whole chain	On-ramp / off-ramp, on-chain address screening, the last mile
28 CPN	Who solves the N² problem of on/off-ramp channels?	One rulebook plus one network of participants — the seat Visa holds; SWIFT's messaging layer with a settlement layer of its own; the deciding factor is neutrality	Network neutrality
29 The final review	Did the main line actually stick?	Twenty-eight questions, one per chapter, plus one integrative question, all with answers; whichever one stalls you names the chapter to reread	None

Extension · Frontier (30–31, Optional) and the Appendix

Chapter	Question it picks up	Solution in one sentence	New concepts
30 Agentic payments	The payer is no longer human, and three old assumptions fail	At the authorization layer, scoped credentials (tokenization, narrowed); at the settlement layer, a division of labor: shopping on a human's behalf rides cards, machine-to-machine rides stablecoins; the "who sets the rules" seat opens up for the third time	Agentic commerce, scoped payment credentials, x402, machine-to-machine payments
31 The frontier radar	Four base layers on the supply side are changing over at once	ISO 20022 swaps the messages; instant-payment interlinking is local-in local-out run as public infrastructure; wholesale CBDC is a reshuffle of the hierarchy; Pay by Bank hinges on institutions and incentives	ISO 20022, instant payment interlinking, wholesale and retail CBDC, Pay by Bank
Appendix A (32)	How do all the rails covered stack up side by side?	Every rail placed on the remittance/purchase × domestic/cross-border grid, all answering the same four questions — clearing, settlement, settlement asset, finality; not part of the main line	None

Previous: (none — this is the starting point) Next: Chapter 1 · The Nature of Payment: Money Never Actually Moves