Appendix A: The Four-Quadrant Solution Comparison
Purpose: take every rail this course has covered and set it into one 2×2 — by "who initiates" and "does it cross a border" — with every solution answering the same question: where do clearing and settlement land. Builds on: Chapter 2 (clearing and settlement, finality, RTGS and DNS), Chapter 3 (the hierarchy of money, central bank reserves). This is an appendix: outside the main line, placed after the two optional chapters. It is a horizontal index only and introduces no new judgment tools. The few solution names in the tables that never appeared in the main text (Visa Direct, mBridge, and the like) are examples only and are not in the final review.
1. First, a Measuring Stick¶
More than twenty solutions appear below. They look wildly varied, but measured with the same set of questions, the differences collapse into just a few kinds. Every time a new solution shows up, ask these four:
| # | Question | Why it matters |
|---|---|---|
| 1 | On whose books is clearing computed? | Decides who holds the rule-making power |
| 2 | On whose ledger does settlement land? | Decides which tier the money ultimately rests on |
| 3 | Whose liability is the settlement asset? | Decides how hard the money is (the hierarchy of money) |
| 4 | When does finality happen — and can the money be clawed back afterward? | Decides who carries the risk in the meantime |
The fourth question is the easiest to skip and the sharpest at telling solutions apart. Arrival is not finality — the layer the user sees and the layer between institutions can be hours or even days apart.
Nail down the two axes first, or the four quadrants will bleed into one another:
- Remittance (push / credit): the payer initiates; money is pushed from the payer's account toward the payee.
- Spending (pull / debit): the payee (the merchant) initiates, requesting a debit against the payer's account.
"Push" and "pull" describe who initiates, not which way the funds flow. Cards are the textbook pull, yet the funds still flow from issuer to acquirer to merchant — the authorization path and the money path run in opposite directions.
What payment, clearing, and settlement each ask
The three words get used interchangeably all the time, but they happen at different times, at different institutions, on different ledgers:
| Dimension | The question it answers | Where you can see it |
|---|---|---|
| Payment | What did the payer do; what does the instruction look like | The user interface and the message format |
| Clearing | Who is computing who owes whom, and what the computation produces | The clearing house's books |
| Settlement | On which ledger the money actually moves, and from what moment it can't be undone | The central bank's ledger, or some shared ledger |
The one-line mnemonic: payment is the action, clearing is the counting, settlement is the money moving. Every solution below is laid out along these three, so the differences compare at a glance.
2. Quadrant 1: Domestic + Remittance (Push)¶
This quadrant is the baseline for the whole course; the other three are variations on it.
| Solution | Clearing: on whose books? | Settlement: on whose ledger? | Settlement asset | Finality and reversibility |
|---|---|---|---|---|
| Book transfer | No interbank clearing happens; two lines change together on the bank's own ledger | No settlement needed — payer and payee are owed by the same bank | That bank's deposit liability | Instantly final |
| Batch net (ACH, SEPA SCT, China's BEPS) | The clearing house accumulates batches and computes multilateral net positions | Net amounts move across central bank reserve accounts at end of day | Central bank reserves | Settled but not final: no finality until the return window closes |
| Large-value RTGS (Fedwire, TARGET2, CHAPS, China's HVPS) | Transaction by transaction, no netting | Each payment moves gross across central bank reserve accounts | Central bank reserves | Instantly final, irrevocable |
| Instant rails · central-bank-ledger type (FedNow, Pix, SCT Inst via TIPS) | Transaction by transaction, in real time | Each payment moves across central bank accounts in real time | Central bank reserves | Instantly final, irrevocable |
| Instant rails · prefunded-pool type (US RTP) | Transaction by transaction, in real time | Real-time transfers inside a shared pool run by The Clearing House; the pool sits at the Fed | Funds pledged into the joint account — not counted in the bank's own reserves, earning no interest | Instantly final, irrevocable |
| Instant rails · deferred-net type (India's UPI, China's IBPS) | Transaction by transaction, in real time | The clearing house nets first, then settles in batches through the central bank's RTGS | Central bank reserves, but deferred to the settlement window | Seconds on the user side, not yet final between institutions; the receiving bank carries intraday exposure |
| Wallet internal ledgers (Alipay / WeChat Pay balances, neobank internal transfers) | The operator's own ledger | No cross-institution settlement happens; the money never leaves one customer-funds account or FBO account (For Benefit Of — a pooled account held for the customers' benefit) | A wallet balance is the operator's liability; a neobank balance is nominally the partner bank's liability, with ownership proven by the internal ledger (see neobank anatomy) | Instantly final internally — but what you hold is not a bank deposit in your own name |
Three causal rules — take them and measure any new domestic rail that appears:
One: within the instant rails, whether arrival is final comes down to a single check — is the money for the interbank leg already in place? Rows 4 and 5 both move money that is already there; in row 6 the receiving bank fronts the funds and squares up afterward, so a window opens between clearing and settlement. RTP makes the point best: it doesn't touch the central bank's ledger, yet it is still final — so the key variable is whether the money is in place, not whether the central bank's ledger is involved. The non-finality in the batch-net row has a different cause — the return window, not absent money.
Two: the faster, the less reversible. Batch net comes with return windows measured in tens of days; RTGS and the instant rails essentially cannot be undone. This is not designer laziness — a return window is only possible while the money hasn't truly moved. Once each payment settles in real time, reversal means clawing money back out of the payee's account. The price: fraud defense has to move from after the fact to before it.
Three: the wallet row does not sit on the same tier. The settlement asset in rows 2 through 6 is a central bank liability; row 1 has no cross-institution settlement, so "settlement asset" doesn't apply; the last row is a private company's liability. In user experience it is the fastest; in the hierarchy of money it is the softest. The row earns its place in the table because it comes back later under two other names: the neobank's internal ledger, and stablecoins.
3. Quadrant 2: Domestic + Spending (Pull)¶
| Solution | Clearing: on whose books? | Settlement: on whose ledger? | Settlement asset | Finality and reversibility |
|---|---|---|---|---|
| Four-party-model cards (Visa, Mastercard) | The network computes multilateral net positions at end of day | Issuers pay acquirers the net amounts, moved across reserve accounts via Fedwire or the Fed's National Settlement Service (NSS); the money never enters the network's own books | Central bank reserves (for the member-bank leg) | Merchant paid T+1 to T+2; chargeback window 120+ days |
| Three-party-model cards (Amex, Discover) | One company both issues and acquires; clearing is an internal ledger | The company pays the merchant directly | That company's liability, until the payout lands | Same order of magnitude; disputes decided by a single party |
| Direct debit (ACH debit, SEPA DD, UK Direct Debit) | Batch net | Central bank ledger at end of day | Central bank reserves | Very long return windows: unauthorized consumer ACH up to 60 days; SEPA DD gives consumers an 8-week no-questions refund |
| Pay by Bank / merchant payments on instant rails (Pix merchant codes, UPI merchant codes, iDEAL) | Exactly the same rails as their namesakes in Quadrant 1 | Same as above | Central bank reserves | Instantly final, essentially irrevocable |
| Wallet QR payments (WeChat Pay, Alipay) | Balance payments run on the wallet's internal ledger; the linked-card portion goes through NetsUnion and on to the banks | Splits two ways by funding source | Balance is the operator's liability; linked card is a bank deposit | Instant internally; the linked-card leg follows the rail beneath it |
This quadrant holds one counterintuitive fact, worth its own paragraph:
Row 4 sits in the "pull" quadrant, but mechanically it is a push.
Pix merchant codes, UPI merchant payments, iDEAL: in every one, the user confirms and initiates inside their own banking app. The merchant merely displays a payment request; the instruction to move money comes from the payer.
That explains two things you can observe in the market: merchants love it (low fees, instant finality, no chargebacks), and consumers are wary of it (pay the wrong person and the money is gone). One mechanical fact, two conclusions depending on which side you stand.
As for why cards can charge 2% while nothing else in this quadrant can — that is an institutional question about interchange, not a mechanical question about clearing and settlement. Go back to interchange.
4. Quadrant 3: Cross-Border + Remittance (Push)¶
Put this quadrant's root constraint up front, or every solution below will look like it was invented out of thin air:
No connection exists between two countries' central bank ledgers, and there is no global central bank. Every solution here works around this fact. None of them solves it.
| Solution | Clearing: on whose books? | Settlement: on whose ledger? | Settlement asset | Finality and reversibility |
|---|---|---|---|---|
| Correspondent banking + SWIFT | No unified clearing exists; each hop clears locally on its own | Each hop lands on a local central bank ledger, or on the correspondent's nostro / vostro accounts | Switches hop by hop: local central bank reserves, or the correspondent's deposit liability | Final hop by hop; 1–5 business days end to end; arrival amount uncertain (fees deducted along the hops) |
| Local-in, local-out (Wise, Airwallex) | Netted on the provider's internal ledger | Each end runs one local clearing and settlement; the cross-border piece is the provider periodically rebalancing its own funds | Both ends: local central bank reserves; the middle leg: the provider's own position | User side can be instant (the provider fronts the money); the books between institutions are squared afterward |
| Intra-bank cross-border at a multinational (HSBC Hong Kong → HSBC UK) | The bank's own internal ledger | Never leaves the bank; the branches in the two countries adjust internal positions | That bank's liability | Instantly final — but only covers the bank's own customers |
| CIPS (cross-border RMB) | Cleared transaction by transaction inside the system | Ultimately lands on China's central bank ledger (via HVPS) | RMB, central bank money | Final per transaction; RMB only, with indirect participants connecting through direct ones (correspondent banking embedded inside) |
| Card network push (Visa Direct, Mastercard Send) | Network netting + currency conversion | Member banks pay one another through a settlement bank | Bank deposits in the settlement currency | Receiving side: minutes to days, depending on the receiving issuer |
| Stablecoins (USDC / USDT over a chain) | The chain's consensus is the clearing | Settlement completes in the same transaction | The issuer's liability | Final once the chain confirms, with timing set by the consensus mechanism; the on- and off-ramps at each end still run through Quadrant 1 |
| Instant-rail interlinking (PayNow–UPI, Nexus) | Each country's instant rail runs its own leg | Each lands on its own country's central bank ledger | Local central bank reserves at both ends | Seconds on the user side; the FX and position layer in the middle is carried by the settlement arrangement |
| Wholesale CBDC / mBridge | Rearranged directly at the central bank tier | A ledger shared by several central banks | Central bank liabilities | Pilot stage; not yet at scale |
This quadrant has only one causal rule, but it measures every new solution:
Cross-border solutions differ in one thing only: who holds accounts on both sides at once.
In correspondent banking it is the correspondent; in the Wise model it is Wise itself; in a multinational bank it is the bank's own two branches; in the card networks it is the network's membership; in stablecoins it is the chain. Find that role, and the rest — fee structure, speed, failure modes — can all be derived.
It also explains the line this course keeps repeating: the money doesn't cross the border; the information does. Outside the stablecoin and CBDC rows, the funds in every other row move locally, inside their own borders.
5. Quadrant 4: Cross-Border + Spending (Pull)¶
This is the most counterintuitive of the four quadrants, because its mainstream answer is not "a cross-border solution":
The mainstream move reduces cross-border spending to "local spending + cross-border repatriation" — Quadrant 2 plus Quadrant 3, rather than inventing a Quadrant 4.
| Solution | Clearing: on whose books? | Settlement: on whose ledger? | Settlement asset | Finality and reversibility |
|---|---|---|---|---|
| International card transactions (Visa, Mastercard) | Network netting, with currency conversion in the same step | Issuers pay acquirers the net amounts through a settlement bank, in the settlement currency | Bank deposits in the settlement currency | Merchant T+1 to T+2; chargeback window 120+ days |
| Local acquiring entities (Adyen, Stripe, dLocal setting up in-country) | Collapses into Quadrant 2: cleared by the local card network or local rails | Local central bank ledger | Local central bank reserves | The cross-border part is deferred to the merchant repatriation step, which runs through Quadrant 3 |
| Cross-border acceptance of local payment methods (a PSP fronting iDEAL, Pix) | Entirely local clearing | Local central bank ledger | Local central bank reserves | Same as above; the border is crossed at repatriation |
| Stablecoin collection (Bridge, BVNK, for merchants) | Cleared on-chain | Settled on-chain in the same moment | Issuer's liability | Final on chain confirmation, no chargebacks; converting to local currency takes an off-ramp, whose cost and availability vary widely by country (see the last mile in the chapter on what stablecoins solve) |
| Wallet cross-border acceptance (Alipay, WeChat Pay at overseas merchants) | The user side is debited on the domestic wallet ledger | The overseas merchant is settled in local currency by a local acquirer | User side: the operator's liability; merchant side: local bank deposits | Instant on the user side; FX and the cross-border position are carried by the wallet operator |
Three things worth remembering:
One: only rows 1 and 4 are true cross-border acceptance. The other three rows all push the border out to the repatriation leg. There is a reason — a local issuer's card processed through a foreign channel gets declined noticeably more often, and to a large e-commerce merchant, a few percentage points of approval rate matter far more than a few basis points of fees.
Two: what makes the stablecoin row special is not speed — it is chargebacks. It is the only option in this table with no chargeback mechanism in a cross-border setting. For merchants, a virtue; for consumers, a hole — and one of the main barriers keeping it out of consumer retail acceptance.
Three: at small ticket sizes, a cross-border card payment beats a wire on both speed and cost — and not because the technology is stronger. The network nets in the middle, compressing thousands upon thousands of cross-border transactions into one daily net amount between institutions. N cross-border payments squeezed into 1 — that is the entire secret.
6. The Four Quadrants in One View¶
| Remittance (push) | Spending (pull) | |
|---|---|---|
| Domestic | Clearing and settlement in standard form: batch net, RTGS, or instant rails — pick one of three | Rides on the domestic rails, with a layer of acceptance and dispute rules on top |
| Cross-border | Every difference traces to "who holds accounts on both sides" | Mainstream: reduced to domestic spending + cross-border repatriation |
Three rules that run through the whole table:
| Rule | What it says |
|---|---|
| Clearing and settlement can separate — and someone carries the risk for exactly as long as they stay apart | Zero separation: RTGS, Pix, RTP, on-chain. Longest: multi-hop correspondent banking (1–5 business days). Return windows are a separate axis: money can still be pulled back after settlement completes |
| The settlement asset's tier decides how hard the money is | Central bank reserves > commercial bank deposits > stablecoins and wallet balances |
| Only two run worldwide | Push: SWIFT plus correspondent banking. Pull: Visa and Mastercard. Each sidesteps having to unify national clearing |
Finally, back to the measuring stick. Put any new solution in front of you and ask: on whose books is clearing computed, on whose ledger does settlement land, whose liability is the settlement asset, when does finality happen. Answer all four, and which quadrant of the grid it belongs to — and where its costs hide — falls right out.
7. Why Domestic Spending Grew Differently in Six Countries¶
Inside the domestic-spending quadrant, six markets run very different clearing and settlement paths. The differences are not random; every one traces back to an institutional or structural cause. The method is one sentence: when you see a market difference, look for the institutional difference first.
| Country | The visible difference | The cause |
|---|---|---|
| China | UnionPay handles cards, NetsUnion handles wallets — two networks running in parallel | Non-bank players once ran private direct connections into every bank; in 2017 NetsUnion pulled that into a single hub. One network per type of institution — the thing that most sets China apart |
| US | Cards keep account-to-account down; the strongest cashback-and-rewards culture anywhere | The Durbin Amendment capped only debit interchange; credit cards were left alone, so issuing-side revenue never took the hit |
| Europe | Local debit networks and account-to-account have room to live | The IFR caps interchange (0.2% debit, 0.3% credit); with card revenue squeezed, other methods can finally compete |
| Japan | An extra prepaid layer on top of cards and debit | Cash use stayed high for decades, and the government pushed cashless with rebate campaigns; convenience stores and transit IC cards (Suica) form a system of their own |
| Brazil | Pix reached the whole country in two years | The central bank could force institutions above a size threshold to connect, and the incumbent card ecosystem was weak. Both conditions held at once: the power to mandate adoption, and an incumbent not good enough to defend itself |
| Philippines | Wallets spread first; banks connected later | Thin branch coverage and a large unbanked population meant wallets simply skipped the bank layer — most people's first financial account was GCash |
Read the six rows vertically and the causes collapse into three types:
| Cause type | What it decides | Where it shows up |
|---|---|---|
| How regulators treat interchange | Cards' revenue headroom — and with it, whether cards can keep other methods down | US, Europe |
| The central bank's power to mandate, and its will to push | Whether a new rail can spread fast | Brazil, China |
| How far the incumbent infrastructure reaches | Whether the new method stacks on the old system or skips it | Philippines, Japan |
These three travel beyond the table. Given a new market, ask how its regulators treat interchange, whether the central bank can force adoption, and how far the incumbent bank-and-card coverage reaches — those three answers largely fix the shape of its payments.
8. Hierarchy Text You Can Paste Straight into a Mind Map¶
The four quadrants of payment solutions
├─ The measuring stick (four questions)
│ ├─ On whose books is clearing computed
│ ├─ On whose ledger does settlement land
│ ├─ Whose liability is the settlement asset
│ └─ When does finality happen / reversal window
├─ Domestic · Remittance (push)
│ ├─ Book transfer | internal ledger | no settlement needed | instantly final
│ ├─ Batch net ACH/SEPA SCT | multilateral netting | central bank reserves | not final until return window closes
│ ├─ Large-value RTGS Fedwire/TARGET2 | gross, per transaction | central bank reserves | instantly final
│ ├─ Instant · central-bank-ledger type FedNow/Pix | real time per transaction | central bank reserves | instantly final
│ ├─ Instant · prefunded-pool type RTP | real time per transaction | central bank reserves in the pool | instantly final
│ ├─ Instant · deferred-net type UPI/IBPS | real-time clearing, deferred settlement | receiving bank carries intraday exposure
│ └─ Wallet internal ledger | operator's ledger | no cross-institution settlement | operator's liability
├─ Domestic · Spending (pull)
│ ├─ Four-party model Visa/MC | network netting | issuer → acquirer | chargebacks 120+ days
│ ├─ Three-party model Amex/Discover | internal clearing | company pays merchant directly
│ ├─ Direct debit ACH debit/SEPA DD | batch net | unauthorized ACH 60 days, SEPA DD 8 weeks
│ ├─ Pay by Bank / merchant codes | mechanically a push | instantly final, no chargebacks
│ └─ Wallet QR | balance on internal ledger / linked card via NetsUnion
├─ Cross-border · Remittance (push) ※ root constraint: no global central bank
│ ├─ Correspondent banking + SWIFT | local clearing hop by hop | nostro/vostro | 1–5 business days, fees per hop
│ ├─ Local-in local-out Wise/Airwallex | internal netting | local at both ends | capital tied up to buy speed
│ ├─ Intra-bank cross-border at a multinational | internal ledger | instant | own customers only
│ ├─ Card network push Visa Direct | network netting | minutes to days
│ ├─ Stablecoins | clearing and settlement merged on-chain | issuer's liability | on/off-ramps still local
│ ├─ Instant-rail interlinking PayNow–UPI/Nexus | each end runs its own local rail
│ └─ Wholesale CBDC / mBridge | rearranged at the central bank tier | in pilot
│ └─ Causal rule: the only difference is "who holds accounts on both sides"
└─ Cross-border · Spending (pull) ※ mainstream: reduced to domestic spending + cross-border repatriation
├─ International card cross-border | network netting + FX | settlement still via the banking system
├─ Local acquiring entities Adyen/Stripe/dLocal | collapses into domestic spending
├─ LPM cross-border acceptance | entirely local clearing
├─ Stablecoin collection Bridge/BVNK | the only no-chargeback option
└─ Wallet cross-border acceptance | debited at home + settled locally abroad
Previous: Chapter 31 · Frontier Radar: The Four Base Layers Being Replaced Next: (none — this is the end of the course)