Part IV — Digital Money

Chapter 23 · CBDC Design Choices

Chapter 22 ended on the money flower and a promise: that the empty petals were where the interesting engineering lived. This chapter walks the petal a central bank actually chose. Every CBDC decision — retail or wholesale, direct or two-tier, token or account, programmable or free — is a trade between adoption and control, between privacy and traceability, between a payment that works and a banking system that survives it. India's e-rupee and Europe's digital euro are the worked examples, and their designers disagree.

📖 Chapter 23 of 35 ⏱ ~65 min read 🎓 Session 29 · CILO 4 Core 75% Stable 25% Volatile 0% 📝 Quiz · 30 MCQs

🎯 Learning Objectives

1. The Design Space

A CBDC is not one product; it is a family of products parameterised by a handful of choices. Get the vocabulary straight and the arguments write themselves. Five axes span the space:

  1. Retail or wholesale — who may hold it: everyone, or financial institutions only?
  2. Direct or two-tier — does the central bank face every citizen, or issue through intermediaries?
  3. Token or account — is value held in the instrument (cash-like) or in a registered identity-bound balance?
  4. Programmable or free — can rules be attached to where, when and on what the money spends?
  5. Offline or online-only — must it work without connectivity, accepting the double-spending risk that follows?

No combination dominates. That is the central lesson of this chapter, and the reason the trade-off matrix of the hands-on is the session's actual deliverable. Every axis you push toward control (traceable, capped, programmable, account-based) costs adoption. Every axis you push toward adoption (private, capless, offline, token-like) costs control and — the danger nobody can ignore — disintermediates the banks, because a risk-free central bank liability that is too attractive pulls deposits out of the institutions that lend. The design problem is not building a digital rupee. UPI already moves rupees digitally and magnificently. The design problem is building one that citizens adopt and banks survive.

📝
Why bother, with UPI already there? The RBI's own answer, from the concept note onward: UPI moves commercial bank money. The e-rupee is a settlement asset with no counterparty, a defence of monetary sovereignty in a world of foreign CBDCs and dollar stablecoins, an offline instrument for the unconnected, and a programmable rail for purpose-bound transfers. Whether those uses justify the build is precisely what Chapter 24's evidence will test.

2. Retail or Wholesale

Retail CBDC is for the public: wallets on phones, P2P and P2M payments, the digital analogue of the note in your pocket. Wholesale CBDC is for financial institutions: the digital analogue of the reserve balance, used to settle high-value interbank transactions. The two share an issuer and almost nothing else. Retail lives or dies on adoption — merchants, wallets, habit. Wholesale lives or dies on integration — does it settle faster, cheaper, or with less risk than the existing RTGS?

India runs both, and the asymmetry is instructive. The wholesale pilot (from November 2022) settled secondary-market government securities trades — delivery-versus-payment in central bank money, removing counterparty exposure — and has since extended to the Unified Markets Interface with tokenised certificates of deposit. Its success is measured in basis points of settlement efficiency, invisible to citizens. The retail pilot (from December 2022) is measured in users and daily transactions, and is where the adoption problem of Chapter 24 lives. A central bank that cannot solve retail adoption can still have a perfectly useful wholesale CBDC — and several effectively do.

3. Direct or Two-Tier

A direct CBDC makes the central bank the ledger-keeper for every citizen: it issues, it onboards, it runs wallets, it handles lost phones. No serious project chose this. A two-tier (indirect or intermediated) CBDC keeps the central bank as issuer and liability-bearer while banks and licensed non-banks do everything that touches the customer — KYC, wallets, support. The RBI's concept note said plainly that the central bank has "no comparative advantage" in customer-facing tasks, and a direct model would disrupt the financial system while burying the RBI under onboarding. The e-rupee is two-tier: the RBI creates tokens, nineteen banks and a growing set of non-bank providers distribute them.

Two-tier distribution is the first disintermediation control, but not the last. The standard trio:

  1. No interest. An interest-bearing CBDC is a risk-free deposit account at the central bank, and would drain bank funding in any stress episode. The e-rupee pays nothing — exactly like cash, and exactly like the digital euro is planned to.
  2. Holding caps. Indian pilot wallets are capped — ₹1,00,000 at most participating banks — so flight into central bank money is bounded by construction. Europe is fighting the same war at €500–3,000 (Section 6).
  3. Two-tier incentives. The intermediary bearing the KYC cost is the bank whose deposits the wallet drains. The concept note conceded the point with unusual candour: banks "would restrain themselves from distributing CBDCs if they find it as a threat to their bank deposits." Distribution design must overcome its own distributors.
💡
Pro Tip: The holding cap is the single number that tells you a jurisdiction's real position on the adoption-versus-banking trade-off. A cap of a few thousand units says: this is a payment instrument, not a store of value. No cap says: we are willing to compete with our own banks. Watch the number, ignore the press release.

4. Token, Privacy and the Law

Chapter 22 drew the token/account line; here is what it costs. A token-based retail CBDC is a bearer instrument — whoever holds the token owns it — and is therefore the cash-like design: no identity check at the moment of payment, offline-capable in principle, familiar. An account-based design ties every balance to a verified identity at every transaction: safer to police, useless for cash-like privacy. India chose token-based for retail (the concept note also considered account-based for wholesale).

But token does not mean anonymous, and here the design meets the law. The RBI's stated compromise is managed anonymity: anonymity for small value, traceability for high value — mirroring cash, where large transactions already demand PAN. Three frictions follow. First, the threshold has never been published; a policy choice about how much privacy citizens enjoy is being made administratively, outside any transparent process. Second, the enabling law is thin: the Finance Act 2022 amended the RBI Act — digital form within "banknote" (Section 2), the issue right (Section 22), a new Section 22A exempting digital notes from physical-only provisions, legal tender under Section 26 — but says nothing about privacy thresholds, intermediary liability, or redress. Third, even an "anonymous" token transaction runs on infrastructure that logs wallets, devices and timestamps — and the custodian model gives the token service provider visibility over every movement. Under Puttaswamy, informational privacy is a fundamental right; under the DPDP Act, someone in this chain is a data fiduciary; no CBDC-specific rule reconciles any of it. The design works. The legal frame around it is a pilot too.

5. Programmability and Offline

Programmability attaches a small rulebook to a token: who may receive it, which merchant categories may accept it, in which geography, until when. When the conditions are met, the merchant receives ordinary money and the rules fall away. The promise is leak-proof delivery: Odisha's Subhadra scheme paid 88,000 women in programmable e-rupee; PDS food-subsidy pilots in Gujarat, Puducherry and Chandigarh credited e-rupee redeemable only at fair-price shops and identified merchants; a tenant-farmer credit pilot restricted input loans to input stores. Against decades of subsidy leakage, this is a serious instrument.

The cost is definitional. Money that expires is not money — it fails the store-of-value function, and a currency whose units can die is a voucher system wearing a currency's clothes (Chapter 22's taxonomy is the reference). The RBI's implicit answer is a boundary: programmability attaches to purpose-bound issuances — welfare, credit-linked transfers — while the ordinary e-rupee in an ordinary wallet remains free. Whether that boundary holds as governments discover the convenience of conditional money is the live question. Offline capability is the quieter sibling: transactions without connectivity, cash-like resilience for the unconnected — bought at the price of double-spending risk, managed by technical checks and small value limits, with multiple solutions still under exploration in the Indian pilot. A retail sandbox (opened October 2025) now lets fintechs build against both features.

⚠
Warning: Do not let the welfare use case soften the principle. The tool that expires a food subsidy on the fifteenth is the tool that could expire a stipend, a refund, a payment to a dissident supplier. Programmability's defenders must argue the guardrails, not just the leaks it stops.
📋 Stable content — Reviewed: November 2026 · Next review: May 2027

6. What the Major Projects Chose

The same axes, five sovereign answers — as of late 2026. Re-verify the euro row at each review; the trilogue outcome will move it.

ProjectModelFormKey controlsStatus (late 2026)
e-rupee (India)Two-tier; 19 banks + non-banksToken (retail); no interestWallet cap ~₹1,00,000; managed anonymity; denominations like notesRetail + wholesale pilots since Nov–Dec 2022; ~1 crore users by Apr 2026
Digital euro (Eurozone)Two-tier via PSPsNot yet issued; design final pending lawNo interest; holding limit €500–3,000 under negotiation; waterfall fundingRegulation in trilogue; pilot H2 2027; potential first issuance 2029
e-CNY (China)Two-tier; six state banksToken; anonymity "controllable"Caps and quotas; expiry-capable features trialled for stimulusLargest pilot by user base; staggered city rollout continues
Sand Dollar (Bahamas)Two-tierToken; tiered wallets by KYCLower tiers capped, less identifiedLive since 2020 — first retail CBDC; modest uptake
eNaira (Nigeria)Two-tierTokenTiered wallets; zero-fee tier for the unbankedLive since 2021 — the cautionary tale of top-down adoption

The digital euro row deserves its paragraph, because it is the world's most public display of this chapter's trade-off. The Council fixed its mandate in December 2025 wanting the ECB to set the holding limit inside a legislated ceiling; the European Parliament's position (adopted July 2026) pulls calibration back toward the legislators. The ECB tested €500 to €3,000 and found even €3,000 would put only nine of 2,025 banks below full liquidity buffers under an extreme crisis scenario — so the number is not about solvency; it is about who owns the knob. Meanwhile the governing council will not decide issuance until the law passes, and its own calendar (pilot in the second half of 2027, first issuance during 2029, roughly €1.3 billion to build) hangs off the assumption that the regulation is adopted by end-2026. India built first and legislated thin; Europe legislates first and builds under the law. Two rational answers to the same design space.

7. Reading a Design Choice

Turn the chapter into a method. When you meet any CBDC proposal — a government press release, a central bank concept note, a startup's pitch — run four steps. One: place it on the five axes; nothing else makes sense until the position is fixed. Two: find the holding cap and the interest decision; they reveal the true adoption ambition better than any speech. Three: ask who bears the cost of the design — which bank's deposits, which citizen's privacy, which merchant's compliance. Four: name the failure mode: disintermediation if too attractive, disuse if too capped, surveillance if too traceable, exclusion if too offline-blind.

Chapter 24 turns the method on India's own numbers. The e-rupee's design choices are now made; the question becomes whether the evidence — users, circulation, transactions against UPI's continent-scale flows — shows a currency being adopted or a pilot being administered. The design chapter ends where the practice chapter begins.

🛠 Hands-On: Design Trade-off Matrix

You are the RBI's CBDC design desk. For each of three policy objectives, fill the matrix: retail/wholesale · direct/two-tier · token/account · programmable/free · offline/online-only · interest/no-interest · cap level (pick a number). Then answer the trap.

  1. Objective A — Rural inclusion: payments for districts with patchy connectivity, low smartphone penetration, and heavy cash habit.
  2. Objective B — Wholesale settlement: cutting counterparty risk in interbank government-securities trades.
  3. Objective C — Fertiliser subsidy: a subsidy that leaks heavily today — diverted to non-input purchases and ghost beneficiaries.

Steps

  1. Fill the seven-column matrix for each objective. No cell may be left blank; "either" is not an answer.
  2. For each cell, write one sentence naming what the choice trades away.
  3. The trap: for Objective C, the minister wants the subsidy to expire in 30 days. Using Chapter 22's functions of money, argue both for and against — and state your verdict.
  4. Extra credit: take the losing side of the digital-euro holding-limit fight (Council: ECB sets it; Parliament: legislators set it). Write the strongest three-sentence case for the side you disagree with.
View Solution / Walkthrough
ColumnA — Rural inclusionB — Wholesale settlementC — Fertiliser subsidy
Retail / wholesaleRetailWholesaleRetail (beneficiary wallets)
Direct / two-tierTwo-tier — banks and business correspondents must onboardTwo-tier — RBI issues, banks holdTwo-tier with Aadhaar/mobile onboarding, no bank account required
Token / accountToken — cash-like, no identity check per paymentAccount — institutions are known partiesToken with identity binding at issue (purpose-bound)
Programmable / freeFree — programmability would complicate first adoptionFree — settlement has no purpose conditionProgrammable — input stores only, validity window
OfflineYes, mandatory — with small value limits to bound double-spendingIrrelevant — institutions are always connectedYes, useful — input shops in rural blocks
InterestNone — cash parity, and no deposit drainPolicy-rate aligned if at all — wholesale is a settlement assetNone — it is a transfer, not savings
CapModerate (₹10,000–25,000) — enough for a week's cash needsUncapped for eligible FIs — settlement scaleIssue-sized — each credit is its own cap

3. The expiring-subsidy trap. For: an expiring credit is leak-proof — it cannot be hoarded, re-sold or diverted; unused value returns to the scheme; the purpose (seasonal inputs) is itself time-bound, so the expiry matches the need. Against: money that expires fails the store-of-value function by design — it is a voucher, not currency; a poor household that cannot reach an input shop in 30 days loses the benefit entirely; and normalising expiry in one scheme builds the machinery for expiry everywhere. Verdict: acceptable if the instrument is framed as a purpose-bound benefit credit — not as e-rupee — with a realistic validity window and automatic re-issue on lapse. The guardrail is the framing, not the technology.

4. The losing side. If you favour legislative caps (Parliament's instinct): the strongest case for Council/ECB calibration is speed and expertise — deposits move in a crisis in hours, not in legislative sessions; the ceiling can be written in law while the exact number stays with the institution that watches liquidity daily and is accountable to the Eurogroup. If you favour ECB calibration: the strongest case for Parliament is democratic ownership — the limit is a decision about how much of citizens' financial lives sits outside the banking system, and a five-year mandate is the wrong instrument for a knob that should answer to voters, not to the body whose own project it caps.

Key Takeaways

1

Five axes span CBDC design — retail/wholesale, direct/two-tier, token/account, programmable/free, offline/online — and no combination dominates.

2

Every project chose two-tier. The disintermediation trio — no interest, holding caps, bank distribution — is the price of a banking system that survives its own CBDC.

3

The holding cap is the honest number: it states a jurisdiction's real adoption ambition. India: ₹1 lakh. Europe: €500–3,000, still being fought over.

4

Managed anonymity is a policy choice made administratively — the threshold was never published, and the Finance Act 2022 is silent on privacy, liability and redress.

5

Programmability buys leak-proof welfare at the cost of money that can stop being money; the guardrail is the framing of purpose-bound credits, not the technology.

Test Your Understanding

Choose one best answer per question. The design matrix of the hands-on is fair data for any item. Aim for 24/30 before moving to Chapter 24.

1. The five axes of CBDC design are:

2. The central design problem of a retail CBDC, as this chapter frames it:

3. The RBI's own case for an e-rupee when UPI already works:

4. The success metric that separates retail from wholesale CBDC:

5. India's wholesale e-rupee pilot, from November 2022, settled:

6. The RBI's stated reason for rejecting a direct model:

7. The standard disintermediation controls on a retail CBDC:

8. Why does no major CBDC pay interest on holdings?

9. The concept note's candid admission about distribution:

10. Watch the number, ignore the press release — the number is:

11. What India's token choice for retail CBDC means at the counter:

12. Managed anonymity, defined — and its transparency gap:

13. The Finance Act 2022 did all of the following EXCEPT:

14. Even an anonymous-designated token transaction still generates:

15. The constitutional anchor for privacy in the CBDC debate:

16. What happens to a programmable e-rupee's rules when the eligible merchant is paid?

17. Match the programmable pilots: Subhadra and the PDS food pilots belong to:

18. The definitional cost of expiring subsidy money:

19. Offline CBDC payments trade what for what?

20. What the retail sandbox opened in October 2025 allows:

21. As of late 2026, e-rupee wallets are offered by:

22. When an e-rupee app scans a UPI QR code:

23. The digital-euro holding-limit fight, in one line:

24. The ECB's technical finding on a €3,000 holding limit:

25. The digital euro's governing sequence:

26. The India-Europe contrast, compressed:

27. The first-mover pair, correctly matched:

28. In the four-step method for reading a CBDC proposal, the honest signals are found at step:

29. Which is NOT one of the chapter's four CBDC failure modes?

30. Chapter 24 asks, of the e-rupee's evidence: