Where you are. You have the two lower tiers in token form: a non-bank issuer’s claim backed by reserve assets, and a commercial bank’s deposit. Module 1’s map has a tier above both, and it is the obvious place to ask the question next.
Central bank money is the settlement asset. Module 1 established why: it is the only claim that is not somebody’s credit risk, which is what makes it final. Module 2 built the system that moves it between banks and spent a whole lesson on how much liquidity that system needs.
So: what if central bank money were a token on a shared ledger?
The idea in one paragraph
A central bank digital currency is a direct claim on the central bank in token form, and the question splits immediately into two projects with almost nothing in common. The wholesale version gives banks tokenised reserves for settling with each other, which is a change of plumbing in a system whose participants already hold central bank accounts, and it slots in beside the machinery of module 2 without disturbing anybody else. The retail version gives the public a direct claim on the central bank, which is a change to the structure of banking itself: if anybody can hold the safest asset directly, the reason to keep money at a commercial bank weakens, and the deposits that fund lending are the thing at risk. The first is an engineering project. The second is a constitutional one, and that is why the two have moved at completely different speeds.
Wider than the screen; scroll it sideways.
Where they have actually got to
Three launches against 146 explorations is the number to sit with. This is not a technology that failed to work; the launched ones work. It is a technology whose retail case nobody has been able to make convincingly enough to carry the political cost.
What the exercise shows
The exercise adds a central bank to the shared ledger, settles an interbank transfer in wholesale central bank money, and compares the account chain with module 2’s real-time gross settlement path.
The result is deflationary in a useful way: the chain is nearly the same. Both settle in central bank money, both are final on settlement, both require the paying bank to have the balance. What changes is the mechanics of the record rather than the economics of the claim.
That is the honest case for wholesale CBDC, and it is a real one. It is not that settlement becomes possible; it already was. It is that settlement in central bank money becomes available inside the same transaction as a securities leg, which is what makes atomic delivery against payment possible in central bank money rather than in a commercial bank’s liability. Module 5’s clearing house would like that very much.
Why the retail version stalls
Three objections, and none of them is technical.
The first is disintermediation: if the public can hold central bank money directly, deposits leave commercial banks, and deposits are what fund lending. Every serious retail design carries holding limits or tiered remuneration to blunt this, which is an admission of the problem rather than a solution to it.
The second is privacy. A central bank account for everybody is a record of everybody’s payments held by the state, and the designs that avoid this add intermediaries back until the result looks a good deal like the banking system it was going to replace.
The third is demand. In a country with working instant payments, it is genuinely hard to say what a member of the public gets from a retail CBDC that they do not already have. Module 2 built those rails, and they work.
Check yourself
1. Why do wholesale and retail CBDC belong in different conversations?
Because wholesale changes the plumbing among participants who already hold central bank accounts, while retail changes who may hold the safest asset at all. The first is an engineering change; the second alters the funding structure of commercial banking. They share a name and very little else.
2. Three countries have launched a retail CBDC out of 146 exploring one. What does that ratio suggest?
Not that the technology failed, since the launched ones work. It suggests the retail case has been hard to make convincingly enough to carry the political and structural cost, which is consistent with the three objections: disintermediation, privacy, and unclear demand where instant payments already exist.
3. The exercise finds the wholesale chain is nearly the same as the RTGS path. So what is the point?
That settlement in central bank money becomes available inside the same transaction as another leg, which makes atomic delivery against payment possible in the top tier rather than in a commercial bank’s liability. The economics of the claim do not change; what changes is what it can be combined with.
4. Why does this course date the Nigerian adoption figure rather than simply quoting it?
Because it comes from a 2024 report reflecting roughly 2023 data and is widely repeated as though current. It is a real finding about an early period, and presenting a several-year-old measurement as the present state of a fast-moving field is the exact failure the course’s dating discipline exists to prevent.
Do this
Settle in central bank money on the shared ledger.
python3 code/cbdc.pyThe starter adds a central bank to the shared ledger and leaves you the wholesale settlement: move the balance between two banks’ accounts at the central bank, refusing when the payer is short. The assertions check that the transfer settles finally, that an unfunded one is refused, and that the resulting account chain matches the RTGS path in length.
The completed version is in solutions/cbdc.py.
What you can now do. You can place all four kinds of digital money on the two-tier map, settle an interbank transfer in tokenised central bank money, and explain why the wholesale version moves while the retail version argues. The next lesson leaves money behind and asks what happens when the thing in the token is not money at all.