Are CBDCs a good idea?

Start with a fact most people miss: there are already two kinds of money in your life, and only one of them is public. The cash in your pocket is central-bank money — a direct claim on the state, with no credit risk; if the issuer is the central bank, it cannot bounce. The number in your banking app is commercial-bank money — a claim on a private bank, backed by that bank’s solvency and a government deposit guarantee. Almost all the money you actually use is already digital. It is just private digital money.

A retail CBDC is the missing third box: central-bank money in digital form — the digital analogue of cash, a direct claim on the state that ordinary people can hold and spend. That is the whole concept. And it is worth being precise about what it is not. It is not crypto: it is centralized and state-issued, the opposite of a permissionless blockchain. It is not a stablecoin: that is a private token pegged to a currency, issued by a company. And it is not inherently a surveillance tool, nor inherently private — the privacy of the ledger and the programmability of the money are design choices, not properties that come baked into the idea. This is the move that resolves most of the shouting: the two camps are arguing about two different CBDCs, and the technology can be built as either one.

For the deeper question of what money even is — commodity, fiat, credit, public versus private — the dedicated home is Is money real, or just a shared illusion? And the institution doing the issuing is the subject of What do central banks actually do? A CBDC sits exactly at the seam between those two: it is a new central-bank liability aimed at the public. The long arc that got us here — coins, gold, fiat, the disappearance of cash — is a story in its own right; this walkthrough takes that arc as given and asks the narrower, live question: is building this particular box a good idea?

To see why a state might build its own money, start with why payments concentrate. A payment network has powerful network effects: you want to be where everyone else already is, so payments tend to pile up onto a few rails — a couple of card networks, a couple of Big-Tech wallets. Left alone, the market for payments tilts toward natural monopoly. That is exactly the condition under which economists treat a service as a candidate for public provision — the way roads or the postal system are: a universal, neutral option that doesn’t extract rents and doesn’t gate access on owning the right account. A CBDC is, on this reading, public infrastructure for the digital-payment age.

So what could a CBDC actually do? Three things, in ascending order of how strong the argument is.

1. Financial inclusion. A public account with no minimum balance, no bank required, reachable from a basic phone — for the roughly 1.4 billion adults worldwide without a bank account, that is a genuine welfare gain. This is the case the marketing leads with, and it is real.

2. A public money option in a cashless world. As physical cash — the one form of public money the public can hold — disappears, a CBDC keeps a public alternative to private bank money and Big-Tech wallets alive. Less a new capability than a refusal to let the public’s only sovereign money quietly vanish.

3. Monetary-sovereignty defense — the strongest professional case. This is the one the Bank for International Settlements and most central banks actually lead with, and it gets the most weight. The domestic payment and unit-of-account layer is starting to be colonized from two directions: dollar-pegged private stablecoins, and the prospect of a foreign CBDC — an internationalized digital yuan — running inside other countries’ economies. If a small country’s citizens drift onto a private dollar token or a foreign state’s money for everyday payments, that country has quietly ceded control of its own monetary system. A public CBDC is how the state keeps a seat at its own table. The case is defensive, not transformational — and that distinction is the whole game.

But notice what’s missing. The first two wins — inclusion and a cashless-age public option — are largely achievable without a CBDC. M-Pesa did inclusion with mobile-network money; India’s UPI and Brazil’s Pix did instant, near-free, universal payments with fast-payment systems sitting on top of ordinary bank accounts. No new form of central-bank money was required. So the payment-efficiency case is real but mostly solvable by cheaper means. The one argument that genuinely needs a public instrument is sovereignty — and that is exactly why the verdict later turns on which kind of country is asking. The financialization and electronic-payment build-out that set this stage is the backdrop of Economic History Ch.18 §18.3 (Financialization).

The sovereignty argument sits next to a larger question this walkthrough doesn’t reopen: whether the state should be the one creating money in the first place — the territory of Is Modern Monetary Theory crank or serious? The cross-border contest — whether the dollar’s dominance, a foreign CBDC, or a private stablecoin ends up running on your rails — is a sibling question; here it matters only as the threat the sovereignty case is defending against. For the deeper what-is-money grounding, see Is money real, or just a shared illusion?

The surveillance hazard — taken seriously, then bounded. A CBDC ledger can be state-readable and programmable, and that is a genuine civil-liberties hazard, not a fringe worry. Financial privacy is a real freedom: what you buy reveals who you are, and a money the state can read and switch off is a coercion tool. The e-CNY proves the capability ships, and “trust us” is not a privacy guarantee. The honest hinge is that this is a design choice, not a property of the concept. Privacy-by-design is technically possible — tiered anonymity that makes small payments cash-like below a threshold, data minimization, no direct central-bank access to identities, and hard legal limits on programmability — and the European Central Bank’s digital-euro design explicitly targets cash-like privacy for low-value payments. So both halves are true at once: the threat frame is right that the capability is real and a hostile state would use it; it is wrong that the outcome is inevitable, because a liberal democracy can build the privacy in. Whether it will is a political choice, not a technical certainty — which is exactly why the concern stays serious even after the design answer.

The bank-run hazard — the mechanism the central bankers fear most. Today the banking system is two-tier: the public holds commercial-bank deposits, and banks hold reserves at the central bank. A retail CBDC inserts a new option — a direct, default-free claim on the central bank that the public can hold. In normal times that competes with bank deposits: if a perfectly safe public account is one tap away, banks have to pay more to keep your money, cheap deposit funding shrinks, and credit gets pricier. In a crisis it is far worse. The instant a bank looks shaky, depositors can flee to the perfectly safe CBDC instantly and at scale — no queue outside the branch, just a phone. Every wobble can become a system-wide run faster than any in history. This is the digital amplification of the 2008 wholesale-funding runs, where money fled a stumbling bank in days; a CBDC could compress that into minutes.

The canonical modern demonstration of how fast funding flees a wobbling bank is 2008 — the spine of Economic History Ch.19 §19.1 (September 2008), and the run-acceleration logic the disintermediation worry borrows is the same one explored in Did economists cause the 2008 crisis? A CBDC doesn’t invent the bank-run risk; it removes the friction that used to slow it down.

The design space that bounds the run risk. The central-bank research literature converges on three mitigations, and they are real:

• Holding limits — a cap per person, set high enough for everyday payments but far too low to drain the banking system in a panic.
• Tiered or zero remuneration — no interest above a threshold, so the CBDC is a payment instrument, not a savings vehicle that out-competes bank deposits in calm times.
• Intermediated, two-tier architecture — banks and payment providers distribute the CBDC and run the customer accounts; the central bank doesn’t hold retail accounts directly, preserving the banking system’s role.

The catch, and it is load-bearing: a CBDC capped, un-remunerated, and intermediated enough to be safe for banks is also capped enough to be a modest payment instrument, not a revolution. The mitigation and the use-case pull against each other. You can have a CBDC that protects the banking system or one that displaces it, but the safe version is the small version — which already tells you the enthusiasts’ “payment revolution” and the realistic, well-designed product are not the same thing. The institutional-design lens for trading off these constraints is Economics Ch.18 §18.5 (Property Rights and design). And the idea of a public money so safe it strips the banks of deposits has an ancestor: the “narrow banking” / Chicago-Plan tradition — Friedman’s interest in 100%-reserve money — whose lineage sits in History of Economic Thought Ch.10 §10.1 (Friedman’s monetarism), where the libertarian, anti-state-money strand the surveillance frame also draws on is situated (not endorsed).

The CBDC sits at the recent end of a long intellectual argument over who should control money; the era-cluster view of that fight is the counter-revolution cluster of the money lineage. And because a CBDC reshapes the central bank’s direct relationship to the public, it bears on what central banks are for at all — What do central banks actually do?

When a public payment rail earns its weight. A CBDC is worth building where two conditions line up. Either the existing payment system is weak or exclusionary — a large unbanked population, no good fast-payment system — so the inclusion and efficiency gain is large; or there is a real monetary-sovereignty threat, with private stablecoins or a foreign CBDC poised to colonize the domestic payment layer, so the defensive public-option gain is real. And in either case the design has to protect privacy and bound the run risk. Where those hold, the modest, well-designed CBDC clears the bar.

When it’s a solution in search of a problem. Where private rails already work well — cards, FedNow, near-universal bank access, the United States and much of Western Europe — the marginal payment gain is small, the surveillance and run risks are non-trivial, and the only live motive is the sovereignty defense, which reasonable people weigh differently. The skeptic camp, and the US political verdict, says the sovereignty motive alone doesn’t clear the bar in such a country; the defensive camp, and the European Central Bank, says build the option before stablecoins entrench, because building after the network tips is far harder. This is the live, genuinely contested edge — and the parameter that resolves it is how fast private stablecoins and foreign CBDCs are judged to threaten the domestic payment layer.

Both extremes oversell — each in its own specific way. The enthusiast’s “payment revolution / programmable-money future” oversells an upside that the safety-vs-usefulness design constraints keep modest: the version that protects the banks is the small version. The alarmist’s “surveillance tyranny / end of cash and freedom” oversells an inevitability that good design and political will can prevent: the capability is real, the dystopia is a choice. The mainstream sits between them — not out of lazy centrism, but because each extreme is wrong about a different, nameable thing. The post-2008 challenger debate this lands in — digital money, stablecoins, state-versus-private money — is mapped in History of Economic Thought Ch.17 §17.5 (the expanding frontier).

The whole “solution in search of a problem” assessment depends on what payment infrastructure a country already has — the electronic-payment build-out narrated in Economic History Ch.18 §18.3 and the post-crisis monetary regime in Ch.19 §19.3. A country with weak rails and a country with FedNow are not asking the same question, even when they use the same three letters.