Whose Frontier? What the BIS Left Out of Its Own Comparison

Whose Frontier? What the BIS Left Out of Its Own Comparison

At Jackson Hole this year, the BIS’s General Manager compared two ways of putting money on a blockchain and found the bank-issued version safer, sounder, and more deserving of the future. The comparison would be more convincing if it had included the third option sitting in plain view.

Digital Finance frontier

The setup

On 28 August, Pablo Hernández de Cos, General Manager of the Bank for International Settlements (BIS), used his Jackson Hole panel to lay out a choice for the “next frontier” of money: stablecoins on one side, tokenised bank deposits on the other. Judged against three criteria — singleness, interoperability, and financial integrity — tokenised deposits win comfortably in his account, because they “operate within the two-tier system,” preserving “the tight link between deposit-taking and credit provision” and thereby “reducing the risk of disintermediation and supporting resilience.” Stablecoins, by contrast, are found wanting on nearly every count: fragmented across chains, exposed to run risk, vulnerable to illicit use.

Nothing in that assessment is obviously false; stablecoins do have real interoperability and integrity problems. The trouble is what the comparison leaves out. A speech about “the monetary frontier” that weighs two forms of private money against each other, and never seriously puts genuinely public digital money — a retail central bank digital currency — into the same frame, has already decided the terms of the debate before making the case. Tokenised deposits don’t win because they’re the best available design. They win because they’re the only option on the table that keeps the existing two-tier, bank-intermediated system fully intact.

That’s worth taking seriously rather than dismissing as an oversight, because the same institutional move — treating the preservation of the current bank-based system as a self-evidently correct starting point, rather than one of the things actually being tested is one already documented in the digital euro’s own case (Helmich, SSRN & 2026b).

The missing third option

The absence isn’t for lack of institutional familiarity: the BIS has published extensively on retail CBDC, and this very speech leans on its own 2026 Annual Economic Report chapter on “anchoring trust in money” as its intellectual foundation. The tools to include a genuine public-money option in the comparison were sitting on the shelf. They weren’t used.

The timing sharpens the point: De Cos gave this speech in the middle of the EU’s active trilogue negotiations over the digital euro, arguably the most consequential live retail CBDC debate among advanced economies right now. A speech framed explicitly around “the monetary frontier” for “advanced economies” does not mention the digital euro once, nor any other retail CBDC, by name.

Capacity and price, treated as one thing

The clearest example sits in the speech’s own account of what happens to bank funding under stablecoin competition: “If stablecoin reserves were predominantly held as wholesale bank deposits, retail funding would give way to concentrated, more rate-sensitive wholesale liabilities. This would raise banks’ marginal funding costs and thereby tighten lending conditions.”

Read that sentence twice. It moves from “funding costs rise” to “lending conditions tighten” as though the second follows automatically from the first, and treats stablecoin reserve composition as one undifferentiated thing. It isn’t. Niepelt, Rey, Cecchetti and Vives’s 2026 CEPR report on digital money formalises the underlying distinction as a neutrality result: when a central bank recycles displaced deposit funding back to banks on terms equivalent to what those deposits previously offered, banks’ capacity to lend is largely unaffected, only the price of that funding changes. For a central bank digital currency, that recycling is structural: the transaction runs directly through the central bank’s own balance sheet, leaving it as the direct counterparty with unilateral power to lend the funds straight back. Applied to the digital euro, this distinction shows the ECB’s own published simulations are considerably more measured than the disintermediation narrative built on top of them (Helmich, SSRN & 2026b).

Whether the same recycling holds for stablecoins depends on a specific, discretionary design choice. A stablecoin purchase is an ordinary interbank payment; the central bank facilitates settlement but does not automatically retain the funds, so whether a funding shock gets offset depends on how much of an issuer’s reserves sit directly at the central bank versus in government bonds, where the effect runs through diffuse market forces instead. The Bank of England’s own systemic-stablecoin regime makes this distinction concrete: a mandatory 30% minimum held at the central bank, and, more tellingly, a purpose-built liquidity facility letting the Bank lend against issuers’ gilt holdings, a deliberate bridge for exactly the portion of reserves that wouldn’t otherwise recycle back to banks at all (Helmich, 2026a). That facility is not a neutral technical detail. It is evidence that the “tightened lending conditions” outcome the BIS speech treats as a fact about stablecoins is really a fact about policy choices, present in some jurisdictions and not in others, that the speech never once discusses.

The speech’s own internal logic makes a related double standard visible without needing an outside source to prove it. A few paragraphs after warning that stablecoin-driven deposit outflows would “tighten lending conditions,” the same speech acknowledges that tokenised deposits carry a parallel risk: “round-the-clock operability could quicken deposit outflows and might require additional backstops to safeguard financial stability.” The identical risk counts as a strike against stablecoins’ fitness to be money, and as a mere “challenge” to be managed for tokenised deposits. The verdict depends on which technology is carrying it.

Even the speech’s own cited research cuts against its prose. Hofmann, Kaldorf and Rottner’s 2026 BIS working paper finds only a “modest net output effect” from stablecoin adoption overall, a considerably smaller claim than “tighten lending conditions” implies.

“Tight link” as a privilege, not just protection

The corrected picture from the previous section sharpens a second point the speech never confronts: what, precisely, is being protected when a “tight link between deposit-taking and credit provision” is preserved.

Niepelt, Rey, Cecchetti and Vives’s 2026 CEPR report on digital money puts a number on it. Using US data, they estimate the portion of bank net interest margin exposed to this substitution risk at roughly 0.2% of GDP over the past fifty years. Niepelt’s companion working paper traces the figure to a gross deposit-taking profit of roughly 0.67% of GDP, of which about a third, the portion attributable to market power rather than competitive pricing, is what’s actually at risk. The same report is candid about what this implies politically: it describes the resulting dynamic as “a contest over institutional power,” banks and their allies defending a rent, not a stability buffer.

Read against that literature, “the tight link between deposit-taking and credit provision” is not a description of a stability mechanism. It is a description of the mechanism by which banks fund themselves more cheaply than ordinary market competition would allow, because they are the ones creating the deposits they then hold as their own funding. A tokenised-deposit-based monetary architecture cannot disrupt this arrangement at all: by construction, the money never leaves the banking system, so there is no reserve-composition question and nothing for a backstop to bridge. Where the UK’s stablecoin regime at least makes the trade-off visible, in the specific choice of how much of an issuer’s reserves must sit at the central bank versus in gilts, and the explicit liquidity facility built to cover the gap, a tokenised-deposit architecture removes the trade-off from view entirely by keeping every deposit inside the banking system by design. That is not a more resilient answer to the same question. It is a structure that guarantees the question never has to be asked.

This is worth stating plainly: the speech’s own language invites the opposite reading. “Resilience” implies a system-wide good. What the evidence actually says is narrower: banks retain a funding advantage worth a calculable share of GDP, defended in the language of stability rather than institutional interest.

An institutional interest, not a personal one

The case made so far rests on the substance of the speech alone: what it compares, what it leaves out, where its logic slips between price and capacity. But it is worth briefly asking who is making these comparisons, and from where.

The BIS describes its own mission as supporting “central banks’ pursuit of monetary and financial stability through international cooperation,” and it hosts the Basel Committee on Banking Supervision, the world’s standard-setter for bank capital and liquidity regulation, as one of its permanent committees. De Cos himself chaired that committee from 2019 to 2024, before becoming BIS General Manager, a career spent inside the institutions that strengthen and supervise commercial banks; one data point among many, not a special case. An institution whose core work is strengthening the two-tier banking system is not a neutral referee when the question on the table is whether the two-tier banking system should remain the default architecture for the next generation of money. That does not make its technical analysis wrong. It does mean the analysis should be read the way any interested party’s analysis should be read, with the interest disclosed rather than assumed away. A speech from its General Manager that finds the bank-preserving option safer is not surprising; the bias is predictable, given whose interests the BIS is built to serve.

What a genuinely neutral comparison would look like

None of this means the BIS’s three criteria (singleness, interoperability, integrity) are the wrong ones to judge a monetary instrument by. The problem is not the test; it is that the test was applied to only two of the three live options, and a second question, whose funding and whose privilege each design protects, was folded silently into the first without ever being named as its own axis.

A fairer comparison would keep those two questions apart. On the technical axis, CBDC, tokenised deposits and stablecoins can be judged evenhandedly against the same three criteria the speech already uses. A retail CBDC, settled directly on the central bank’s own ledger, arguably outperforms both alternatives on singleness by construction: redemption at par is the starting condition, not something engineered or backstopped. That is the strongest technical case for the option the speech does not examine.

On the second axis, the right question is not whether a design protects some existing interest, every design does, but whether that protection is declared openly or dressed up as stability. Tokenised deposits protect bank funding by construction, since the money in question never leaves the system. Stablecoins protect it only to the degree a jurisdiction chooses to build a recycling mechanism, as the UK’s liquidity facility and MiCAR’s absence of one already show doing differently. A central bank digital currency protects it, or fails to, entirely at the discretion of whoever controls the recycling terms, a live question in the EU’s own digital euro negotiations right now (Helmich, SSRN & 2026b).

That decision is the same each time, whichever instrument it attaches to: who sets the criteria by which a central bank’s balance sheet allocates credit back into the economy. There is precedent for making that decision openly. Postwar France, and similar arrangements in Germany, Italy and Belgium, ran credit-guidance frameworks in which elected authorities, not central banks, set the priorities for favourable financing terms, leaving day-to-day lending decisions to banks (Monnet, 2018). The lesson from that history is not that central banks should direct credit themselves. It is that when a central bank’s design choices end up allocating credit either way, as tokenised deposits, stablecoins and CBDC all eventually do once reserves or backing assets are recycled at scale, those criteria are a matter for democratic institutions, not a technical detail for the BIS or any central bank to settle alone.

Conclusion

None of this argues that stablecoins are secretly the safer choice, or that the BIS has got the technical merits backwards. On singleness, interoperability and integrity, tokenised deposits may well be the better instrument, at least until public blockchains solve their own integrity problems. The case made here is narrower: a comparison that leaves out the option most likely to unsettle its own conclusion is not neutral, whatever technical language it is dressed in, and one that treats bank funding costs and lending capacity as the same thing has smuggled a policy preference in as a technical finding.

That preference has a name, and it runs through the digital euro debate, the UK’s stablecoin regime, and now the BIS’s own framing of the monetary frontier alike: preserve the two-tier banking system’s funding advantage first, and treat the alternative as a technical problem to solve rather than a political choice to make openly. It is a coherent preference, held by capable people, for defensible reasons. It is not, however, the neutral starting point it is so often presented as.

Jackson Hole is, appropriately enough, a good place to notice this. Two centuries ago the valley’s money was decided by whoever showed up with something worth trading. The next generation of money will be decided by whoever gets to set the terms of comparison before the argument even starts. That is worth naming as the actual frontier, whatever technology ends up crossing it.

References

BIS (2026), Anchoring Trust in Money: Innovation Beyond Stablecoins, Annual Economic Report 2026, Chapter III, Bank for International Settlements, June.

De Cos, P H (2026), “Pushing the Monetary Frontier: Stablecoins and Tokenised Deposits,” speech at the Jackson Hole Economic Symposium, Bank for International Settlements, 28 August.

Helmich, P (2026a), “Britain’s Synthetic CBDC,” OMFIF, August.

Helmich, P (2026b), “Whose Interest Does the Digital Euro’s Holding Limits Serve?”, SSRN Working Paper, forthcoming Rosa & Roubini.

Hofmann, B, M Kaldorf and M Rottner (2026), “The Macroeconomics of Stablecoins,” BIS Working Papers, no 1363.

Monnet, E (2018), Controlling Credit: Central Banking and the Planned Economy in Postwar France, 1948–1973, Cambridge University Press.

Niepelt, D (2026), “Central Bank Digital Currency and Monetary Architecture,” CEPR Discussion Paper DP21141, University of Bern and CEPR, 8 February.

Niepelt, D, H Rey, S Cecchetti and X Vives (2026), Digital Money, Barcelona Report 8, CEPR.