From Fiat-ed Coins to Decentralised Central Banking
Deposit tokens don't work and Stable Coin don't serve any purpose in Decentralised Finance (DeFi)
This article hopefully is what your question was about with respect to a comment I made on social media, in my typically restrained tone of course:
“Deposit tokens do not exist and cannot exist, because a “deposit token” is a contradiction in terms. This should be obvious if one simply considers what deposits are—and what tokens are. You can use blockchain to record deposit transactions, but you cannot tokenize a deposit. If you can, then pigs can also fly.”
Ever since JP Morgan started this idea, I always saw red when that term popped up. A deposit is a different currency from a banknote or funds held on a central bank’s RTGS system. We say "deposit" like calling somebody Joe — but Joe has a full name, and so does a deposit: it’s called on-demand deposit or sight deposit. And it has an ugly twin called a time deposit (in TradFi).
Legally, you have a bank account into which deposits can be made or from which withdrawals can be taken. The account is governed by a bilateral legal arrangement between the account holder and the bank. If there is more than one account holder, they must hold it jointly — i.e., each individual owns everything jointly.
If money is transferred, the right to request payment on demand (i.e. now, pronto) from the bank is eliminated. What gets transferred is funds, not "my deposit" — because that deposit no longer exists.
What changes hands between banks are my bank’s central bank reserves. The legal claim (the deposit) is extinguished by my bank, and the receiving bank creates a new deposit — unless it finds a reason not to. But there is no direct 1:1 relationship in the actual funds each bank manages.
It is therefore legally not possible for the bank or the account holder to create a token that would enable an ownership transfer of my claim against the bank. This is not property but a right and that right only exist between then bank and me. My bank only recognises this claim when I make this claim. Any document I create that says I sell you my claim so take my document and it will be fine you can withdraw my money will be rejected by the bank (hopefully). We can create tokens all day long but they cannot be turned into anything that changes what I just said. I may go to prison for fraud selling my deposits this way which presumably creates unnecessary risk for my bank. Fun idea though.
But let’s say law is irrelevant (and regulation and stuff)— we can do some financial engineering despite these old-fashioned laws. If we want to take the token and move it freely outside JP Morgan, then the bank and I, logically speaking, would have to agree. This is similar to selling a loan, which sometimes requires the borrower’s consent.
So let’s say the SWJP Coin is a super-stable token. Each token represents 1 unit of the SW–JP bilateral arrangement that entitles SW to boss JP around, and both SW and JP agree that anyone with this token can boss JP around as if they were me but strictly limited to the equivalent number of tokens they present. Here we already encounter a problem. You can’t prove you hold the tokens without handing them over which means you no longer possess what is required to make the claim. In the absence of applicable legal frameworks, this is inherently risky. If JP Morgan were to go bankrupt at that very moment, you’d have no way to verify your entitlement and likely lose everything. They are a very fine bank so it’s a very small risk of course.
Now let’s say you bank with Citi and I send you this token as payment. You’re an account holder with Citi, but that account can’t hold my deposit. But let’s ignore that too and pretend we find a way: you now have deposits with Citi and you have something that is not a Citi deposit — even though it’s denominated in the same currency.
The mere fact that your record distinguishes two types of dollars makes it impossible to assume they have the same value — unless the law gives you a good reason. But the law says: “Ah, that was your bloody idea, not mine.”
We do, in fact, have such an instrument that was created: bonds. But that only works because the law says we must treat them as property, which can then be transferred. However, the law does not say we can enforce that for deposits against the world.
At that moment, by accepting this payment, you’ve agreed to put yourself in a very vulnerable position in any market crisis — because whoever gets to a troubled bank first gets their money out without a loss. And that additional layer, even with tech, will put you one second behind me — and one second is all I need to make sure you bear all the losses in a crisis but not me.
Hence, deposit tokens not only privatize money and thereby destroy what they claim to represent — they also expose the holder of such a token to structural fragility.
That kills the whole idea. And we don’t need to POC our way around it.
But let’s have a look what it looks like when JP Morgan’s blockchain people take an axe to English and reasoning talking about this topic
“A foundation for stable digital money”
Kinexys by J.P. Morgan has launched a USD J.P. Morgan Deposit Token (JPMD) proof-of-concept (PoC) on public blockchain, as an alternative to stablecoins for native cash settlement and payments use cases for J.P. Morgan institutional clients. [..]
A Deposit Token is an electronic payment instrument issued by a bank, representing funds that have been deposited by a customer. It can be deployed on public and private blockchain networks and is intended to be transferable among the issuing bank’s direct customers as well as between their eligible customers.”
A payment instrument?
FCA, please do us the favour:
“payment instrument
(1) (in BCOBS) any personalised device or personalised set of procedures agreed between the banking customer and the firm used by the banking customer to initiate an instruction or request by the banking customer to the firm to make a payment. [..]
So this means if I have £100 in my account and give a payment order that the money on the account is not a payment instrument.
Instead:
The £100 is a claim (debt) reflected as a deposit liability by the bank.
The payment instruction I issue is a tool to direct the bank to discharge part of its debt to me by transferring value to someone else.
The payment instrument is therefore the form and method by which that direction is expressed and acted on.
Rhetorical question: Can JP Deposit Token be an electronic payment instrument issued by JPM?
I don’t think that’s possible. I mean in a way that the law or the banking sector refers to it. A payment must involve bank acceptance. Otherwise, a token transfer cannot be a valid payment. If JPM doesn’t control validation on Base, they cannot treat a blockchain movement as a payment instruction that discharges a deposit liability. That undermines the core legal mechanics of a deposit.
Neither is the token a payment instrument nor can it be treated as one and it’s not a token because the token cannot be used to transfer value. The clients have no control over the tokens.
The JPMorgan Deposit Token (JPMD) cannot be a payment instrument because deposits are not legally transferable, and payment instruments must operate over transferable rights or instructions.
The JPMD can at most be a representation of a personal deposit claim. But since such claims are not transferable, and payment instruments require a mechanism of discharging transferable obligations, the JPMD fails to qualify — both legally and functionally.
The language used by JPMorgan obscures this fundamental incompatibility. It cannot ever become a payment instrument unless law is changed or the entire framework of deposit accounting is reconstructed to support transferable liabilities — which would require a major shift in banking law.
Imagine I walk into a JPMorgan branch and ask:
“Excuse me, which account here has the highest balance?”
The branch manager replies:
“Who are you? We can’t disclose that. Is the account in question yours?”
I say:
“No, of course not. Mine is the next one after that. Here are some tokens.”
Now, I want to redeem funds — not because the deposit was made in my name, but because someone else tokenized their account, used the token to pay a bill, and I ended up with it. How much have you got for me?
JPMorgan now has to decide whether I’m entitled to receive the funds.
Let’s set aside the KYC and AML nightmares for a moment.
The point is this: tokenization exposes JPMorgan to new risks they are not compensated for. The token behaves like a bearer claim, but JPMorgan still holds the underlying liability.
The tokenization creates a risk that the person causing the risk doesn’t bear but JP Morgan does and thereby all other clients. That risk doesn’t vanish. It’s absorbed by the institution and ultimately priced into the experience of every other customer. That destroys this idea again economically.
In economic terms, the model is inefficient. In legal terms, it’s incoherent and impossible. And in practical terms, it’s unfair — because it allows one party to extract rents from the system while bypassing the constraints everyone else must follow.
Suppose you receive a letter saying:
“Dear Borrower,
We’ve sold your loan to a third party. Please send your future payments to an account in the Bahamas. Do it quickly!”
Your first reaction is: This could be fraud. Even if it's not, you have no clear legal basis to trust the claim. The safer option is to keep paying your original bank — because you have a contractual relationship with them, and you expect them to handle any assignment internally. That way, if the sale was real, the bank can forward the proceeds. If it wasn’t, you’re protected.
For that same reason, tokenizing deposits cannot work.
But that’s not even the biggest problem. That was only the Base problem.
“JPMD – and Deposit Tokens in general – match the novel and distinctive properties of stablecoins, most notably the ability to conduct peer-to-peer transactions with programmability. As such, Deposit Tokens and stablecoins can be applied to similar use cases. However, Deposit Tokens differ from stablecoins in key areas including, interest payouts, and deposit treatment.”
That implies two incompatible properties:
Peer-to-peer suggests bearer logic: whoever holds the token can spend or transfer it.
Interest-bearing requires registered ownership: the issuer must know exactly who owns it to credit interest and perform regulatory duties (e.g., tax reporting, anti-money laundering, etc.).
A token cannot be both:
Freely transferable without identity binding (bearer), and
Legally eligible to receive interest (registered)
because:
Bearer assets can't reliably support interest payout, taxation, or legal claim enforcement.
Registered assets can't support trustless peer-to-peer movement without constant identity tracking.
Peer-to-peer would require that every transaction includes live KYC and identity reassignment. Since this limits the potential counterparts the claim that this thing competes with Stablecoin is not only contradictory, it’s structurally misleading.
In traditional commercial banking, interest rates are private. So are all institutional clients receiving the same interest and if not, then interest compensation could not take place via this.
Paying differentiated interest on a fungible token is structurally incompatible. You can’t tokenize deposits and still do bespoke yield management without:
Making the token non-transferable,
Killing programmability,
Or faking uniformity while settling interest separately off-chain.
Interest and token fungibility are in conflict. The balance is held at an entity — let’s say the New York branch versus Hong Kong. Not every client is onboarded with every entity. But the model creates a mismatch between the bank’s internal records and its balance sheet versus the token position if you allow a payment from a Hong Kong client to a New York client.
At what point did the asset leave one entity and arrive at the other? And how do you reconcile if any token transfer takes place after business hours — for example, in Hong Kong? If the closing balance from the prior day and the opening balance today are out of sync, we have a reconciliation break that causes a mismatch in the bank’s ledger.
If a blockchain token can move without the bank’s approval, visibility, or timing, then the bank loses control over its books and records — which would not be permitted. There are further breaks in the nostro accounting. Hong Kong has the client funds, but it also has a deposit relationship with the US entity. Both the client funds and the nostro position should now be reduced — but without a client instruction or an implied instruction. These two processes cannot run independently.
If a deposit is moved from Hong Kong to New York, it changes not only entities but also the jurisdiction of taxation. The client may owe withholding, reporting, or declaration obligations.
But if the token just moved on Base (i.e., the public blockchain) and not on JPMorgan’s systems, then the client’s liability has shifted in effect — but not in law. This creates a mismatch between economic activity and tax visibility.
Addressing this requires a completely new mechanism to avoid unreported flows — and also creates a mismatch between cost centers: which company has done what work, and who can charge whom for what. This represents a high risk of inadvertently facilitating tax evasion, or necessitates a massive investment in replicating bookings that the system was designed to avoid. It makes true efficiency impossible.
This would explain why this initial launch on Base is described as a PoC. JPM defined the purpose as follows:
Deployment of JPMC’s smart contracts on a public blockchain.
Issuance, transfer, and redemption of JPMD.
Trusted execution of JPMD transactions using approved-listed blockchain addresses.
Really? This franchise was set up in 2020 ish with around 100 employees according to the google search AI. And 5 years later they still need a PoC to test how to deploy a smart contract on an Ethereum chain? The minimum infrastructure requirements for this are sufficiently met by an iPhone.
What are they saying then?
We are participating in the narrative, I guess,
even if the infrastructure itself is neither new nor adequate for what we would need. That’s more like it.
And so they finish this overview: product availability is subject to internal review and continued engagement with our regulators.
So how likely is it then that JPM is serious about this whole thing?
Hm?
To be fair, a lot is written these days about Stablecoin that makes me wonder …
Reuters offers this
“Stablecoins are pegged to highly liquid assets such as the U.S. dollar and the tokens can drive demand for U.S. Treasuries by requiring issuers to hold large, liquid, and safe reserves to support a 1:1 peg to the greenback.”
Water is wet. They way they write about highly liquid assets sounds like they don’t know what a dollar is? Peg is also technically the wrong word.
Reuters found some help by brokerage [firm?] Bernstein:
"Once passed into a law (likely by the end of summer), we expect stablecoins to evolve from the money rail of crypto to the money rail of the internet,"
What on earth does this mean? Have you ever seen a money rail and where does it go? Bernstein, money is not like broadband, FYI.
Anyway, I previously spoke about what is needed to make blockchain a tokenization black belt:
Transfers must be initiated by the receiver, not the sender. This solves all ALM issues. It creates an ability to represent tiered holding structures without friction. No more pseudonymity traps — the sender can’t fabricate an origin story. Receiver defines intent — “Send me X amount with this reference hash.” Built-in transaction reconciliation — the reference allows the receiver to trace and verify the origin. KYC is solved by design — the receiver knows who they expected money from. Fraud window collapses — unsolicited transfers lose power, phishing dies, misattributed funds disappear.
Let me share another idea.
This one is brilliant — even by my standards!
I’m very sincere. This is a key precondition for any hope that tokenization could work. There is more though: I have some fun ideas about what the required structure is. We can basically do without Chainlink and Chainalysis (names selected solely for alliteration effect, that space basically), because we can solve for what companies like them do in a fully trust-minimized way. Trust me!
Crypto Markets Contradict Core Financial Theory. But That Doesn’t Mean the Theory Is Wrong.
We are wrong how we apply it,
What we observe in crypto directly contradicts some foundational principles of financial theory. Tether dominating despite high cost, an undifferentiated product. It doesn’t guarantee redemption, and it’s expensive to move on Ethereum. And yet, Ethereum ‘leading’ despite structural cost inefficiency
And any explanation cite is not more a cute story not an argument..
It was first to support real DeFi and NFTs,
It has the largest developer ecosystem,
It is considered "legitimate" by institutions.
None of this means anything. Staking as “security” is BS. It is mostly a simulation of responsibility, not actual capital risk. And validators don’t do anything other than following EVM rules. It’s illusionary to think this adds value.
In crypto, the real asset is the herd, and you can’t take it with force. Try, and they’ll just walk to another pasture.
Ethereum-like systems, where each asset has bespoke logic (smart contracts), break the clean separation between trading and settlement. This distorts the price signal and introduces randomness that traditional financial infrastructure avoids.
But Stablecoin are somehow infrastructure-ish and we do not have Coca Cola Shares doing something like this for say a stock loan involving Apple Shares and helping out DTCC a bit. But that is exactly true for Stable Coin.
But why?
The EVM Ignores the Off-Chain collateral of Stablecoin — So It Doesn’t Exist On-Chain
The Ethereum Virtual Machine (EVM) doesn’t “see” or “verify” what backs a stablecoin.
It doesn’t know if 1 USDC equals $1.
It doesn’t know if Circle is solvent.
It doesn’t know if tomorrow exists.
On-chain, the EVM only recognizes:
Token contracts with internal logic (balances, transfers),
Not external guarantees or legal contracts.
On-chain, “stable” is not a function of any of any off-chain collateral.
Stablecoins claim to represent an external state ("$1").
But that state is never validated by the system that moves the coin. Therefore, on-chain it has none
So even if someone says: “Don’t worry, USDC is backed 1:1 by Treasuries.” A DeFi protocol cannot and should not differentiate that from: “My cousin promises he’ll pay you back.” because the DeFi ability to exercise the right of redeeming on their respective premise is the same: none.
In decentralized systems:
A “stablecoin” is just a token with a
price = 1label.That label is used as an input to protocol logic (e.g. collateral ratio, liquidity pool math).
The actual stability is not enforced by anything on-chain unless there’s a mechanism (e.g. DAI’s liquidation engine or crypto overcollateralization).
The reason is that we need this info in token form to enable the DeFi constructl
Fiat-ed Coins is the answer when Backed vs. Unbacked is Irrelevant On-Chain
If I use USDT or a made-up SWisNr1 in a lending protocol,
the smart contract doesn’t verify backing — it uses the peg as a logic rule.The utility of a stablecoin within DeFi comes from:
Token liquidity,
Protocol recognition (
token.symbol == "USD"),User expectation of
price = 1
Tether makes billions by arbitraging interest on reserves, but the users don’t get value from the reserves, they get value from the symbolic function of 1 USDT = 1 USD inside the DeFi machine.
And this function doesn't require collateral.
No Functional Difference in Smart Contract Context
If I mint an unbacked token that I promise will always be worth 1,
And I make it usable in DeFi protocols,
Then it’s functionally indistinguishable from a “backed” coin like USDT.
Unless you leave DeFi and go redeem it off-chain. But SWisNr1 is only a technical placeholder it has no value otherwise.
The “Backing” Doesn’t Even Benefit If You Exit to Fiat. The stable in Stablecoin is a data attribute — not a price stabilization mechanism.
Stablecoins don't exist because people want dollars. They exist because DeFi needs a conversion convention in the form of token — and USD via USDT just happens to fill that slot. But the collateral is irrelevant, the value can be declared even if it has none. Their function is computational, not economic.
Liquidity pools reveal this structure
A pool of Apple ↔ 1USD
A pool of Mercedes ↔ 1USD
Then: Apple / Mercedes = (Apple / 1USD) × (1USD / Mercedes)
The stablecoin cancels out. It is the pivot, the intermediate constant, not the economic object.
The question is not whether a stablecoin can be redeemed at par, but whether its “1” can be used as a computational anchor. That’s why algorithmic stablecoins function — not because they are solvent, but because they are syntactically coherent as long as we use them correctly. They resolve cross-asset relationships through a shared unit, not a shared store of value.
DeFi doesn’t need “backing” — it needs predictable behavior.
The so called peg is a label but works without backing. We don’t need to give a company 100 billion to remind us of ‘1 USD’. That is the value proposition of Tether and all the rest including the collateral is folklore to make us forget that there is no need for collateral.
An unbacked token that simply allows to technically express a relation against a dollar is equally useful inside a smart contract if we use it only for this purpose. And we save a 100 billion or 1 trillion within a few years.
Stablecoin are a confusion. Backed or unbacked makes no difference. DeFi has no stablecoins, only a technical reference to 1.
Backing = TradFi thinking.
DeFi = Token as ledger stat
This is post-stablecoin, post-staking, post-oracle, and post-trading
Think about it. I am very serious. This must change if this market wants to survive.
We collectively handed over hundreds of billions to entities like Tether for the privilege of pretending that a token labeled “USD” is worth $1 — even though the software and protocols never check if it's true. There is no need at all. Billions are locked in reserves that could be used for productive investment, all to simulate what a single data point in a smart contract already does. Try to get any of that money back. Good luck.
I’m not telling you how to spend your money — I’m only saying this is an optional expense, and it offers no value over and above any generic token. That’s irrational.
This is 100% true.
I will do Decentralised Central Banking next time.
This is brilliant, Swen, thank you!! 🔥 I love that you not only peel back the layers of hype and educate on financial plumbing, you also give us all a new set of questions for the tool box.