The two-tier system makes sense only when you can see what each tier actually does at the operational level. Reserves at the central bank are not a generic bank account. Commercial bank deposits are not a passive store of value waiting to be lent out. The mechanics of both differ from the textbook story most readers picked up at some point, and the tokenisation conversation is much sharper once you have replaced the old model with the operational one.
The reserve account at the central bank
A reserve account is a deposit account held by an eligible institution at the central bank. Access is restricted, by statute and by central bank policy. In most jurisdictions, only commercial banks with full banking licences hold reserve accounts directly. A handful of non-bank participants, typically clearing houses, central securities depositaries, and select payment system operators, may hold reserve accounts under specific designation regimes. The set is small and gated for a reason. The central bank's balance sheet is the apex of the payments hierarchy, and admitting a counterparty to it confers settlement-grade trust by association.
What reserves are at law. They are deposit liabilities of the central bank, denominated in domestic currency. They are settlement assets for the institutions that hold them. They earn interest, or do not, at a rate set by the central bank as one of its policy levers. Reserves at the Bank of Japan, the HKMA, the Bank of Korea, the Federal Reserve, and the European Central Bank (ECB) each have their own remuneration regime, and the differences matter to bank treasury teams in ways that look invisible from outside the bank.
What reserves cost. The opportunity cost of holding reserves is what the bank could have earned by deploying that liquidity elsewhere, net of what the central bank pays on the balance. Required reserves, where they exist, are the minimum a bank must hold against its deposit liabilities under prudential rules. Excess reserves are voluntary, held for intraday liquidity and to meet payment obligations. Liquidity coverage ratio (LCR) and net stable funding ratio (NSFR) requirements interact with reserves in ways that drive a great deal of bank treasury behaviour, well above what the headline policy rate alone would suggest.
Why reserves sit at the apex. They are the only settlement asset that carries no counterparty credit risk in domestic currency, because the issuer is the lender of last resort and prints the currency. A bank receiving a reserve transfer is paid in the only money that cannot bounce. Every tier below this tier, by contrast, settles by promising to pay reserves later, with all the credit risk that promise implies.
Commercial bank deposits as private credit money
A deposit is created in two ways, and the second is the one Web3 readers tend to miss. The first is the obvious one. You deposit cash or transfer in money from elsewhere, and the bank credits your account, receiving an asset and recognising a corresponding liability to you.
The second is loan creation. When a bank extends a loan, it credits the borrower's deposit account with the loan proceeds. No prior deposit is required. The bank simultaneously books an asset, the loan receivable, and a liability, the new deposit. The two are created together by the same accounting entry. This is how the bulk of broad money in modern economies is created. Central banks influence the rate at which this happens through interest rates, capital and liquidity rules, and macroprudential tools. They do not, in any direct sense, lend reserves out to be redeposited and re-lent in the way the old textbook money multiplier story implied.
The Bank of England's 2014 quarterly bulletin made the operational reality explicit. Authors Michael McLeay, Amar Radia, and Ryland Thomas set out the standard contemporary view, which is now shared across central banks, BIS policy staff, and academic monetary economists. The textbook model where banks act as passive intermediaries, taking in deposits and lending them out, mapped poorly onto how banking has actually worked for decades. The Bank of England paper is the standard external citation when the question comes up in a meeting.
What this means for tokenised cash
A tokenised deposit on a bank's chain is a representation of that bank's deposit liability. Minting a tokenised deposit token does not require the bank to receive new reserves into its account at the central bank. It requires the bank to recognise a deposit liability, which it does in the ordinary course whenever it receives a payment or extends credit. The token sits on the chain, the deposit sits on the bank's books, the two move together by reconciliation as covered in Tokenisation, defined.
Reserves do still come into the picture when a tokenised deposit moves between banks. If a tokenised deposit is transferred from a customer of Bank A to a customer of Bank B, Bank B is now owed by Bank A. That interbank position has to settle somewhere. In current production designs, the settlement happens on the conventional wholesale rail, in central bank reserves, often at end-of-day or on a continuous basis. In a tokenised wholesale CBDC (wCBDC) design, the settlement happens on the chain itself, with the wCBDC token moving from Bank A's wallet to Bank B's wallet in tandem with the deposit token transfer. The legal substrate is unchanged. Only the rail differs. This is the design space Project Ensemble and the BIS unified-ledger thesis have been working through.
The implication is structural. The tokenisation of deposits does not, on its own, expand the supply of bank money. It re-expresses the existing deposit liability on a programmable rail. New deposits are still created when banks lend, and that has not been re-engineered by the move to tokens. The mechanics of money creation sit upstream of the chain, not on it.