The two-tier system is straightforward in principle and slippery in practice, because the vocabulary travels poorly between Web3 and TradFi audiences and the headline labels on tokenised products often elide the legal distinction the chapter has been insisting on. This part collects the misreadings that show up most often in production discussions, then walks through the worked examples in APAC and the United States that illustrate the tiering at work.
Common confusions
A tokenised deposit is not a stablecoin. The tokenised deposit is commercial bank money on chain, issued by the bank, recorded on the bank's books, with deposit insurance and prudential supervision in scope to the extent the structure preserves them. A payment stablecoin is an e-money or equivalent instrument issued by a non-bank or specialised issuer, with reserves rather than a bank deposit liability backing it. The two sit on different parts of the regulatory map. See Stablecoin types for the full taxonomy, and Chapter VII for tokenised deposits specifically.
A wholesale CBDC is not a retail CBDC. Wholesale CBDC is tokenised central bank money for use among institutions with reserve account access, an extension of the existing reserve system onto programmable rails. Retail CBDC is tokenised central bank money issued to non-bank holders, a structurally new instrument that competes, in some accounts, with bank deposits. The two are entirely different policy objects. Most APAC central banks are leaning into wholesale work and approaching retail with caution.
The token does not change which tier the money sits in. A bank deposit is commercial bank money whether it is recorded only in the bank's general ledger, recorded in the ledger and on a permissioned chain, or recorded across a multi-bank chain operated by a consortium. The legal claim is the same. The settlement and operational mechanics may be entirely different, and the differences are where most of the engineering and legal work lives.
Reserves are not "the central bank's deposits at commercial banks". Reserves run the other way. They are deposit liabilities of the central bank, owed to commercial banks. The direction of the arrow matters because it determines who is the creditor and who is the debtor, and confusing the two corrupts every analysis that follows.
Deposit insurance is not a guarantee that all deposit money is risk-free. It is a guarantee that covered depositors are made whole up to a cap, on a defined timeline, when an insured bank fails. Wholesale flows, corporate balances, and most institutional deposit positions sit far above the cap. The way the system manages credit risk for those flows is through the wholesale settlement layer, the lender-of-last-resort framework, and supervisory oversight, not through the insurance scheme.
Where it appears in production
Project Ensemble is the worked example of the tiered model in APAC. The HKMA operates the wholesale CBDC layer for interbank settlement. Participating commercial banks operate the tokenised deposit and tokenised asset layers. The architecture is explicit about which tier each instrument sits in, and design discussions through 2024 and 2025 have refined the interface between the layers in ways that map onto the BIS unified-ledger research. Hong Kong's broader payment infrastructure, including CHATS, sits underneath as the conventional rail. The HKMA's wholesale CBDC and tokenisation page is the public-facing reference for the programme.
The Payment Innovation Project in japan sits in the same conceptual space, with megabank consortia working with the Bank of Japan and the FSA Japan on tokenised deposit infrastructure. Mizuho and Nomura participate alongside JSCC for the post-trade piece. The architectural conversation is the same. Which tier of money settles which leg of the trade.
In the United States, the Federal Reserve's Project Cedar and the Regulated Liability Network work explored the tiered model on permissioned rails, with central bank money and commercial bank money represented as distinct token classes interoperating across the same ledger. JPMorgan's Kinexys Digital Payments runs tokenised deposits in production today, settled in commercial bank money, with the wholesale settlement layer continuing to run on Fedwire in central bank money. The two layers are not merged. They cooperate.
Digital Asset's Canton Network is increasingly the connective tissue between tokenised deposits, tokenised central bank money pilots, and tokenised collateral and securities. The Tokenised collateral use case is one of the cleanest examples of why the tiering matters. Collateral movements need atomic delivery against payment, and the payment leg has to settle in something with credible finality, which lands you back at the question of which tier.
What to read next
Chapter IV, Settlement finality, picks up the question of when a transfer in either tier is legally final, which is the question this chapter has implicitly been pointing at every time it used the word "settle". Chapter VII goes deep on tokenised deposits as the commercial bank money on chain category. Chapter VIII covers wholesale CBDC and the design choice between a bank-issued tokenised deposit and a central-bank-issued wholesale CBDC for the same wholesale settlement use case. Chapter IX takes the editorial wedge, agentic commerce on tokenised rails, and asks which of these money tiers is actually accessible to AI agents transacting on behalf of human or corporate principals. The institution pages for the HKMA, Bank of Japan, and Bank of Korea are the most direct way to track which tier each APAC central bank is prioritising in its tokenisation work, and the BIS and BIS Innovation Hub pages are where the policy framing for the tiered model is being set.