This new report from PositiveMoney, a progressive UK nonprofit, has a great treatment of the question:

Fundamentally, there are two main types of payment systems or transfer mechanisms: token-based and accountbased systems (Kahn et al, 2018, pp.8-11). Both are record keeping arrangements, but they are distinguished by their identification requirements.

The primary example of a token-based payments system is cash, where what really matters is the identification of the payment instrument: making sure that the note or coin is not counterfeit. It also constitutes a decentralised payment system, as the clearing process occurs bilaterally when the money instrument is transferred between the transacting parties. In this case, the cash itself is the record, acting “as a device that summarizes past production, trade and consumption decisions” (Kocherlakota, 1998, in ibid. p. 8).

The most important account-based payment systems in most modern economies are commercial bank deposits and central bank reserves (CBRs). What is crucial for account-based systems is the identification of the accounts of the transacting parties so the transfer of funds can happen and the transaction to be cleared. This record keeping is done by a single trusted party (such as the central bank when banks transact using central bank reserves) or by third parties (for example, when transactions happen between two people who have accounts at different banks), usually for a fee.

An account-based CBDC requires a centralised payment system to clear transactions (controlled by a trusted party or trusted third parties), whereas a token based system could be decentralised, with clearing happening between those conducting any given exchange.

Crypto folks will recognize these approaches under the names permissioned and permissionless. Two quick points:

1. Any serious proposal for a digital dollar should be either token-based, or include a token-based component in addition to an account-based option. Digital cash must have at least feature parity with physical cash.
2. It’s fascinating that the proponents of CBDC who seem to have all the energy these days are not the usual suspects, but modern monetary theory types who want to are no fans of commercial banking.