Yesterday, Sweden’s central bank, the Riksbank, published a report on the second phase of its retail central bank digital currency (CBDC) pilot. It found that it could successfully integrate with commercial banking partners for distributing the e-krona. The central bank has some reservations about the DLT technology used, although the identified issues appear addressable and might provide insight for CBDC tests performed elsewhere.
Distribution of the e-krona is designed to be via commercial banks in a two-tier system. The central bank integrated the CBDC network with the legacy systems of consumer bank Handelsbanken and banking IT provider Tietoevry. The work demonstrated how the token-based e-krona can exist on a separate parallel DLT network but still integrate with the banking systems.
Riksbank explored two wallet options, a centralized one hosted by banks and a user-hosted wallet that worked more similarly to cash. One of the challenges with user wallets is that their identity is often a string of incomprehensible text such as “0x629a673a8242c2ac4b7b8c5d87”. Given this makes it tricky to send payments to another person, the central bank created human-readable aliases for the wallet addresses, using a centralized solution.
User-hosted wallets allow low value offline usage, which may increase penetration rates. However, the central bank has concerns about potential money-laundering risks, requiring further research.
Addressable DLT performance challenges
Performance testing of the DLT technology was one of the goals of the trials. R3’s Corda enterprise blockchain was used and found to perform satisfactorily at low transaction volumes. However, the central bank commented that Corda is “not specifically designed for a so-called retail CBDC that the e-krona would be (the platform is essentially designed for other types of financial transactions).” A more detailed reading indicates they believe the issues discovered are quite addressable.
The central bank observed a significant performance reduction when there’s a longer historical transaction chain. Corda uses the unspent transaction output (UTXO) model similar to bitcoin. For a nine kronor transaction, if you use a ten kronor coin, you receive a one krona coin as change. And the one krona coin has the transaction history, including all the history of the ten kronor coin. That’s something that doesn’t happen with account balances.
Apart from long transaction histories, UTXO also results in the fragmentation of balances in a wallet. This is similar to the numerous small coins you have as change from physical cash transactions. Imagine a long list of those coins in a digital wallet. This fragmentation of balances makes retailer wallets unwieldy and adds to the volume of transactions because several coins are used for single payments. In contrast to the bitcoin UTXO model, Ethereum uses an account balance model.
While phase 2 of the pilot did not test the potential solutions, there are straightforward proposals. For example, a retail trader who has received a lot of token payments could regularly merge these into a larger token. And tokens with a long transaction history could be redeemed for a new token.
It’s notable that this fragmentation issue may not be limited to blockchain solutions. MIT’s digital dollar work in the United States on Project Hamilton also uses the UTXO model but not blockchain. And MIT is now working with the central banks of England and Canada.
Corda’s storage of the token’s transaction history also creates confidentiality and data privacy issues. The professional participants on the network might be able to see the information, and Swedish banking legislation precludes participants from sharing confidential data. However, when it comes to data privacy, the central bank is unconvinced that blockchain transactions are compliant, requiring either a technological solution or a legal change.
Accenture was Riksbank’s technology partner for the pilot project.