Should central bank digital currencies (CBDCs) become reality, all that might be required to safeguard bank funding structures would be a digital euro holding limit of €3,000 per person, says the European Central Bank (ECB). The recommendation is based on a study published under the ECB’s Occasional Paper Series that focuses on the impact CBDCs could have on the balance sheets of banks and central banks.
The paper was motivated by an often-cited concern around the introduction of CBDCs: a successful retail CBDC could lead to retail customers shifting part of their funds away from bank deposits to central banks, thus leading to an increase in banks’ funding risks and a decrease in bank lending.
Applying a constraint optimisation model and using euro area bank balance sheet data, the study simulated the introduction of a fictitious digital euro in Q3 of 2021. The impact was analysed under different liquidity risk tolerance scenarios. With the baseline scenario, described as where banks have “an intermediate liquidity risk appetite and wish to keep half of the bank-specific voluntary liquidity buffers they hold in excess of the regulatory minimum”, a digital euro holding limit of €3,000 per person was found to be sufficient to limit significant changes to bank funding structures and prevent the need for additional Eurosystem funding.
The same analysis was made with data from 2019, when reserves were lower, and with the model specification adjusted to focus on a “segmented rather than perfectly functioning interbank market” and a “bank run where there is not interbank market”. Results were similar to that from the baseline scenario.
A time and place
Although the model “could be used to guide policy makers’ decision making on the design and timing of a future digital euro introduction”, ECB cautions that “economic situations, market rates and bank balance sheets are subject to change” – particularly if a CBDC introduction were anticipated. Therefore, should the time come for a CBDC to become reality, policy makers should rerun the model.