The concern about CBDCs for retail payments is that they will compete for deposit funding thereby crowding out banking and reducing output.
The paper found a cash-like CBDC would be more effective than a deposit-lie CBDC promoting both consumption and welfare. It could even crowd-in banking even where bank market power was absent. In its calibrated model the maximum benefit could be an increase of bank intermediation of 5.8% and the capture of 25% of the payment market. A deposit-like CBDC could crowd-out intermediation up to 2.6% and grab a market share of about 16.7%.
Properly designed therefore, the paper found that the concerns about retail CBDCs are not warranted.
The model used
The starting assumptions were that banks finance investments by issuing deposits to households. Goods produced by investments are sold to households in a frictional retail market. Households have a portfolio of cash and deposits to finance trade.
Three types of transaction were studied:
1. Where cash is the only viable payment options (eg. offline payments)
2. Where only deposits and credit can be used to pay (eg. online payments)
3. Where cash, deposits and credit can be used to pay (eg. most physical retail stores)
As deposits are used in transactions, the implied liquidity premium lowers banks’ funding costs.
The economic cycle sees bank deposits being used to fund investments that produce goods that are bought in retail transactions. The results is inter-market price linkages across different types of retail transactions and there is a feedback effect from retail transactions to deposit creations. The model also allows cash and deposits to compete directly (type 3).
General equilibrium effect
The general equilibrium effect of CBDC is decomposed across three channels:-
1. Payment Efficiency Channel (PEC). Interest bearing CBDC lowers the opportunity costs of holding payment balances and increases payment efficiency promoting aggregate demand for consumption and investment.
2. Price effect. Increased aggregate demand raises price levels thereby reducing the quantity of trade in all types of transactions, including where CBDCs are not used.
3. Bank funding costs. CBDCs may induce banks to increase interest paid on deposits increasing banks’ funding costs and reducing consumption and investment. This could crowd-in bank intermediation even if there is no market power.
If a cash-like CBDC is interest bearing it lowers the opportunity cost of holding payment balances and, therefore, people will buy more goods in transactions where a CBDC is used.
There is likely to be an indirect effect of banks increasing interest on deposits causing households to buy more transactions where deposits are used. This could lead to a spill over effect from cash to non-cash transactions. The PEC, leading to higher consumption demand, will induce banks to create increased deposits to finance production in order to clear the goods market. This indirect effect creates a feedback effect from transactions to deposit creation.
The paper found that cash used in type 3 channels will create a positive effect through the PEC than can outweigh the other two channels leading to higher consumption, intermediation and welfare.
The conclusion was that deposit-like CBDCs may not promote consumption and banking. They can’t be used with type 1 payments and therefore do not increase payment efficiency. They increase price levels reducing type 1 consumption, further worsening the allocative efficiency.
The PEC may also be weaker because cash is not interest bearing. It can, therefore, introduce an endogenous reduction in the opportunity cost of holding cash so that there is no spill over effect from non-cash to cash transactions. The result is the crowding-out of banking and lower output.
Findings and recommendations
The model was calibrated with US data. Cash-like CBDCs with a nominal interest rate of 5% were studied and the maximum effect was found to be an increase in bank intermediation of 5.8%, an increase in retail consumption of 3.5% and CBDCs capturing 25% payment market share. These increases were driven by the PEC.
Deposit-like CBDCs weakened the PEC gains. At maximum effect there was a 16.7% payment market share and banking was crowded out by 2.6%.
The two design recommendations in the paper are that a retail CBDC should serve type 1 (offline) and 3 payments, and that they should be interest bearing, or have other perks, to reduce the opportunity cost of holding payment balances.