The impact of a central bank digital currency on payments at the point of sale

Introduction

According to the Bank of Canada’s Methods-of-Payment (MOP) surveys in recent years, more than 80% of retail consumer purchases in Canada occur at physical locations.1 This suggests that such point-of-sale (POS) transactions are critical for the relevance of a central bank digital currency (CBDC).

This note summarizes recent research on the impact a CBDC could have on payments at the POS (Engert, Shcherbakov and Stenzel 2024).2 We aim to answer these questions:

• How would consumers and merchants react to the issuance of a CBDC?
• What share of the payments market at the POS would a CBDC capture?
• How would a CBDC affect use of the major methods of payment at the POS, that is, cash, debit cards and credit cards?

Model and simulation features

We develop a structural, empirical equilibrium model of interactions between consumers and merchants at the POS. Consumers and merchants interact in two stages in the model. In the first stage, consumers decide which payment methods to adopt, and merchants decide which methods to accept. Consumers make their adoption decisions based on the expected benefit from using each method, given their understanding of merchant acceptance practices. The model also takes into account a wide range of empirical features, including:

• consumer perceptions of different payment methods
• consumer awareness of merchant acceptance practices
• consumer costs of adopting and using different payment methods
• population demographics
• merchant costs of accepting different payment methods
• credit card rewards
• whether consumers carry a balance on their credit card or pay it off each month

For their part, merchants could miss out on sales if consumers are unable to use their preferred payment methods and, as a result, take their shopping elsewhere. This prospect, along with cost considerations, influences merchant acceptance decisions. In the second stage of the model, consumers and merchants interact at the POS to complete a series of transactions and thus determine which payment methods to use. The ability to use a given method reflects the adoption and acceptance choices made in the first stage.

We estimate the model using rich survey and diary data on Canadian consumers and merchants. These data include consumers’ perceptions (or ranked evaluations) of key features of the major payment methods—that is, the ease of use, security and cost of cash, debit cards and credit cards. Merchants are price takers in the model, meaning that they accept prevailing market prices. The model focuses on small and medium-sized enterprises (SMEs) for model tractability and especially because the available merchant data come from such firms. However, in our analysis, we also discuss the likely implications of including large retailers.

With our model of consumer and merchant decision-making, and knowledge of consumer perceptions of the key features of different methods of payment, we simulate the introduction of two kinds of a CBDC: a debit-like version and a best-of-both (cash and debit) version. The debit-like CBDC incorporates consumers’ perceived features of debit transactions regarding ease of use, security of use and affordability. The best-of-both CBDC incorporates the features of either cash or debit that are most preferred by consumers regarding ease of use (including universal offline capability), security of use (including anonymity) and affordability.

Adoption cost frictions

A key feature of our analysis concerns frictions that can deter the adoption of a new payment method, such as a CBDC. On the one hand, the public nature of a CBDC might suggest that consumers would bear moderate adoption costs, which could be affected by policy decisions such as subsidies or price regulation. On the other hand, consumers would likely encounter substantial non-monetary costs, which include learning about the CBDC and its related technology, as well as acquiring and managing a digital wallet. Costs faced by consumers could also reflect the central bank recovering expenses related to development as well as the ongoing maintenance, distribution and marketing of a CBDC.

Therefore, we compare the effects of adoption cost frictions with an ideal CBDC baseline. To simulate this ideal baseline case, we impose a small adoption cost just for those consumers who use only cash and add a CBDC (i.e., a digital payment method) to their cash-only payment bundle. We set the adoption cost for this payment bundle equal to the estimated mean net cost of adding debit to a cash-only bundle, which is equal to $0.33 per month in our data. (The median consumer in our dataset obtains a net benefit of $0.74 per month when adding debit to a cash-only bundle.) And we set the cost of adding a CBDC to any other bundles of payment methods held by consumers—[cash, debit] and [cash, debit, credit]—to zero. Few consumers in our dataset have cash as their only payment method. This means that, for the vast majority of consumers in this ideal case, the net adoption cost of a CBDC is zero.

As noted, some costs would likely be borne by consumers when adopting a novel payment method like a CBDC. So we also conduct simulations that include an adoption cost friction. In these scenarios, adding a CBDC to any payment method or bundle is $x more costly than in the above ideal CBDC case, where x varies from negative $2 per month to $5 per month. Technically, we impose this adjustment cost, and consumers then value the benefit of the CBDC and decide whether to adopt it, based on their assessment of its net benefit. As a frame of reference, the average fee for a bank account in Canada is about $6 per month.3 Therefore, these simulations consider costs that do not appear unusual in the market and correspond to a fraction of typical costs.

CBDC introduction and the effect of adoption cost frictions

Financial inclusion—access to financial products and services—in Canada is high, with widespread access to major payment methods. So a new payment method such as a CBDC is unlikely to lead to meaningful expansion of the payments market because it is essentially covered. Accordingly, when assessing the introduction of a CBDC, we consider the size of the payments market in our analysis to be fixed, and we focus on how market shares of the major existing methods of payment—cash, debit cards and credit cards—are affected by the new method.

Table 1, Table 2 and Table 3 present key simulation results for the introduction of the debit-like CBDC and the best-of-both CBDC.

Table 1: Results for a CBDC with an adoption cost friction of zero (ideal CBDC case)

Note: CBDC is central bank digital currency. A simulated debit-like CBDC incorporates features of debit transactions regarding ease of use, security of use and affordability. A simulated best-of-both CBDC incorporates the features of either cash or debit that are most preferred by consumers regarding ease of use (including offline capability), security of use (including anonymity) and perceived cost of use.

Table 2: Results for a CBDC with an adoption cost friction of $1.50 per month

Note: CBDC is central bank digital currency. A simulated debit-like CBDC incorporates features of debit transactions regarding ease of use, security of use and affordability. A simulated best-of-both CBDC incorporates the features of either cash or debit that are most preferred by consumers regarding ease of use (including offline capability), security of use (including anonymity) and perceived cost of use. The annual subsidy for a debit-like CBDC to move from this case to the results in Table 1 is about $440 million per year. This is calculated as follows: $1.50 (adoption cost friction per consumer per month) × 34 million adult Canadians × 0.72 (target adoption rate) × 12 months. The target adoption rate of 72% is the consumer adoption rate that results from the ideal CBDC scenario, where the adoption cost friction is zero.

Table 3: Results for a CBDC with an adoption cost friction of $3.00 per month

Note: CBDC is central bank digital currency. A simulated debit-like CBDC incorporates features of debit transactions regarding ease of use, security of use and affordability. A simulated best-of-both CBDC incorporates the features of either cash or debit that are most preferred by consumers regarding ease of use (including offline capability), security of use (including anonymity) and perceived cost of use. The annual subsidy for a debit-like CBDC to move from this case to the results in Table 1 is about $880 million per year. This is calculated as follows: $3.00 (adoption cost friction per consumer per month) × 34 million adult Canadians × 0.72 (target adoption rate) × 12 months. The target adoption rate of 72% is the consumer adoption rate that results from the ideal CBDC scenario, where the adoption cost friction is zero.

Chart 1 and Chart 2 show how consumer adoption changes as adoption cost frictions vary.

Chart 1: Consumer adoption of a debit-like CBDC

Chart 2: Consumer adoption of a best-of-both CBDC

• For the ideal CDBC case, where the adoption cost frictions are zero (Table 1), consumer adoption of the debit-like CBDC is widespread (72% of consumers adopt this CBDC), merchant acceptance is moderate (44% of merchants accept it at the POS) and this CBDC is used for 16% of transactions (19% of value).
• For the best-of-both CBDC in this ideal environment, we see greater consumer adoption (78%) and merchant acceptance (91%), and usage is 22% of transactions (27% of value). Note that this CBDC has better outcomes than the debit-like CBDC for the entire range of adoption cost frictions (Table 2 and Table 3). We consider results for the best-of-both CBDC to be the theoretical maximum, but in practice such outcomes are not feasible, for several reasons:
  • This simulated best-of-both CBDC has some features of cash that are not readily attainable, such as anonymity of use and universal offline capability.
  • The best-of-both CBDC embeds features of cash and debit on an individual consumer basis—each consumer gets what they most prefer—which is unachievable in practice given the variety of consumer preferences.
• As adoption cost frictions increase, consumer adoption of a CBDC falls sharply, as shown by the blue lines in Chart 1 and Chart 2. In Table 2, we see that with adoption cost frictions of $1.50 per month, for example, consumer adoption of the debit-like CBDC is 43%, merchant acceptance is 25%, and usage is 11% of transactions (13% of value). For the best-of-both CBDC, we have moderate consumer adoption (50%) and merchant acceptance (42%), and usage is 16% of transactions (19% of value). With adoption cost frictions of $3 per month, usage for both types of CBDC is less than 10% of transactions (Table 3).4

Usage shares for the debit-like CBDC in the model simulations are drawn mainly from pre-existing usage shares of debit and credit cards, with relatively limited impact on cash use. For the best-of-both CBDC, cash shares are less robust. Note that there is little change in consumer adoption of debit and credit cards in these simulations. However, their use at the POS is negatively affected, which can lead to merchants reducing their acceptance of those payment methods. This latter impact is driven by cost-benefit considerations in the model. Merchants’ expected profits are influenced by their understanding of average consumer payment preferences and by the acceptance costs that they face. So, if consumer use of a payment method that is relatively expensive for merchants (such as credit cards) declines—even slightly—then some price-taking merchants in the model respond by not accepting that payment method. This reduction, in turn, undermines its use at the POS.

Simulations of related models (based on Huynh et al. 2020) that do not include this particular channel for merchant response suggest that CBDC issuance would be unlikely to significantly affect credit card use. Similarly, if our model included large retailers with some degree of market power, these substitution effects would likely be weakened (discussed further below). The bottom line here is that the potential impact of a CBDC on the usage shares of cards—particularly credit cards—is likely exaggerated in these simulations.  

Subsidies for consumer adoption of a CBDC

In reality, consumer adoption of a novel payment method is unlikely to be costless, so cases that include adoption costs are more relevant than the ideal case. If authorities wish to support the introduction of a CBDC under such circumstances, they could provide subsidies to offset adoption cost frictions. Our simulations suggest that the cost of such subsidies could be large. With a debit-like CBDC and adoption cost frictions of $1.50 per month, for example, moving consumers to ideal CBDC outcomes would require a subsidy of about $440 million a year (Table 2).

Responses by financial institutions and payment service providers

After a CBDC is issued, incumbents in the payments market—financial institutions on the acquiring and issuing sides of the market as well as payment service providers—are likely to respond to maintain their market shares. As noted earlier, one implication of CBDC issuance in our simulations is that merchants potentially reduce their acceptance of debit and credit cards. Accordingly, we assess incumbent responses aimed at restoring debit and credit card acceptances to 90% of their pre-CBDC levels.

Consider the ideal CBDC case with a best-of-both CBDC (Table 1). Suppose that incumbents respond by lowering merchant acceptance fees along with merchant and consumer usage costs to restore 90% of the pre-CBDC acceptance levels of debit and credit cards. This leads to a 9% reduction in the costs associated with credit cards and a 3% reduction in the costs associated with debit. The ideal CBDC case with debit-like features triggers only a 5% reduction in credit card costs and a 0.8% reduction in debit costs.

However, consumer adoption costs reduce the price adjustments needed to restore incumbents’ market shares. With moderate adoption cost frictions, a CBDC fails to gain substantial market share and has little impact on market prices through competitive pressure. For example, when adoption cost frictions are $1.50 per month, a debit-like CBDC induces close-to-zero incumbent price responses.

Related considerations

A taste for variety

A striking aspect of some of our simulations is that a CBDC attracts market share at the expense of payment methods that already exist in the market, even though consumers perceive the CBDC to be very similar or even identical to those payment methods. To some extent, adoption of a CBDC that is similar to existing methods is due to a characteristic of consumer preferences in models that are widely used to study such (two-sided) markets. That is, consumers in such models have a preference or taste for variety in products, leading them to perceive benefits from any new payment method, even if it is identical to existing methods. This is typically rationalized as curiosity, a desire for security or resilience, or the effect of persuasive advertising. Other things being equal, this preference supports simulated consumer adoption, merchant acceptance, and use of a new payment method such as a CBDC.

The effect of large retailers

As noted, merchants in the model are price takers, and our merchant data focus on SMEs. How would inclusion of large retailers affect our results? The acceptance practices of large retailers would probably be less responsive to a change in consumer behaviour than those of smaller retailers. This is for two reasons:

• Large enterprises have a greater ability (market power) than SMEs to protect their profit margins in response to modest changes in consumer payment preferences without changing their acceptance practices.
• Most large retailers tend to accept all major existing payment methods, including cash, debit cards and credit cards.

These factors suggest that, after the introduction of a CBDC, large retailers are likely to have more robust acceptance practices for existing payment methods than SMEs. Therefore, including large retailers in the model would be unlikely to lead to a meaningful gain in simulated CBDC outcomes compared with those discussed above. Further, in a scenario in which all large retailers accepted a CBDC (perhaps due to subsidies that target large-retailer acceptance), the main outcome that would be substantially affected is their acceptance of the CBDC, with only minor effects on consumer adoption and use. This is consistent with our results showing that it is consumer behaviour that largely drives outcomes for payments at the POS.

Conclusion

We develop a structural model of the payments market, featuring decisions involved in consumer adoption, merchant acceptance, and use of different payment methods. The model incorporates a wide range of empirical features, including consumer perceptions of different payment methods, consumer awareness of merchant acceptance practices, population demographics, and merchant costs and acceptance decisions. We use this model to simulate the introduction of two types of CBDC: a debit-like version and a best-of-both version that embeds those features of cash and debit most preferred by consumers. Our key conclusions are as follows:

• In the absence of adoption cost frictions, the simulations suggest that consumer adoption and merchant acceptance of a CBDC could be significant and use of a CBDC at the POS could be considerable. Usage shares of a CBDC come primarily from a corresponding decrease in debit and credit card use, but with an increased share from cash when the CBDC incorporates some cash-like features, including anonymity and universal off-line capability (which are unlikely in practice). Inclusion of large retailers in our dataset would probably weaken these effects, so market shares of credit cards would likely be more robust in practice than in these simulations.
• An absence of consumer adoption costs is unlikely in reality. Our simulations suggest that even modest adoption cost frictions substantially reduce consumer adoption, merchant acceptance, and use of a CBDC. Consumer subsidies could offset the effects of these frictions, and the costs of such subsidies could be large.
• If a CBDC does not include compelling features, it would have little effect on market outcomes. But developing and launching a competitive CBDC or providing subsidies to promote its adoption would be costly and would likely motivate a response from incumbent financial institutions and payment service providers to recover their market shares. Our simulations suggest that this incumbent response does not have to be substantial to restore their market shares. This means that the competition benefits of a CBDC in terms of market price or cost reductions are small.

References

Engert, W., O. Shcherbakov and A. Stenzel. 2024. “CBDC in the Market for Payments at the Point of Sale.” Bank of Canada Staff Working Paper No. 2024-52.

Halaburda, H., S. J. Kim, O. Shcherbakov and A. Stenzel. 2024. “Interchange Fee, Market Structure and Excessive Intermediation in the Payments Market.” Bank of Canada mimeo.

Henry, C., M. Shimoda and D. Rusu. 2024. “2023 Methods-of-Payment Survey Report: The Resilience of Cash.” Bank of Canada Staff Discussion Paper No. 2024-8. 

Huynh, K. P., J. Molnar, O. Shcherbakov and Q. Yu. 2020. “Demand for Payment Services and Consumer Welfare: The  Introduction of a Central Bank Digital Currency.” Bank of Canada Staff Working Paper No. 2020-7.

Huynh, K. P., G. Nicholls and O. Shcherbakov. 2022. “Equilibrium in Two-Sided Markets for Payments: Consumer Awareness and the Welfare Cost of the Interchange Fee.” Bank of Canada Staff Working Paper No. 2022-15.

Acknowledgements

We thank Kim P. Huynh for comments and discussion of the issues considered in this work. We also thank Scott Hendry, Francisco Rivadeneyra and participants in Bank of Canada seminars and workshops for their comments. For excellent editorial assistance, we thank Maren Hansen and Meredith Fraser-Ohman.

Endnotes