Centralized fraud controls can price out small instant payments
Centralized fraud prevention can deliver lower welfare than decentralized controls when flat fees exclude small payments, Bank of Canada research finds. Access rules and transaction caps determine whether coordination delivers its benefits.
Coordination meets the access fee
Yuteng Cheng and Francisco Rivadeneyra of the Bank of Canada and Charles Kahn of the University of Illinois model instant-payment pricing, eligibility and fraud prevention.
Centralized operators coordinate sending-bank screening and receiving-bank monitoring, but flat fees can exclude socially valuable small payments.
Mandating access without a minimum transaction size and setting the efficient upper cap restores the model's efficient allocation, with least-cost tie-breaking for adoption-maximizing operators.
Without those mandates, decentralization delivers higher welfare in a numerical illustration: sending-bank fees exclude receiver-side costs, preserving access.
A simple bank liability split generally cannot deliver efficient prevention.
Equal shares need equal flexibility
With access fixed, optimal liability falls more heavily on the bank with more flexible prevention technology; a 50/50 split requires equal adjusted curvatures.
When sender costs raise access fees, liability tilts toward receivers even with symmetric technologies.
The UK splits reimbursement costs equally across sending and receiving providers, generally caps claims at £85,000 and requires most claims resolved within five business days.
Canada's planned Real-Time Rail requires centralized fraud services.
The authors' findings are not official Bank positions.
Access carries a financing constraint
The paper makes access and fraud prevention inseparable design choices.
Its efficient-access result requires separate financing when a fee affordable to the smallest payments cannot cover costs.
The governance ranking is conditional, not a blanket case for centralization or decentralization.