Distributed ledger technology experiments in retail payments: Evidence from Turkey

Abstract

A growing number of central banks are exploring the potential introduction of a central bank digital currency (CBDC), and authorities are conducting distributed ledger technology (DLT) proof-of-concept experiments and pilot programmes. While these pilot projects have confirmed the feasibility of using DLT platforms for large-value wholesale payments, it is not yet clear whether these platforms will provide major advantages in terms of speed and scalability for small-value/large-volume retail payments. In Turkey, the existing instant payment system has already demonstrated that it can settle as many as 1,980 transactions per second (TPS) and 11.5 million transactions per day, across its 25 participant institutions. Moreover, simulations in test conditions suggest that it has the potential to deliver thousands of transactions per second, and that it is scalable to hundreds of participant institutions. Any retail payment system built on a DLT platform must therefore, at the very least, perform at a comparable level. Using actual retail payment data from the day on which the peak TPS was observed, this paper describes a novel experimental investigation into the TPS and scalability performance of two open source DLT platforms (Hyperledger Fabric and Quorum) and Turkey’s instant payment system for retail payments, with the aim of providing valuable insights for the design of alternatives for future CBDC launches. The study presents several empirical findings. First, it demonstrates that the speed and scalability of selected DLT platforms are not yet mature enough to establish an enterprise retail payment solution. Secondly, the experimental design enables us to determine precisely which DLT platform is facing a performance and capacity bottleneck and under what circumstances. One of the most critical findings is that platforms with a modular architecture and more efficient transaction life cycle perform relatively better, although still lack the requirements for achieving reliable and efficient retail payments. Lastly, we suggest that it could be worth developing an integrated approach in which the central instant payment system and DLT-based digital currency solution communicate via a bridge platform.