India's payment system has begun answering a basic question: who pays for it? The government's decision to introduce a Merchant Discount Rate (MDR) on a narrow segment of higher-value Unified Payments Interface (UPI) transactions marks a shift from the zero-MDR architecture that helped drive digital-payment adoption. It also raises a second question: could some of the underlying cost of higher-value payments be reduced through a different settlement architecture?
The Cost of Keeping UPI Free
UPI's explosive growth was accelerated by a consequential policy choice. In December 2019, the government zeroed out the Merchant Discount Rate on person-to-merchant UPI transactions, removing the processing fee that merchants would otherwise pay for accepting digital payments. Annual UPI transaction value rose from ₹21.3 lakh crore in 2019–20 to over ₹260 lakh crore by March 2025, and to roughly ₹314 lakh crore across more than 24,000 crore transactions in 2025–26 alone.
That growth was never free; its costs were absorbed elsewhere. Banks, payment providers and other participants continued to bear infrastructure, fraud-management and processing costs, while the government partly offset these through an incentive scheme for low-value merchant transactions. The broader financing question has therefore persisted even as users and merchants experienced UPI as effectively free.
The latest MDR framework begins to change that arrangement. It preserves zero MDR for the overwhelming majority of low-value merchant payments while allowing a narrow higher-value segment to contribute towards the cost of the payments ecosystem.
The distribution of UPI transactions explains why such segmentation matters. Only about 4 percent of person-to-merchant UPI transactions in 2025–26 exceeded ₹2,000, yet these accounted for nearly two-thirds of P2M UPI payments by value. The vast majority of transactions are therefore low-value, even as a small higher-value segment accounts for most of the money moving through merchant UPI.
This opens a wider policy question. Beyond deciding how much MDR should apply to higher-value transactions, policymakers can examine whether those payments could be settled through an architecture with lower underlying costs.
A Different Settlement Architecture
A central bank digital currency, known in India as the e-rupee or e₹, is a direct digital liability of the Reserve Bank of India, rather than a claim on a commercial bank transferred through UPI's existing architecture. The distinction matters because e₹ transactions settle directly and finally in central bank money.
Retail e₹ payments would still carry costs. Wallet infrastructure, merchant acceptance, fraud prevention, cybersecurity, customer support, compliance and conversion between bank deposits and e₹ all require resources. The key issue is whether the different settlement architecture can reduce the end-to-end cost of processing some merchant transactions.
That is an empirical question worth testing.
The higher-value segment now being brought within the MDR framework offers a natural place to do so. Policymakers could test e₹ settlement for a defined subset of merchant transactions above ₹2,000 while leaving the low-value, high-volume core of UPI unchanged.
Such a pilot could also draw on infrastructure that already exists. The Reserve Bank of India has enabled interoperability between e₹ wallets and UPI QR codes, allowing the digital rupee to use much of the merchant acceptance network that UPI has built. The principal change would occur at the settlement layer behind the transaction.
This creates the basis for a relatively clear comparison: higher-value payments processed through the existing UPI architecture with MDR, alongside comparable payments settled through e₹.
The Constraints on Digital Rupee Settlement
A digital rupee pilot would nevertheless confront risks that matter even if it covers only a limited share of merchant payments.
The first is bank disintermediation. If consumers move deposits into e₹ wallets to make payments, wider adoption could reduce balances held with commercial banks. At sufficient scale, that could affect banks' funding base and their capacity to extend credit. A limited pilot would therefore need to measure how much money moves out of bank deposits, for how long, and with what consequences.
The second is privacy and data governance. A larger e₹ footprint raises questions about traceability, anonymity and who can access transaction-level information. Expansion into routine merchant payments would require clear rules governing the collection, retention and use of payment data.
The third is scale. Retail digital rupee circulation remains tiny relative to UPI's transaction volumes. An architecture that works in a controlled pilot would still have to demonstrate reliability, resilience and cost efficiency at substantially greater transaction volumes.
The fourth is behavioural inertia. UPI's dominance reflects a network effect built over several years. QR interoperability reduces the infrastructure barrier, although consumer adoption, merchant familiarity and confidence in a different form of digital money would still have to develop.
Together, these constraints make a limited pilot a more useful starting point than a large-scale shift.
From Fee Debate to Policy Pilot
India has now chosen one way of financing a narrow higher-value segment of UPI: larger merchant transactions will contribute through MDR while the low-value core remains protected.
That decision also creates an opportunity to compare settlement models before the financing architecture becomes entrenched.
A digital rupee pilot could establish whether direct settlement in central bank money materially lowers the total cost of higher-value merchant payments. The comparison should cover the full economics of the two systems: processing and infrastructure costs, merchant acceptance, payment failures, fraud, settlement efficiency, effects on bank deposits and the cost of operating the e₹ infrastructure itself.
The purpose of the exercise would be to establish where each settlement model works most efficiently and whether some higher-value transactions can be processed at lower cost through e₹.
The MDR debate is usually framed around who should bear the cost of digital payments. A second question deserves equal attention: can the architecture itself reduce that cost before policymakers decide how to distribute it?
Zero-cost UPI was an effective strategy for building scale and widespread acceptance. As the system matures, different categories of transactions may warrant different financing and settlement models. With MDR now returning for a narrow segment of merchant payments, India has begun answering the question of who pays. The next step is to determine whether the underlying cost of some higher-value payments can also be reduced.