Days before a controversial UPI fee kicks in, a new IIT Bombay technical report argues the charge should be slashed — and that banks, not merchants, should foot much of the bill. The report, "UPI at a Crossroads: Reintroducing the MDR" by IIT Bombay professor Ashish Das and former RBI monetary policy adviser Praggya Das, was released on October 11, 2026.
The study recommends cutting the merchant discount rate (MDR) on UPI transactions from the 0.40% fixed by the UPI Steering Committee on September 15 to 0.08% — matching NPCI's own proposal. The 0.40% framework, which applies to merchant transactions above ₹2,000, is scheduled to take effect from October 15, 2026, though trade bodies have asked the government to roll it back.
The report proposes splitting the 0.08% as 0.06% for third-party application providers (TPAPs) such as PhonePe and Google Pay and 0.02% for payment service providers (PSPs), which are mostly banks. It argues banks should surrender their MDR share because they already benefit disproportionately from low-cost CASA deposits: banks booked more than ₹4.85 lakh crore in net interest margin through CASA deposits in FY26. The authors suggest banks set aside about 3% of that — roughly ₹15,000 crore — to run and improve the UPI ecosystem instead of taking MDR revenue from offline transactions.
The study also warns that if merchants pass the MDR on to consumers through higher prices, small-ticket users would end up cross-subsidising large-ticket customers, eroding trust in a public digital infrastructure. It suggests restricting the chargeable base to merchants with annual turnover above ₹50 crore — about 90% of NPCI's proposed MDR base — and questions the reasonableness of a ₹300 transaction charge cap for P2M payments of ₹75,000 and above.
Why it matters for founders
Fintech and consumer founders should model for a lower — or reworked — UPI fee regime. An MDR that bites only large merchants changes unit economics for payment-led businesses, and any rollback or revision before October 15 will reset pricing assumptions across the board.