Memo
India's payments network monetises the wrong leg
The answer
India’s payments rails have separated volume from value, and then priced the volume side at zero. In 2026Q2, merchant payments were 63.9% of all transactions but only 23.0% of the rupees moved. Person-to-person transfers were the mirror image: 30.8% of transactions, 71.2% of value. The merchant leg is the only leg a merchant discount rate could ever be charged on, and under the zero-MDR regime it earns nothing.
The consequence, sized on one player’s disclosed data: 50,705,249 registered merchants transacting 487 times a quarter each, Rs 206,685 of GMV per merchant per quarter, at Rs 0 of transaction revenue.
Four supporting arguments
1. The base is enormous and still compounding. UPI has grown from 3,248 million transactions in 2021-07 to 23,658 million in 2026-07: a 49% five-year CAGR, off a series that starts at 0.1 million in 2016-07. Even now, growth has not decayed to maturity: the most recent year-on-year reading is 21.5% on volume and 19.1% on value (2026-07).
2. Growth is arriving in the leg that cannot be charged for. Between 2018Q1 and 2026Q2, Retail contributed 64% of all volume growth, 24.6 billion of the 38.5 billion additional quarterly transactions. Growth is not merely large, it is concentrated in the merchant leg that zero-MDR prices at nothing.
3. Growth is arriving as small tickets, which is the expensive kind. The average UPI transaction has fallen to Rs 1,263, down 22.7% from January 2019. Every incremental transaction adds switch, fraud and support cost while adding no fee income. Volume growth without price is a cost line, not a revenue line.
4. The revenue foregone is quantifiable, and it is not small. On the merchant GMV of 10.48 lakh crore in 2026Q2, again, one player: a 10bps MDR would generate Rs 1,048 crore a quarter, 30bps Rs 3,144 crore, and 50bps Rs 5,240 crore. The policy choice is therefore not “should payments be cheap” but “who funds a Rs 3,144 crore-a-quarter subsidy, and for how long”.
So what: four monetisation pathways, ranked
| Pathway | Mechanism | Why it can work | Principal risk |
|---|---|---|---|
| Credit on UPI | Route pre-approved credit lines and RuPay credit cards over UPI, where interchange is permitted | Converts a zero-fee rail into a distribution channel for a fee-bearing product; the merchant relationship is already there | Credit risk sits with the lender, not the app; underwriting thin-file borrowers at this volume is unproven |
| PPI / wallet interchange | Wallet-loaded UPI transactions carry permitted interchange | Already regulator-sanctioned; no behaviour change asked of the payer | Small share of transactions; economics depend on load patterns |
| Distribution and cross-sell | Use the payment relationship to distribute insurance, mutual funds, lending | Highest-margin option; the app knows cash-flow behaviour no bank sees | Regulatory perimeter, and the trust cost of monetising a utility |
| Cross-border and merchant SaaS | Inbound remittance corridors, and paid tooling for merchants (settlement, reconciliation, capital) | Fees are acceptable where the payer is not the Indian consumer | Corridor-by-corridor build; slow to scale |
The recommendation. Treat UPI as customer acquisition, not as a product. The defensible model is a merchant-side flywheel: use the 487 touchpoints a quarter to underwrite working capital and sell software, where price is chargeable, rather than lobbying for an MDR that is politically unavailable. Rank credit-on-UPI first because it is the only pathway that scales with the existing transaction base rather than against it.
What would change this answer
- MDR returns for large merchants. A tiered MDR: zero for small merchants, priced above a threshold, would make the merchant leg directly monetisable and reverse the ranking above.
- The 30% market-share cap is enforced. A binding cap redistributes volume rather than growing it, and changes who can afford the subsidy.
- Credit losses on UPI-originated lending. If loss rates on this channel run above unsecured norms, pathway one collapses and distribution moves to first.