← India FS Pulse

Memo

Zero MDR is not zero cost: the state pays about seven basis points, and it is falling

Generated 2026-11-05 by analysis/08_rail_cost.py. Every figure below is interpolated from a computed value, so the prose cannot drift from the data.

The answer

Zero MDR is not zero cost. It is 7.0 basis points, paid by the taxpayer, and it is falling. In FY2023-24 the government paid acquiring banks Rs 3,631 crore in incentive against Rs 51.5 lakh crore of national merchant volume. Set beside the 30 basis points NPCI permits on merchant UPI and that statute has held at zero since January 2020, the state is funding roughly 23.5% of the discount rate the market is not allowed to charge.

The appropriation is now falling in absolute terms, not only per rupee. FY2024-25 was approved at Rs 1,500 crore, 58.7% below the Rs 3,631 crore actually paid in FY2023-24, against a merchant base that was still growing. No rate is computed for that year here: the outlay covers twelve months and the published value base covers 10, and a ratio across mismatched periods is not a rate.

Three supporting arguments

1. The subsidy per rupee has fallen in every year measured. The blended rate went from 8.68bps in FY2021-22 to 7.05bps in FY2023-24, a 18.8% decline. Nobody legislated that. The payout grew 2.6 times while the merchant base it covers grew 3.2 times, and the distance between those two multiples is the whole mechanism: the appropriation is set in rupees and the base compounds in percent.

2. The statutory rate and the effective rate are different numbers, and only one of them is money. The scheme pays 15.00bps, but only on merchant transactions at or under Rs 2,000 to small merchants. Blended across all merchant value the state pays 7.05bps, which implies that at least 47.0% of national merchant value sits inside that band. That is a floor rather than an estimate: 20% of every claim is withheld unless the acquiring bank holds technical declines below 0.75% and uptime above 99.5%, so any entitlement left unclaimed pushes the true eligible share up rather than down.

3. It reframes the question from “why is the rail free” to “for how long”. The merchant leg carries 63.9% of transactions and 23.0% of value at an average ticket of Rs 425. Those economics are not zero-revenue. They are revenue of last resort, appropriated one year at a time, at an effective rate nobody has committed to holding. A business model resting on it is underwriting a budget line, not a price.

So what

Method and its limits

Both tables come from one PIB release, transcribed in a browser because PIB serves HTTP 403 to scripted requests and publishes these figures as chart images with printed data labels rather than as text. The blended rate divides incentive actually paid by all national merchant value, so both sides are measured and neither is modelled. The national merchant share it implies, 25.8% of UPI value, is cross-checked against PhonePe Pulse’s own merchant share of value, 23.0%: a gap of 2.7 percentage points. Announcing a cross-check without printing its result would be worth nothing, so it is printed. The two are not the same quantity and are not expected to match: PIB measures the whole country, Pulse measures one operator’s book. The rate above is computed on the national figure, because the incentive is paid against national merchant value and not against PhonePe’s.

What is not known: how merchant value splits above and below the Rs 2,000 ceiling, what share of it belongs to small merchants as the scheme defines them, and how much entitlement went unclaimed on the performance conditions. Those three unknowns are precisely why the eligible share above is published as a floor and not as a figure. The series ends at FY2024-25 and is not extended by interpolation.