← India FS Pulse

Memo

The only comparable rail repeats India's shape and has not monetised the transaction either

Generated 2026-11-05 by analysis/09_pix_comparator.py. Every figure below is interpolated from a computed value, so the prose cannot drift from the data.

The answer

The one rail comparable to UPI does the same thing, and after six years it still does not earn from the transaction. In 2026-10, Brazil’s merchant leg was 45.3% of all Pix transactions and 11.3% of the value. India’s merchant leg in 2026Q2 was 63.9% of transactions and 23.0% of value. The shape repeats. Two central banks, two regulators, one design, and in both cases the busy leg is the cheap leg.

That matters because it converts this report’s central claim from an interpretation into an observation. The split is not a consequence of India’s zero-MDR statute, because Brazil never had an MDR to remove and arrived at the same place.

Three supporting arguments

1. Brazil’s merchant leg migrated exactly as India’s did, from a standing start. P2B was 5.2% of Pix in 2020-11, the rail’s first month, and 45.3% in 2026-10. Nothing redirected it: the merchant leg is simply where instant payments go once the habit exists. India’s own growth bridge says the same thing from the other end, that the merchant leg produced most of the volume growth. Two independent series, one behaviour.

2. The layers built on top of the free rail are real and still immaterial. Brazil did what the “just add products” answer prescribes: recurring debits, open-finance payment initiation, and contactless. Together they account for 0.36% of merchant transactions, still below the 1% line, which is to say not yet a business. It is also compounding fast: 16 times its 2025-10 share in twelve months. Small and accelerating is a different statement from small and static, and only the first of those is an option on something. Meanwhile the merchant leg consolidated into dynamic QR, now 83.6% of it, which is a distribution channel rather than a revenue line: it carries an amount and a payload into a merchant’s own system, and it is free.

3. Brazil is further along the same curve, so this is not a maturity gap India will grow out of. On 2026-07, the latest month both countries publish, Brazil ran 49 instant payments per banked adult against India’s 24, so Brazil leads by 2.0 times. Both use the same denominator: total population less ages 0 to 14, times account ownership (86% for Brazil and 89% for India, Findex 2024 and 2024). The more mature rail is the one with the smaller merchant value share. Volume does not convert into transaction revenue by getting bigger.

So what

Method and its limits

Brazil’s central bank publishes the transaction cross-tab as keyless open data, labelled by payer and receiver type, so the merchant leg is read from the publisher’s own P2B label rather than derived. No currency is converted anywhere in this module: every figure is a share of its own country’s total, which is why no exchange rate appears and none is assumed. The intensity comparison is the one place a denominator is used, and both sides use the same one.

What this cannot support: it is one comparator, not a survey of instant rails. India’s merchant split is PhonePe’s book while Brazil’s is national, so the two are the same measure on different populations, and the levels should be read as approximate even though the direction is not in doubt. Brazil’s series carries a small number of early rows with no initiation method recorded, kept as a separate bucket rather than dropped, because dropping a grouping key removes its transactions from the denominator and quietly inflates every other share. The fraud table stops at 2026-09 while the transaction table runs to 2026-10, so the two are not read against each other here.