India runs two payments markets: one that spends, one that sends: India FS Pulse

Generated 2026-08-22 by analysis/05_geo_gap.py. Every figure below is interpolated from a computed value, so the prose cannot drift from the data.

The answer

Digital payments in India are not one market. In 2026Q2, the ten largest states carried 79.6% of all transactions, but the more useful split is not size. It is what kind of transaction a state runs. Measuring each state’s merchant share of its own transactions separates retail economies from remittance economies, and the two carry different economics and need different strategies. Across the material states the merchant share ranges over 12 percentage points.

Two supporting arguments

1. Merchant-heavy states are the urban retail economies. The highest merchant shares sit in Delhi (68.5% of its own transactions), Haryana (67.9% of its own transactions), Karnataka (67.5% of its own transactions). These are dense, high-income, high-merchant-density states where the everyday retail leg dominates: the leg that generates processing cost and, at zero MDR, no revenue.

2. P2P-heavy states are the remittance economies. At the other end sit West Bengal (56.4% of its own transactions), Bihar (56.6% of its own transactions), Assam (57.3% of its own transactions). A larger share of person-to-person transfers is the signature of money being sent rather than spent: labour-exporting states receiving inbound remittance flows.

So what

Merchant acquisition and lending should follow the merchant-heavy states. That is where merchant density and repeat transaction frequency support underwriting.
The zero-MDR burden is geographically concentrated. Processing cost sits disproportionately where the merchant share is highest, which is where a tiered MDR would land hardest and where a small-merchant exemption would matter most.
Do not read this as adoption. It is a mix measure, not a penetration measure.

Method and its limits

Computed from PhonePe Pulse per-state category files for 2026Q2, cross-checked against the separately-fetched country file: the 36 state files reconcile to the national totals at a ratio of 1.000000 on both transaction count and value, and imply a national merchant share of 63.9%.

Three limits, stated plainly. This is one operator’s mix, not the market’s, though a within-state mix ratio is far less sensitive to PhonePe’s uneven regional footprint than a between-state volume ratio would be, which is the main reason this measure is preferred. There is no population denominator in the open data used here, so this is deliberately a composition measure, not a per-capita one. And states below 1% of national volume are excluded when naming extremes, so a small union territory cannot top the ranking on a thin base.

What would change this answer: a second operator’s state-level category split. If Google Pay’s mix inverted this ranking, the finding would be about PhonePe’s distribution rather than about India. Nothing open publishes that today.

A note on the previous version

This module previously ranked states by transaction share divided by value share, calling it an “intensity index”. That quantity is identical to national average ticket divided by state average ticket. It restated ticket size and nothing else. Measured against the real merchant share it correlates at -0.05, i.e. not at all. The old index ranked Assam most merchant-intense; Assam is in fact among the most P2P-heavy states here. The exhibit was replaced rather than relabelled.