Reserve Bank of India - Payment System Indicators (Monthly Volume and Value)

It is the Reserve Bank of India's own statistical readout on the payment system it supervises, published monthly as five typed tables and restructured here into records a database can join. Part I carries the headline volume (lakh) and value (Rs. crore) columns - fiscal year to date, same month last year, and the two most recent months - across three tiers: securities and FX markets cleared through CCIL-operated FMIs, RTGS at the wholesale-customer boundary, and the full retail stack underneath it.

That retail stack is where India's payments story lives: AePS fund transfers, APBS, IMPS, NACH credit and debit, NEFT, UPI including its USSD channel, BHIM Aadhaar Pay, NETC toll collections, card payments split between PoS and other acceptance, prepaid payment instruments, and CTS paper instruments as the legacy baseline. The June 2026 edition puts UPI alone at 2,27,160.66 lakh transactions worth Rs. 28,92,139 crore in the month, inside 2,61,705.81 lakh total digital payments worth Rs. 3,17,64,092 crore.

The remaining four parts make this more than a volume sheet. Part II re-cuts payments by mode and channel, Part III stocks the installed infrastructure at month end, Part IV follows India-issued cards and PPIs abroad, and Part V publishes the domestic fraud series with its ratio expressed in basis points.

One rail first, then the structure around it - exactly as a delivery lands:

# Part I - June 2026 edition - volume in lakh, value in Rs. crore
row=2.6 UPI   vol_FY2025_26=2416168.86   vol_June2026=227160.66
              val_FY2025_26=31423250     val_June2026=2892139

row=Total Digital Payments (1+2+3+4+5)
              vol_FY2025_26=2817448.36   vol_June2026=261705.81
              val_FY2025_26=325103565    val_June2026=31764092

Volume and value ride on the same row, which is the whole analytical trick: divide them once and you have UPI's implied average ticket size, month over month, without stitching two tables together. The fiscal-year-to-date column sits beside the monthly one, so a year-to-date reconciliation is a subtraction, not a rebuild.

The stock side lands just as cleanly:

# Part III - June 2026 edition - stock counts in lakh, as on month end
row=7 UPI QR                              May2026=7797.65   June2026=7925.74
row=1 Number of Cards (Credit + Debit)    June2026=11534.57

# Part V - domestic payment frauds, June 2026
frauds:   volume_lakh=3.53   value_crore=489   ratio_bps=0.150

A single month moves UPI QR acceptance by roughly 128 lakh codes while total cards sit above 11,500 lakh - acceptance infrastructure and instrument supply moving on separate clocks, in one schema.

Six field roles cover all five parts. The payment-system label keys each row to a settlement system, retail mode, channel or infrastructure item - '2.6 UPI', 'Number of PoS Terminals' - preserving the part's own numbering so cross-edition joins stay stable. Volume (lakh) and value (Rs. crore) are the paired measures in Part I and Part II; the lakh-crore convention matters because average ticket analysis needs both unrounded.

Fiscal-year-to-date columns accumulate from April, and the monthly columns run same-month-last-year plus the two most recent months, giving every row its own YoY comparison out of the box. Stock-count columns ('As on March 2026'-style headers) drive Part III, where the unit is outstanding items in lakh rather than period activity.

Every definition was verified against parsed editions during research, backed by published sample rows. Two structural notes ship with any sample because they change what you can claim: the ECS-to-NACH merger effective January 31, 2020, and revised card/PPI definitions from November 2019 onward that make those series incomparable with earlier periods. Deeper columns confirmed during sample preparation fold out under additional fields on request.

Geography: India, scoped to domestic financial transactions - failed transactions, chargebacks, reversals and expired cards are excluded by definition, so the numbers measure completed commerce rather than attempted attempts. Part IV steps outside the border to track international transactions made using India-issued cards and PPIs.

Temporal: the archive reaches back to 1990, though the series is not uniform across that span. The current five-part format with granular retail-mode rows dates from November 2019, and the card and PPI series from that point onward are explicitly not comparable with earlier periods due to revised definitions. The domestic fraud series in Part V runs from September 2022. Figures in recent editions are marked provisional until finalized.

Granularity: monthly totals per payment system or mode, with fiscal-year-to-date aggregates alongside, and month-end stock counts for everything installed - terminals, QR codes, cards, ATMs, CRMs. Roughly 100 data rows per edition means the entire national payments picture fits in memory, not a cluster.

API, files, or your warehouse. Daily, weekly, or hourly.

Pick the channel your team already works in: structured payloads for live lookups on a specific rail or indicator, flat files sized for overnight warehouse loads, or a direct pipe into Snowflake, BigQuery or Redshift. Cadence is yours to set - and to change when your models change.

Every delivery ships the full field dictionary, sample rows for validation, and a schema that holds steady between deliveries.

The source releases as formatted spreadsheet and document tables, roughly 490 KB per edition across about 100 rows in five parts. Turning that layout into the six-role schema above - typed measures, split periods, preserved part numbering, footnote caveats attached to the series they govern - is our pipeline's work, not yours.

• Investors and quant researchers treat the UPI-versus-card split as a public read on Indian digital-payments adoption, watching volume growth in lakh against value growth in crore to see ticket sizes fall as acceptance spreads.
• Payments strategists and market sizers build rail-by-rail India models from one internally consistent official table instead of reconciling press releases, presentations and third-party estimates that never quite agree.
• Fraud and risk analysts benchmark their own book against Part V's national figures - the 'one in every X transactions fraudulent' rate and the fraud-to-payment-value ratio in basis points - rather than against anecdotes.
• Competitive intelligence teams track acquiring-side infrastructure directly: PoS terminal installs, micro-ATM rollout, Bharat QR versus UPI QR counts as month-end stocks.
• Market researchers and consultants ground fintech market maps and India-entry theses in official rail-level volumes and channel splits.

More applications by team live on the investors quants use cases and market researchers use cases briefings.

Investors and quant researchers get the cleanest public panel on India's payment rails, refreshed on their cadence rather than the release calendar; see investors quants use cases. Payments strategists and market sizers get one consistent denominator across every rail they compare. Fraud and risk analysts get a national baseline with the ratio pre-computed in basis points. Competitive intelligence and product teams get the acceptance-infrastructure story - terminals, QR codes, micro-ATMs - as stock counts rather than marketing claims. Data scientists and ML engineers get a long, documented monthly panel whose definitional break points are stated up front instead of discovered mid-backtest; see data scientists use cases. Market researchers and consultants get the official anchor every India fintech deck should cite; see market researchers use cases.

Three things, stated up front.

First, the archive is deep but not uniform. Monthly editions exist back to 1990, yet the granular five-part structure only starts in November 2019, card and PPI definitions changed at that point enough to break comparability with earlier years, and Part V's fraud series begins in September 2022. Tell us how far back you actually need and we will shape the sample to the comparable window.

Second, the units are Indian conventions, kept deliberately. Volume arrives in lakh and value in Rs. crore, exactly as published - converting to millions and billions is a one-line transform we would rather leave to you than round-trip on your behalf and risk silent scale errors.

Third, recent months are provisional. The current edition's figures are flagged provisional in the source footnotes, which matters if you are building a nowcasting model: the latest point can move. Name the rails, indicators and months you care about and the sample comes back shaped to them, caveats attached to the series they govern.