
Two lenders can grow farm credit at almost the same speed and be carrying very different risk. That is the reading problem in the Reserve Bank's latest non-bank numbers. Credit from NBFCs to agriculture and allied activities grew 17.8 per cent year on year in September 2026, against 5.2 per cent a year earlier, in the RBI release of 6 November. Banks, in their own September release, grew agriculture credit 19.5 per cent. Put the two headlines side by side and they look like one story. They are not. The channels lend against different collateral, to different borrowers, on different clocks — and September is exactly the month when that difference starts to matter for the rabi book.
Start with the base, because the headline hides it. A year ago, NBFC credit to agriculture was growing at 5.2 per cent — the slowest of the sectors the RBI reports. The move to 17.8 per cent is a rise of 12.6 percentage points in the growth rate, the sharpest swing in the release. Total NBFC credit grew 15.4 per cent, up from 13.3 per cent, so agriculture outran the book as a whole. But a growth rate is a speed, not a size. The release gives no sector amounts in its text, and a channel whose farm book was growing at 5.2 per cent a year ago does not become a major crop lender because one September print tripled. Risk teams that file this under "agri credit is booming everywhere" will misread both channels.
| NBFC credit growth, year on year | Sep 2025 | Sep 2026 |
|---|---|---|
| Agriculture and allied activities | 5.2% | 17.8% |
| Industry | 10.2% | 7.4% |
| Services | 23.0% | 17.3% |
| Retail loans | 16.8% | 21.9% |
| Total NBFC credit | 13.3% | 15.4% |
Source: Reserve Bank of India, Sectoral Deployment of Credit by NBFC – September 2026, Press Release 2026-2027/1461, 6 November 2026 (provisional data). Table: Farmioc.
The largest acceleration in the release is not agriculture at all. Retail loans grew 21.9 per cent, up from 16.8 per cent, and the RBI names the drivers: housing loans, loans against gold jewellery and vehicle loans all sped up. Two of those three sit close to the farm economy without being farm credit. A gold loan in a district town is often a household's bridge through a delayed harvest payment. A vehicle loan may be a tractor-adjacent purchase, a goods carrier for a mandi trader, or neither. None of it appears in the agriculture line. A lender watching only the sector labelled "agriculture" will understate how much NBFC exposure now rides on rural household cash flows — and a gold loan's collateral does not care whether the crop came in on time, which cuts both ways: it protects the lender and tells him nothing about the borrower's season.
The rest of the release points the other way. Industry credit growth fell to 7.4 per cent from 10.2 per cent, with infrastructure lending, in the RBI's words, still subdued. Services moderated to 17.3 per cent from 23.0 per cent, even though commercial real estate stayed buoyant and trade and transport operators borrowed more. So the NBFC book in September was tilting toward households and agriculture and away from industry. For a bank with on-lending lines to NBFCs, that tilt changes the portfolio it is indirectly holding: less project risk, more small-ticket seasonal and household risk, concentrated in exactly the months when kharif realisations and rabi sowing costs collide in the same household budget.
The bank comparison is worth one line and no more. Scheduled commercial banks grew credit to agriculture by 19.5 per cent year on year in September, against 8.9 per cent a year earlier, on data from 41 banks covering about 95 per cent of non-food credit. Both channels accelerated. But bank farm credit is dominated by crop loans and Kisan Credit Card limits set against the scale of finance, disbursed against a sowing calendar. NBFC farm-adjacent credit is dominated by secured retail and allied-activity lending, disbursed when the borrower walks in. In a season when sowing started late, the bank book's risk arrives on the crop calendar; the NBFC book's risk arrives on the household's cash calendar. They will not default in the same month, and they should not be monitored on the same dashboard.
Three limits belong in any credit note that cites this release. The data are provisional, reported as on the last day of the month. The sample covers NBFCs in the Upper and Middle Layers plus housing finance companies, accounting for about 87 per cent of total NBFC credit on the RBI's 2024-25 banking report — large, not complete. And the release reports growth in outstanding credit, not disbursement quality, purpose at the borrower level, or repayment. A 17.8 per cent growth print is consistent with prudent allied-activity lending and with a channel chasing volume into a difficult season. The release cannot distinguish between them. Branch-level and borrower-level data can.
Split the channel view. Track bank crop credit and NBFC farm-adjacent credit as separate exposures with separate early-warning signals — sowing and arrivals for the first, gold-loan and vehicle-loan delinquency in rural districts for the second.
Read the base before the rate. Any sector note quoting the 17.8 per cent should quote the 5.2 per cent it rose from in the same sentence. Growth off a weak base is a direction, not a level.
Ask what the NBFC agriculture line actually holds. Allied activities — dairy, fisheries, warehousing-linked lending — behave differently from crop credit through a late season. The sector label will not tell you the mix; the lender's own book will.
This is the join Farmioc DataStore is built for: credit signals read alongside sowing, arrivals and price data at district level, so a risk team can see which channel's exposure sits in a district where the season is late — before the overdue report, not after it. September's numbers say both channels lent faster into this rabi. The risk was never in the growth rate. It is in what each channel actually lent against.