The platform's credit model was built for Indian agriculture as it usually is — not as it periodically becomes. In a normal monsoon year, NPA sat at a comfortable 3.8%, within RBI NBFC norms and reassuring to investors. IUSP Partners asked one question that changed the engagement entirely: "What was your NPA in the 2023 Marathwada drought?" After thirty minutes of searching, the answer came back: 17.4%. There was no provisioning plan for that number. There had never been one.
Agricultural credit is not like SME lending, personal loans, or home finance. In those categories, the risk factors are primarily individual: the borrower's income, credit history, debt-to-income ratio. These factors are relatively stable across a portfolio and across time. Stress is idiosyncratic — one borrower fails, not ten thousand simultaneously.
Agricultural credit is exposed to systemic, correlated, geographically clustered risk. When a drought hits Marathwada, it does not affect one farmer in the portfolio — it affects every farmer who grew rain-fed crops in that district that season. The correlation across borrowers in a geographic concentration is close to 1.0 in a severe weather event. This is the opposite of a diversified portfolio.
"A lending model built on normal-year NPA will fail in any stress year — and stress years in Indian agriculture are not exceptional events. They are regular, recurring, and in some regions, becoming more frequent with shifting monsoon patterns."
The platform had 68% of its portfolio in Maharashtra and Karnataka, concentrated in rain-fed Kharif crops (soybean, cotton, jowar). Three consecutive below-normal monsoon years — 2021, 2022, 2023 — had increased the frequency of crop failure. Yet the credit model, provisioning policy, and investor disclosures were all built around the 3.8% NPA benchmark from a normal year.
When IUSP Partners asked to see the stress scenario analysis in the credit risk framework, there was none. The credit policy defined NPA thresholds and recovery procedures. It did not define what would happen to the portfolio if 40% of Marathwada had below-80% rainfall in a single Kharif season — which is precisely what occurred in 2023.
| Scenario | Rainfall (% of Normal) | Crop Failure Rate | Portfolio NPA Impact | Required Provisioning |
|---|---|---|---|---|
| Normal Year | >90% | 3–5% | 3.8% | ₹11.4 Crore (3.8%) |
| Stress Year | 75–90% | 15–22% | 8–10% | ₹24–30 Crore |
| Crisis Year (2023 actual) | <75% | 38–45% | 17.4% | ₹52 Crore |
| Actual provisioning held (2023) | — | ₹9 Crore (3% buffer) | ||
The shortfall: ₹43 Crore in provisioning that should have existed but did not. The company avoided catastrophe in 2023 because of a ₹40 Crore emergency equity infusion from its primary investor — raised under duress, at unfavourable terms, in the middle of a portfolio crisis.
Agricultural credit risk has three dimensions that do not appear in conventional NBFC credit models:
Geography: Risk is not uniform across the portfolio. A borrower in irrigated Nashik growing grapes has fundamentally different risk from a borrower in rain-fed Osmanabad growing soybean. Blending them into a single NPA number hides the geographic concentration.
Season: Kharif (June–October, monsoon-dependent) and Rabi (November–March, irrigated or winter rain) have structurally different risk profiles. A portfolio weighted toward Kharif in rain-fed districts has drought exposure that a Rabi-weighted portfolio does not.
Irrigation source: Canal-irrigated, bore-well dependent, and rain-fed agriculture have NPA volatility profiles separated by orders of magnitude. This single variable predicts crisis-year default better than almost any borrower-level credit factor.
The 2023 Marathwada drought was the worst in six years. Rainfall in Osmanabad and Latur districts was 61% of normal — a crisis-level event. The new early warning system triggered in the first week of August, 10 weeks before the harvest season. By the time the crop failures became visible in October, the platform had already:
In Indian agriculture, stress is not a tail risk — it is a regular event. The question is not whether a drought or flood year will come. It is whether the credit model, provisioning policy, and capital structure are designed to absorb it when it does. A 3.8% NPA in a normal year is a metric. A plan for what happens in a 17% NPA year is a business.
If your agritech or NBFC credit model is built on normal-year data, you may be one bad monsoon away from a capital crisis. Let's stress-test it now.
Contact Us →