India's microfinance book shrinks (−11% y/y) as early arrears fall and old defaults pile up

Implications for lenders

India's lender cap of three lenders per borrower and the RBI's 50% repayment limit mean underwriting has to start from a bureau check across all lenders and an estimate of household income. Lenders in markets without such a limit can apply a similar internal check, which may reduce new arrears but also the number of loans. Collections teams may see legacy loans more than six months overdue, at 17.04% in India, stay on the books well after new lending improves.

Implications

Facts

Size: the 2026 edition of the Bharat Microfinance Report, from the industry body Sa-Dhan with NABARD, uses Equifax credit bureau data to put microfinance loans outstanding at INR 3,33,907 crore at end-March 2026, down 11% from March 2025. ¹ Loan accounts numbered 104 million (1,040 lakh), down 21% from March 2025. ² The report says growth turned positive in the last quarter, at +4% over December 2025. ³ The average loan disbursed in the year to March 2026 was INR 59,473 (up 17% from INR 50,918 in FY2024-25). ⁴

By lender: banks' microfinance loans changed −29% (year to end-March 2026; banks' share of the market 26%), while NBFC-MFIs, the specialised non-bank microlenders, changed −1% (year to end-March 2026; NBFC-MFIs' share of the market 42%). ⁵

Arrears: the share of loans one to six months overdue (portfolio at risk) fell to 2.34%, down from 6.63% in March 2025. ⁶ The share more than six months overdue rose to 17.04%, up from 10.67% in March 2025; the report notes that credit information company data include technical write-offs. ⁷ Data reported by institutions to the self-regulatory organisations give a different measure: a share of loans more than a month overdue, including those more than six months overdue, of 5.75%, down from 7.52% a year earlier. ⁸

Rules: under the RBI's Regulatory Framework for Microfinance Loans (2022), a microfinance loan is a collateral-free loan to a household with annual income up to INR 3,00,000 a year, and a low-income household's monthly loan repayments may not exceed 50% of its monthly income, counting microfinance and other loans. ⁹ The report attributes the fall in loan accounts to lender caps set by the self-regulatory organisations, a maximum of three lenders per borrower (down from four; in force since 1 April 2025), and to tighter credit bureau checks. ¹⁰

Analysis

The Indian numbers show what a debt cap does to a microfinance market in its first year: fewer loans, larger tickets and a sharp fall in new arrears. With loan accounts down 21% from March 2025 and the average ticket up, lenders kept fewer, larger borrowers who passed the bureau check rather than spreading small loans across many lenders. The rise in loans more than six months overdue to 17.04% is the other side of the same correction: loans that went bad during the stress period stay in the bureau data, even after write-off, while the new book performs better. Banks retreated most, at −29%, while NBFC-MFIs held their book roughly flat, a reminder that the specialised lenders, not banks, carry the segment through a downturn. For comparison, the gross NPA ratio of Indian banks as a whole was 1.8% in March 2026.

Cambodia faces the same question with different tools. Its microfinance portfolio was still growing in 2025, reaching USD 5.99bn (up more than 15% y/y) across 1.53 million borrowers at CMA member institutions, and the NPL ratio of non-deposit-taking institutions, mostly MFIs, was 13.5% in 2025. The National Bank of Cambodia's published list of regulations includes an interest rate ceiling on loans but no Prakas setting a debt-to-income limit; over-indebtedness is being addressed through a joint NBC–UN plan of 22 actions. India combines a regulator's affordability rule, income threshold and repayment cap, with an industry cap on the number of lenders, both enforced through credit bureau checks. Definitions differ (portfolio at risk from bureau data in India, NPL ratios from the supervisor in Cambodia), so the comparison is of tools and direction, not of levels.

What it suggests for Southeast Asia: a cap on lenders per borrower or on repayments relative to income works only if every lender reports to and checks the credit bureau, and India's data show the trade-off: credit access shrank in the first year, with fewer accounts, while new arrears fell. For Cambodia, and for Philippine and Indonesian lenders serving the same low-income households, India's first-year result is a reference for setting a limit and for planning the phase-in, including how long legacy defaults stay visible in bureau data.

Counterpoints and uncertainties

The data mix sources: bureau data in the report's sector chapter include technical write-offs, while self-regulatory data cover only reporting institutions and give a single ratio of loans more than a month overdue, including those over six months (5.75%), far below the bureau buckets combined, so the scale of the improvement depends on the data set. The quarterly rise of +4% is one quarter and may not mark a lasting recovery; funding for smaller microlenders and weather risk to rural incomes remain open. Cambodia's market differs in lender mix, currency and supervisory set-up, so India's caps would not transfer unchanged.