MSME Credit Growth in India 2026: Small and Medium Up 20%, Micro Flat

Funded credit to MSMEs with exposure up to ₹100 crore stood at ₹47.4 lakh crore in June 2026, up 12.5% year on year, per CRIF High Mark's MSMEx Spotlight (September 2026). NBFCs lagged by value: their MSME book grew 4.2%, against 13.5% at public sector banks and 13.2% at private banks, though NBFC active loans rose 5.7% while public sector bank active loans fell 9.3%. FlexiLoans, an NBFC, projects a ₹2.4 to ₹2.6 lakh crore MSME lending opportunity for July to December 2026.

MSME credit growth by segment, June 2026

The headline numbers are not the interesting part. This is:

These are bands of credit exposure, not the legal MSME classes, which run on investment and turnover. For a lender they are the more useful cut, because they describe the size of the loan relationship. They also move: a borrower whose borrowing crosses ₹2 crore leaves the micro band for the small one, so some of the small band's growth can be graduation out of micro rather than new lending.

Micro accounts for 84.4% of active MSME loans, and active micro loans fell 1.0% over the same year. The segment holding the overwhelming majority of loan accounts is the one that did not grow.

By sector the same pattern repeats: manufacturing credit grew 17.4%, trading 12.7%, and services 0.2%.

Other datasets frame the same market differently. CRIF counts fund based loans to entities on the commercial bureau. TransUnion CIBIL and SIDBI find that business purpose loans taken by individuals in their own name made up 28% of ₹65.8 lakh crore of commercial credit in March 2026, growing at a 22% three year CAGR against 14% for entities. RBI sectoral data for August 2026 shows bank credit to micro and small industries still expanding, while credit to industry as a whole grew 18.2% year on year. Flat micro exposure in CRIF data is not the same thing as flat lending to micro enterprises.

The two CRIF splits line up with how easily a borrower can be assessed. That is a pattern, not proof of cause, and the timing points to more than one driver.

The growth is where conventional assessment works

Consider what a lender actually has to work with at each end of this distribution.

A borrower with ₹2 to ₹100 crore of credit exposure has audited financial statements, almost certainly a GST registration with a filing history, a credit bureau record with reported tradelines, a chartered accountant, formal banking relationships with large banks, and usually a constitution, private limited or partnership, that produces documentation as a by-product of existing.

Every conventional underwriting instrument functions. Financials can be spread. The bureau can be pulled. GST can be checked. The assessment is work, but it is known work.

A micro enterprise with under ₹2 crore of exposure frequently has none of that. No audited financials. Often no ITR, or a presumptive scheme return that says little about actual income. A thin or absent bureau file. Cash turnover forming a material share of activity. A proprietorship constitution where the personal and business boundary does not exist: the same account pays school fees and buys inventory. And banking that may sit with a cooperative, small finance or regional rural bank whose statement format gets deprioritised by analysis tooling.

Services businesses compound it. No inventory to verify, no fixed assets to value, no stock statement to call for. Cash flow is the entire assessment. That fits the weak services figure, though CRIF's June 2026 edition shows services credit growing 10.7% in the December 2024 quarter alone, so the 2026 stall is not purely structural.

The pattern is consistent with one reading: credit flowed to the borrowers whose financial position could be established with existing tools, and stalled where it could not.

Why micro credit stalled: assessment, risk and timing

Start with timing. CRIF data shows micro exposure credit grew 13.0% in the year to September 2025. Then it turned: micro portfolio outstanding fell 7.4% in the September 2025 quarter, and 3.1% between December 2025 and March 2026 against 3.4% growth a year earlier, a slowdown CRIF's June 2026 edition reads as the potential impact of global uncertainty on domestic MSME credit supply. Assessment difficulty did not appear in a few quarters. What it plausibly decides is which borrowers lenders pull back from first when conditions tighten: the files they cannot defend.

Then test the other explanations, because the conclusion matters.

It is not capital. Lenders are deploying: the small and medium segments grew over 20%, and bank priority sector targets, including a 7.5% sub-target for micro enterprises, push toward MSME regardless.

It is not policy. Formalisation through GST, Udyam registration and the Account Aggregator framework has been pushing the other way for years.

It is not demand. TransUnion CIBIL and SIDBI count 8.7 crore registered MSMEs as of December 2025, of which only 3.6 crore had ever accessed formal credit by March 2026.

What assessment difficulty does bind is the credit note. A lender cannot write a defensible one on a borrower whose income cannot be established, whose obligations do not appear on a bureau, and whose statement may not parse. Faced with that, a rational credit team declines, or underwrites on collateral and guarantee rather than on the business, which caps ticket size and makes the economics marginal.

The alternative that carries the most weight is risk appetite, and it deserves a straight answer. CRIF's September 2026 edition puts early stage delinquency (31 to 90 days past due) in micro at 2.6% in June 2026, against 1.2% in small and 0.6% in medium. All three improved year on year, but micro still runs at more than twice the small rate, and the June 2026 edition noted a more cautious stance among some lender groups. That is real. It is also partly the same constraint seen from the other side: a lender that cannot tell a sound micro borrower from a weak one has to price and limit the whole segment as if it were the weak one. Better assessment does not lower the segment's delinquency; it lets a lender stop lending blind into it.

CRIF itself describes micro, at 84.4% of active loans, as "a massive segment ripe for further credit formalization".

What micro enterprise underwriting needs from the bank statement

To the extent assessment capability is a constraint, it is worth naming precisely what has to exist. Each of these is a concrete gap rather than a general aspiration.

Income derived rather than verified. There is no salary credit and no audited turnover figure. Income has to be built from the account: inflow regularity, source persistence, and net business receipts after removing self-transfers, loan disbursals, related-party credits and refunds. That is a structurally different task from verifying a payslip, and tools built for salaried retail do not do it.

Obligation discovery beyond the bureau. Micro borrowers service chit fund contributions, gold loans from pawnbrokers and local societies, moneylenders and family borrowing, none of which reaches a bureau, alongside loans from regulated lenders that do. The unreported part appears only as recurring debits to stable counterparties in the account, and a capacity calculation that misses them is describing a borrower who does not exist. Finding them is obligation mapping done from the debit side of the account.

The personal and business boundary, held open rather than collapsed. School fees classified as drawings and inventory purchase as business expenditure, in a single account, with both recognised as expected. A retail model reads the inventory purchase as unexplained large spend; a corporate model reads the school fees as an anomaly. Both get the borrower wrong.

Conduct proxies in place of repayment history. Eighteen months of honoured standing instructions, punctuality on rent and fixed commitments, balance discipline around known debit dates. This is the substitute for a bureau repayment record, and it is available in the account.

Format coverage at the long tail. A statement from a cooperative or regional rural bank that does not parse routes to manual review, where the figures get keyed in at summary level and the fraud controls do not run. A partial parse is worse than a failure, because it produces a complete-looking analysis on incomplete data.

Accumulation and seasonality reading. Micro and small borrowers are frequently seasonal, and a six-month window landing inside a peak or a trough produces a materially wrong answer in either direction. Twelve months is the floor.

None of these is exotic. All of them are specific, and a tool either does them or does not.

MSME cluster concentration risk

One further finding in the CRIF data deserves attention from anyone growing an MSME book, because it is a portfolio risk hiding inside a growth story.

The number of districts carrying more than ₹1,000 crore of engineering and machinery credit nearly doubled, from 24 in June 2023 to 47 in June 2026, and those districts now hold 63.5% of that industry's credit, up from 50.9%. High volume wholesale trade clusters more than doubled, from 54 to 119, and hold 71.7% of wholesale credit. In chemicals, such districts rose from 17 to 24 and their share from 47.2% to 53.1%, and CRIF notes a slight uptick in delinquency in them, to 1.5% PAR, that calls for cluster level risk monitoring. Textiles moved slightly the other way at the cluster level, from 18 clusters holding 62.8% to 15 holding 60.8%, while Surat alone rose from 15.6% to 19.4% of the industry's credit. Part of the rise in cluster counts reflects a fixed ₹1,000 crore threshold applied to a market that grew by a third in two years.

The top of each list is not always getting more dominant (the five largest engineering districts slipped from 20.2% to 18.7% of that credit), but the overall shape is clear: credit is consolidating into more large industry hubs. For a lender, that means borrowers in the same cluster are not independent exposures. They frequently share buyers, share suppliers, and depend on the same downstream demand.

Twenty borrowers in a textile cluster, each individually well-diversified by their own concentration metric, may all sit in a chain feeding a small number of large buyers. One of those buyers stretching payment terms by 45 days is a simultaneous liquidity event across twenty exposures. No bureau shows it and no single-borrower analysis shows it, because each file genuinely is diversified relative to its own receipts. It appears only when counterparties are resolved and aggregated across the book.

Cluster concentration is visible in a geography report. Shared-counterparty concentration is not, and it is the one that transmits stress. TransUnion CIBIL and SIDBI make the same point, calling it imperative to build early warning signals specific to cluster shocks.

Growth creates a quality problem before it creates a volume problem

A second-order consequence worth planning for.

If applications grow 20% and credit analyst headcount grows 5%, something gives. What gives first is depth per file, and the section that thins first is the banking section, because it is the most labour-intensive part of the assessment. Reclassifying transactions, consolidating accounts by hand, reconciling totals before any judgement can begin: on a multi-account business case this commonly runs 30 to 45 minutes before the analysis starts.

The result is a book growing on assessments that got progressively shallower, which is invisible until the vintage curves separate two to three quarters later. Cohort signals read from banking, such as throughput deterioration against the origination baseline, obligation accretion and utilisation drift, surface it at month two instead.

The assembly burden is also where the quality of a credit note is actually determined. Banking sections are thin not because analysts think them unimportant but because by the time the figures are trustworthy there is no time left to interpret them.

What this argument does not claim

Assessment capability is necessary for micro lending to grow. It is not sufficient, and pretending otherwise would be selling.

Unit economics bind independently. A ₹6 lakh loan cannot carry the acquisition, assessment and collection cost structure of a ₹60 lakh one, and no analytical improvement changes that arithmetic by itself.

Collection and recovery economics bind. Servicing a dispersed micro book in Tier 3 geographies is a distribution problem as much as a credit one.

Cash remains invisible, and this is one of the most cash-intensive borrower segments in Indian lending. A trader with substantial cash turnover has banking describing a fraction of their activity. Analysis will systematically understate them, and the correct response is to size the assessment to a partial view rather than to treat the visible portion as the whole.

The growth has not been tested through a full credit cycle. The slowdown from late 2025 into 2026 was a first test, and micro is where it showed. Thresholds and models calibrated on recent MSME performance have not been through a downturn, and a segment growing at 20% a year accumulates vintages nobody has stress-tested.

What better assessment does is remove one specific binding constraint: the one that makes a defensible credit note impossible rather than merely expensive.

Where Fiscus fits

Against the capability list above: income derived from inflow regularity and source persistence where no salary credit exists; obligations discovered from recurring debits including lenders absent from the bureau; drawings separated from business expenditure in proprietorship accounts; conduct proxies read from standing instructions and fixed commitments; scanned, native and screenshot statements across banks and formats with documents flagged rather than partially parsed; twelve-month seasonality and accumulation behaviour read month on month; counterparties resolved so cluster and shared-buyer exposure can be aggregated across a book; and bureau cross checks on the banking, with GST turnover benchmarked alongside where the borrower is registered.

The measurable part is the assembly burden, which is what determines whether depth per file survives growth. Fiscus categorises transactions at 95%+ accuracy, validated over 200,000 transactions in joint reviews with partner lenders, and we published the head-to-head benchmark and its protocol so it can be tested rather than taken on trust.

Frequently asked questions

How fast is MSME credit growing in India? Funded MSME credit to borrowers with exposure up to ₹100 crore was ₹47.4 lakh crore in June 2026, up 12.5% year on year, per CRIF High Mark. Small and medium exposures grew 20.3% and 21.3% while micro (up to ₹2 crore) slipped 0.5%. By lender, public sector banks grew 13.5%, private banks 13.2% and NBFCs 4.2%.

Why did micro MSME credit stall in 2026? In CRIF data, micro exposure credit fell 0.5% in the year to June 2026, after growing 13% in the year to September 2025. The decline came in quarterly falls from mid 2025, which CRIF's June 2026 edition links to global uncertainty, alongside higher early delinquency in micro. Thin documentation makes micro files the hardest to defend when lenders turn cautious.

How do CRIF exposure bands differ from MSMED Act classes? CRIF bands borrowers by total credit exposure: micro up to ₹2 crore, small ₹2 to 20 crore, medium ₹20 to 100 crore. The MSMED Act classifies enterprises by investment and turnover: since April 2025, micro means investment up to ₹2.5 crore and turnover up to ₹10 crore. A legally micro firm can sit in the CRIF small band.

Which MSME segments are growing fastest? By CRIF credit exposure band, medium (₹20 to 100 crore) at 21.3% and small (₹2 to 20 crore) at 20.3%, in the year to June 2026. By sector, manufacturing at 17.4% and trading at 12.7%, against services at 0.2%. The split lines up with how readily a borrower's position can be documented, though the timing points to other drivers too.

How much MSME credit do individuals borrow in their own name? A large and growing share. TransUnion CIBIL and SIDBI put business purpose loans taken by individuals at 28% of ₹65.8 lakh crore of commercial credit in March 2026, growing at a 22% three year CAGR against 14% for entities. Assessing a proprietor therefore means reading both the consumer and the commercial bureau.

What is the priority sector target for micro enterprises? Domestic commercial banks and foreign banks with 20 or more branches must direct 40% of adjusted net bank credit, or off balance sheet credit equivalent if higher, to the priority sector, with a 7.5% micro enterprise sub-target. Small finance banks, RRBs and urban co-operative banks share it; foreign banks with under 20 branches do not. NBFCs have no targets.

What data is used to underwrite a micro enterprise? Principally the bank account, since it is the only continuous record produced by a third party rather than asserted by the borrower. Income is derived from inflow regularity and source persistence, obligations from recurring debits including unreported lenders, and conduct from standing instruction success and punctuality on fixed commitments. GST filings provide an independent cross-check where the borrower is registered.

What is cluster concentration risk in MSME lending? MSME credit is concentrating in industry hubs. Per CRIF, 47 districts with over ₹1,000 crore of engineering credit held 63.5% of MSME credit to that industry in June 2026, and Surat held 19.4% of textile credit. Borrowers in one cluster often share buyers and suppliers, so they behave as correlated exposures even when each file looks diversified.

Does better underwriting technology solve the MSME credit gap? It removes one constraint, not all of them. Better analysis lets a lender establish income and obligations for a thin file borrower. It does not fix the cost of acquiring and collecting small tickets, the reach needed in dispersed Tier 3 markets, or the cash activity that never enters a bank account. Each binds on its own.

---

Sources: CRIF High Mark, MSMEx Spotlight, September 2026 (data as of June 2026), the June 2026 edition (data as of April 2026) and the December 2025 edition (data as of September 2025); TransUnion CIBIL and SIDBI, MSME Pulse, July 2026; FlexiLoans MSME Pulse, Festive Edition FY27 (23 September 2026), a company projection. Book a parallel evaluation on micro and small enterprise cases your team has already assessed, including ones you declined.