BANK CREDIT
The ₹2 lakh line
Nearly two in every three bank loans in India have a limit of ₹2 lakh or less. The Reserve Bank calls them small borrowal accounts, and they include crop loans, working capital for small shops, vehicle loans and credit cards. Using the Reserve Bank’s Basic Statistical Returns, we ask how much bank credit they hold, what they cost, and how their number has changed since the end of 2024.
The story continues below
WHERE PEOPLE SAVE
deposit accounts
Each dot on this screen is ten lakh bank deposit accounts: savings accounts, fixed deposits and current accounts. The Reserve Bank’s Handbook of Statistics counts 274.4 crore of them in March 2026, nearly two for every person in India.
WHERE PEOPLE BORROW
loans
The dots now lit are the loan accounts. Banks had 40.3 crore loan accounts in March 2026, one for every seven deposit accounts.
VILLAGE AND CITY
The gap is widest in the villages
The same dots are now sorted by place, with deposit accounts counted at the branch that holds them and loans where the money is used. While villages have nine deposit accounts for every loan, the big cities have five.
THE LINE
Nearly two in every three loans were below ₹2 lakh
One person can hold several credit accounts, and each credit card counts as one account. Farm loans are 33% of them, and personal loans—which includes credit cards—53%. Loans from finance companies, microfinance companies and lending apps are not in these figures, and most of those would fall below the ₹2-lakh line.
BEFORE THE ANSWER
Of every ₹100 that banks have lent, how much do you think goes to these small loans?
THE ANSWER
of every ₹100
Each dot is now an equal share of all bank credit outstanding. Loans of ₹2 lakh or less, hold ₹6 of every ₹100 that banks have lent.
THE OTHER END
loans hold ₹27 of every ₹100
These are the 20,475 loans of more than ₹100 crore each in March 2026. They hold twice as much money as all 33.4 crore loans of ₹5 lakh or less put together.
Interest rates fall as loans get larger. Green dots are money lent at under 8% a year: on loans above ₹100 crore, ₹63 of every ₹100. Terracotta dots are money lent at 13% or more: below the line, ₹35 of every ₹100. ₹32 of every ₹100 below the line is lent at 7% to 9%, much of it crop loans at a subsidised 7%. Loans up to ₹25,000 carry at least 12.4% a year on average, and the table’s top range holds credit cards at 37%.
WHO PAYS THE MOST
loans carry a rate of 13% or more
The loans with the highest rates sit at the bottom of the stack. Nearly half of all loans carry a rate of 13% or more, and together they hold only ₹6 of every ₹100 lent. Credit cards are 8.9 crore of them, 46%, each counted at its contract rate whether or not its holder ever pays interest.
The credit bureau TransUnion CIBIL counted about two cards a holder in March 2026. Excluding cards, that rate applies to one in every three of the other loans.
OVER TIME · JUNE 2017
Small loans, counted every quarter
Each dot represents one crore small loan accounts. In June 2017 there were 11.1 crore, nearly three in every four loans, and they held ₹7.2 of every ₹100 lent.
None of them were yet at the small finance banks, most of them former microfinance lenders. Those banks begin reporting the next quarter, and 25% of the rise since then is theirs.
MARCH 2020
The number of small loans rose while their share of the money stood still
By March 2020 there were 18.8 crore small loan accounts. Villages had added 2.2 crore since 2017, and the big cities 2.7 crore. Their share of the money stayed close to where it began, at ₹7.6 of every ₹100.
DECEMBER 2024
small loans—at their peak
In December 2024 there were two and a half times as many small loan accounts as in 2017, and they were more than seven in every ten loans. They held only ₹6.6 of every ₹100 lent, less than when there were far fewer of them.
The count had changed little for nine months: it was 27.2 crore in March 2024 and slipped in the two quarters after.
Note: December stands above March only because of a merger in October 2024, when a fintech lender’s loans were folded into a small finance bank. About 0.7 crore small app-based credit accounts, booked in Guwahati, entered the bank figures in that one quarter. Without them, the high point would be March 2024.
JUNE 2026
fewer small loans in eighteen months
By June 2026 there were 23.1 crore small loans, down from 27.3 crore in December 2024. They were a smaller share of all bank loans, and of all the money lent, than at any time since 2017. This was not due to small loans simply getting bigger and crossing ₹2 lakh. Loans above ₹2 lakh grew by 2.3 crore, which is almost exactly their normal rate of growth: they had grown by 2.2 crore in the eighteen months before. If millions of small loans had crossed over, that number would have jumped. That didn’t happen. Instead, the data show that bank loans of every size fell by 1.9 crore.
CREDIT CARDS
Cards grew while every other small loan shrank
It is interesting to note that credit cards—which are counted as small loans—kept growing. Small loans apart from credit cards fell by a whopping 4.6 crore, from 20.4 crore in December 2024 to 15.8 crore in June 2026.
WHERE THEY WENT
Villages lost loans filed as farm credit, while big cities lost durable-loan accounts
Between December 2024 and June 2026, every kind of place lost small loans. Villages lost 1.6 crore, 1.3 crore of them under agriculture (which includes group loans for dairy and livestock), mostly at private and small finance banks; public-sector banks added 13.9 lakh village loans.
The big cities lost 1.5 crore, almost all consumer-durable accounts, down 1.4 crore with balances barely moving. CRIF High Mark counts household-goods loans across all lenders rising from 10.3 crore to 11.3 crore in the year to June 2026, suggesting banks changed how they count these accounts rather than stopped lending. Credit cards kept growing.
The village losses track microfinance’s retreat after defaults from mid-2024: borrower caps of ₹2 lakh and four lenders (later three), scarce funding, Karnataka and Tamil Nadu laws against coercive recovery, and write-offs. CRIF counted 10.3 crore microfinance loans in June 2026, against 14.6 crore at end-2024.
Village bank loans of every size fell by 1.0 crore too, so many small loans closed rather than outgrew the line—though fewer people than loans lost credit.
Table view
| Dec 2024 | Jun 2026 | Change | |
|---|---|---|---|
| Villages | 6.37 | 4.80 | −1.57 |
| Small towns | 5.91 | 4.96 | −0.95 |
| Cities | 5.02 | 4.81 | −0.21 |
| Big cities | 9.98 | 8.52 | −1.46 |
| Farm loans, villages | 4.62 | 3.28 | −1.34 |
| Consumer durables, big cities | 2.14 | 0.70 | −1.44 |
| Credit cards, big cities | 5.16 | 5.52 | +0.36 |
| Other personal loans, cities | 1.40 | 1.63 | +0.23 |
From the quarterly Basic Statistical Return, Tables 2.1 and 3.5a, which count each loan at the branch that made it. All scheduled commercial banks except the regional rural banks.
WHOSE LOANS CLOSED
Loans to women fell, most steeply in the east
Women held 11.4 crore loan accounts in March 2024 and 9.9 crore by June 2026, while the amount banks had lent to women rose by 38%.
Loans to women fell in 375 of the 733 districts we can compare. Of the ten districts that lost the most outside Mumbai, eight are places where women hold most of the loans to individuals. That pattern suggests these are the branches where banks keep their group loans to women, so the ranking says where those books are kept rather than where the borrowers live.
A borrower with loans from two lenders holds two accounts, and when lenders stop renewing second loans the count of accounts falls faster than the count of people. The credit bureaus, which count both, report microfinance accounts falling faster than borrowers in the year to March 2026: on Equifax India’s count, published by Sa-Dhan, loan accounts fell 21% while borrowers fell 13%.
Table view
| State | Mar 2024 | Jun 2026 | Change |
|---|---|---|---|
| Maharashtra | 194.0 lakh | 140.6 lakh | −27.5% |
| Bihar | 74.1 lakh | 45.0 lakh | −39.4% |
| West Bengal | 89.8 lakh | 69.0 lakh | −23.1% |
| Karnataka | 94.8 lakh | 75.6 lakh | −20.3% |
| Odisha | 41.9 lakh | 30.8 lakh | −26.7% |
| Madhya Pradesh | 41.6 lakh | 32.3 lakh | −22.5% |
| Jharkhand | 23.9 lakh | 17.0 lakh | −29.2% |
| Uttar Pradesh | 73.0 lakh | 66.6 lakh | −8.8% |
| Rajasthan | 33.7 lakh | 28.4 lakh | −15.6% |
| Tamil Nadu | 184.2 lakh | 179.7 lakh | −2.4% |
| Delhi | 18.8 lakh | 16.4 lakh | −12.8% |
| Chhattisgarh | 12.0 lakh | 10.4 lakh | −13.4% |
| Punjab | 17.6 lakh | 16.5 lakh | −5.9% |
| Tripura | 4.5 lakh | 3.9 lakh | −14.5% |
| Uttarakhand | 5.1 lakh | 4.6 lakh | −10.0% |
| Haryana | 19.1 lakh | 18.8 lakh | −1.4% |
| Gujarat | 32.2 lakh | 32.0 lakh | −0.6% |
| Puducherry | 3.2 lakh | 3.1 lakh | −2.5% |
| Manipur | 1.0 lakh | 1.0 lakh | −4.5% |
| Sikkim | 0.5 lakh | 0.4 lakh | −9.1% |
| Lakshadweep | 0.0 lakh | 0.0 lakh | +47.5% |
| Dadra & Nagar Haveli and Daman & Diu | 0.2 lakh | 0.2 lakh | +10.1% |
| Ladakh | 0.1 lakh | 0.2 lakh | +19.3% |
| Mizoram | 0.4 lakh | 0.5 lakh | +8.0% |
| Andaman & Nicobar Islands | 0.2 lakh | 0.3 lakh | +23.2% |
| Goa | 1.6 lakh | 1.7 lakh | +4.4% |
| Meghalaya | 1.1 lakh | 1.1 lakh | +7.7% |
| Arunachal Pradesh | 0.5 lakh | 0.6 lakh | +20.4% |
| Nagaland | 0.8 lakh | 0.9 lakh | +14.8% |
| Chandigarh | 1.2 lakh | 1.3 lakh | +13.4% |
| Himachal Pradesh | 2.2 lakh | 2.4 lakh | +12.5% |
| Jammu & Kashmir | 4.3 lakh | 5.4 lakh | +25.2% |
| Kerala | 65.0 lakh | 66.4 lakh | +2.1% |
| Telangana | 38.0 lakh | 44.3 lakh | +16.5% |
| Andhra Pradesh | 63.4 lakh | 70.7 lakh | +11.5% |
From the quarterly Basic Statistical Return, Table 1.10, added up across the districts of each state, including the districts the table prints on two rows. The table counts each loan at the branch that made it and leaves out the regional rural banks. Assam is not compared: a fintech’s loans booked in Guwahati entered its figures in December 2024 through a merger into a small finance bank. Each state shows its figures on hover or on keyboard focus.
| District | State | Mar 2024 | Jun 2026 | Change | Women’s share |
|---|---|---|---|---|---|
| Mumbai Suburban | Maharashtra | 61.9 lakh | 37.7 lakh | −39% | 16% |
| Mumbai | Maharashtra | 42.7 lakh | 25.9 lakh | −39% | 7% |
| Chengalpattu | Tamil Nadu | 23.6 lakh | 12.8 lakh | −46% | 64% |
| Vaishali | Bihar | 15.8 lakh | 5.9 lakh | −63% | 82% |
| Wardha | Maharashtra | 10.5 lakh | 4.4 lakh | −58% | 69% |
| Varanasi | Uttar Pradesh | 12.3 lakh | 6.4 lakh | −48% | 58% |
| Ranchi | Jharkhand | 12.8 lakh | 6.9 lakh | −46% | 58% |
| Patna | Bihar | 11.9 lakh | 6.5 lakh | −46% | 39% |
| Jajpur | Odisha | 9.0 lakh | 3.8 lakh | −58% | 71% |
| South 24 Parganas | West Bengal | 12.1 lakh | 7.1 lakh | −42% | 69% |
WHERE THE MONEY GOES
Ten districts hold ₹42 of every ₹100
The Reserve Bank’s annual district return for March 2026 covers 777 districts and counts each loan where the bank reports the money is used. Ten of them, led by Mumbai, New Delhi and Bengaluru Urban, hold 42% of all bank credit and 23% of the loans. Mumbai’s island city alone holds 12%, and with Mumbai Suburban 16%.
Counted instead at the branch that made the loan, the ten largest districts hold 48%, and Mumbai’s two districts 21% rather than 16%, because loans booked at head offices in Mumbai are used across the country. Much of what they hold is credit to companies, finance companies and government bodies, recorded at their offices and spent everywhere. New Delhi district, one of Delhi’s eleven, carries ₹12.4 lakh crore. Many of the ten districts’ loans are credit cards, which banks book at their head offices rather than where the cardholder lives.
At the other end, the 489 smallest districts share 9% of the money between them, though they have 19% of the loans. The concentration is old and easing: in 2010 the ten largest districts held 58% of bank credit and Mumbai alone 22%.
Table view
| Rank | District | State | Share of credit | Share of loans | Running total |
|---|---|---|---|---|---|
| 1 | Mumbai | Maharashtra | 12.19% | 4.37% | 12.2% |
| 2 | New Delhi | Delhi | 5.78% | 0.79% | 18.0% |
| 3 | Bengaluru Urban | Karnataka | 4.54% | 2.33% | 22.5% |
| 4 | Mumbai Suburban | Maharashtra | 4.29% | 8.38% | 26.8% |
| 5 | Chennai | Tamil Nadu | 3.78% | 1.57% | 30.6% |
| 6 | Hyderabad | Telangana | 2.83% | 1.22% | 33.4% |
| 7 | Pune | Maharashtra | 2.78% | 1.45% | 36.2% |
| 8 | Ahmedabad | Gujarat | 2.21% | 0.86% | 38.4% |
| 9 | Kolkata | West Bengal | 1.85% | 0.90% | 40.3% |
| 10 | Gurugram | Haryana | 1.65% | 0.76% | 41.9% |
| 11 | Thane | Maharashtra | 1.27% | 1.87% | 43.2% |
| 12 | Jaipur | Rajasthan | 1.20% | 0.88% | 44.4% |
| 13 | Central Delhi | Delhi | 1.03% | 0.43% | 45.4% |
| 14 | Ernakulam | Kerala | 1.01% | 0.79% | 46.4% |
| 15 | Surat | Gujarat | 0.92% | 0.57% | 47.3% |
| 16 | Gautam Buddha Nagar | Uttar Pradesh | 0.78% | 0.44% | 48.1% |
| 17 | Coimbatore | Tamil Nadu | 0.77% | 0.61% | 48.9% |
| 18 | Medchal-Malkajgiri | Telangana | 0.75% | 0.57% | 49.6% |
| 19 | East Delhi | Delhi | 0.69% | 0.07% | 50.3% |
| 20 | Lucknow | Uttar Pradesh | 0.64% | 0.64% | 51.0% |
From the annual Basic Statistical Return, credit by district. The table view lists the top twenty, and every district is in the download.
State by state
Counted at the branch that made the loan, Maharashtra had 81 loans for every hundred people in March 2026, while Bihar had 9. Maharashtra had ₹4.6 lakh of bank credit per person and Bihar ₹22,200. Two things flatter Maharashtra in that first count. Its branches book loans that are used in other states, and of the credit used in Maharashtra, ₹68 of every ₹100 is used in the two Mumbai districts.
Many of its loans are also credit cards booked in Mumbai and held by people across the country. These quarterly figures also leave out the regional rural banks, which hold 18% of Bihar’s loans and 1% of Maharashtra’s.
Counted where the money is used and with those banks included, Maharashtra has 68 loans for every hundred people and Bihar 12, ₹4.0 lakh of credit per person against ₹25,500. To count the people, we divide each state’s income by its income per person, for the latest year both are published. The result is within 1% of the official population projections for 1 March 2026.
Table view
| State | Loans per 100 people | Credit per person | Population |
|---|---|---|---|
| Maharashtra | 81.3 | ₹4.6 lakh | 12.9 crore |
| Tamil Nadu | 49.9 | ₹2.5 lakh | 7.7 crore |
| Chandigarh | 44.2 | ₹8.5 lakh | 12.5 lakh |
| Kerala | 40.5 | ₹1.9 lakh | 3.6 crore |
| Puducherry | 39.8 | ₹1.6 lakh | 17.0 lakh |
| Telangana | 39.2 | ₹2.7 lakh | 3.9 crore |
| Assam | 38.6 | ₹42,400 | 3.6 crore |
| Andhra Pradesh | 35.4 | ₹1.6 lakh | 5.4 crore |
| Delhi | 35.3 | ₹9.0 lakh | 2.2 crore |
| Goa | 35.0 | ₹2.3 lakh | 15.9 lakh |
| Karnataka | 31.3 | ₹2.0 lakh | 6.9 crore |
| Haryana | 27.8 | ₹2.2 lakh | 3.1 crore |
| Jammu & Kashmir | 24.2 | ₹83,500 | 1.4 crore |
| Punjab | 21.8 | ₹1.3 lakh | 3.3 crore |
| Andaman & Nicobar Islands | 21.2 | ₹1.3 lakh | 4.0 lakh |
| Gujarat | 20.5 | ₹1.6 lakh | 7.4 crore |
| Sikkim | 18.1 | ₹1.1 lakh | 7.0 lakh |
| Odisha | 17.6 | ₹63,100 | 4.7 crore |
| Tripura | 16.9 | ₹37,400 | 42.0 lakh |
| Himachal Pradesh | 16.7 | ₹86,200 | 75.3 lakh |
| West Bengal | 16.4 | ₹70,500 | 10.0 crore |
| Uttarakhand | 16.0 | ₹93,600 | 1.2 crore |
| Rajasthan | 16.0 | ₹85,100 | 8.3 crore |
| Madhya Pradesh | 13.9 | ₹64,100 | 8.9 crore |
| Jharkhand | 11.9 | ₹39,500 | 4.0 crore |
| Chhattisgarh | 10.6 | ₹76,500 | 3.1 crore |
| Arunachal Pradesh | 10.3 | ₹71,700 | 16.0 lakh |
| Nagaland | 10.3 | ₹48,500 | 22.5 lakh |
| Uttar Pradesh | 10.2 | ₹42,400 | 23.9 crore |
| Mizoram | 10.0 | ₹46,900 | 12.6 lakh |
| Bihar | 9.2 | ₹22,200 | 13.2 crore |
| Manipur | 8.7 | ₹45,500 | 32.7 lakh |
| Meghalaya | 8.5 | ₹43,300 | 34.3 lakh |
From the quarterly Basic Statistical Return, Table 1.3, with each state’s population worked out from its income figures. The table counts each loan at the branch that made it and leaves out the regional rural banks. Credit cards are counted where the issuing bank books them, which lifts Maharashtra and Delhi. Assam’s figure is lifted by a fintech’s app-based loans, booked in Guwahati since a merger into a small finance bank in October 2024, and overstates what its own borrowers hold. The population counts everyone, not only adults, and the loans are accounts, not people.
WHERE INCLUSION HAPPENED
Small loans have doubled since 2017
When people say more Indians can borrow now, this is the part of it that happened at banks, counted in loan accounts rather than people. There were 26.2 crore small loan accounts in March 2026, and the Reserve Bank’s quarterly count of them, which leaves out the regional rural banks, has doubled since 2017. Microfinance companies that joined the return when they became small finance banks account for 25% of that rise, and credit cards, which come nearly two to a holder, for 43%.
Loans above the line grew faster still, to 3.3 times their 2017 number. TransUnion CIBIL, which counts people across every kind of lender, put credit-active adults aged 18 to 60 at 28% in March 2026 against 11% in March 2017.
Banks are no longer where most of the smallest loans are made. The Fintech Association for Consumer Empowerment (FACE), the digital lenders’ body, counts 13.2 crore personal loans sanctioned in 2025-26 by the NBFCs that lend mainly through apps. That is 77% of all personal loans sanctioned by number, though only 19% by value, and none of them are in these figures. What is not in doubt is the money: the small accounts hold ₹6 of every ₹100 banks have out on loan.
WHAT IT COSTS
The smallest loans carry the highest rates in the banks’ books
In March 2026 the smallest loans carried at least 12% a year, while the biggest carried about 8%. The small-loan rate is higher still once credit cards are priced at the 37% the Reserve Bank reports for them. Nearly half of all loans carry rates of 13% or more, and 46% of those are credit cards. Access to credit and its cost are separate questions, and on cost the smallest loans in the banks’ books carry the highest rates.
That is a gradient of products as much as of size, with subsidised crop loans at 7% in the same band as cards at 37%. The figures count loans, not people, so they cannot say who pays the most.
The gap has widened: in March 2014 the same tables priced loans up to ₹25,000 at about 12.5% and loans above ₹100 crore at about 11.6%, less than a point apart. The bank rate is also the cheapest formal price such borrowers see: Sa-Dhan, the other microfinance industry body, reports its member lenders outside the banks charging 22% to 24% across the quarters of 2025-26.
WHAT CHANGED
Small loans fell by about 4 crore between 2024 and 2026
Small loans grew for seven and a half years, to December 2024, though the count changed little through 2024. The December high point includes about 0.7 crore accounts that arrived in one quarter when a fintech’s loans were merged into a small finance bank. In the eighteen months since, they have fallen by 4.2 crore, and by more than that among the loans that were there before.
The fall is in the small loans that private banks and small finance banks file under agriculture, many of them group loans of the kind microfinance lenders make. It is also in a count of consumer-durable accounts that fell while the money lent on them hardly moved. It is a loss of loans more than of borrowers: the credit bureaus’ counts suggest many were the second or third loans of people who kept one. The fixed ₹2 lakh line explains little of it.
Among women the count fell most in Bihar, Jharkhand, Maharashtra, Odisha and West Bengal, while the money lent to women rose in every one of them. The next figures will show whether this was a pause, a retreat, or a move to lenders these figures cannot see.
Banks held about 26 crore small loan accounts in March 2026, many of them credit cards and held by fewer people than that. Those accounts hold ₹6 of every ₹100 banks have lent, and the unsecured ones among them carry the highest rates in the banks’ books. Since the end of 2024 there are about 4 crore fewer of them at banks, and where those borrowers went, bank figures cannot say.
WHAT THE RESEARCH SAYS
The counts hold, but they count accounts rather than people, include one merger and place loans where they are booked
The statistical standard is one borrower per person, however many accounts they hold. India reports loan accounts, not borrowers, to the IMF’s Financial Access Survey, which notes that many countries do so for want of data on account holders. TransUnion CIBIL (2026) counts 10.7 crore active credit cards held by 5.2 crore people, and CRIF High Mark (2026) 10.3 crore microfinance loans held by 6.6 crore borrowers. On those ratios the story’s 26 crore small accounts belong to perhaps 16 to 20 crore people.
Re-reading the same tables by bank group, we found that the December 2024 high point includes about 0.7 crore very small accounts. They entered in one quarter when the fintech Slice was amalgamated into North East Small Finance Bank under a scheme effective 27 October 2024, by the bank’s own annual report. The small finance banks’ urban small loans and Assam’s loan accounts step up by the same amount that quarter, and the book has since grown: Assam’s accounts stand 1.0 crore above their September 2024 count. December 2024 stays the published peak, but net of the entry the count stopped growing in March 2024, and the fall since is larger than 4.2 crore, not smaller.
The Reserve Bank’s Financial Stability Report (December 2025) says unsecured retail lending declined sharply after its November 2023 increase in risk weights. The tables agree for the loans the rule covered: small loans in the big cities peaked in December 2023, the month after the circular. The rule never covered farm loans, where most of the fall is, and the Reserve Bank took banks’ microfinance loans back out of it in February 2025. MFIN (2026) records seven quarters of microfinance contraction from mid-2024 before a turn in the March 2026 quarter. In the year to March 2026 every lender type shrank, banks most, though MFIN says much of banks’ later decline is loans reclassified as retail. It attributes the fall to curtailed funding and stricter underwriting under its guardrails.
Cull, Demirgüç-Kunt and Morduch (2018), on 1,335 microfinance institutions worldwide, find unit costs substantially higher for small loans, reflecting the fixed cost of making each loan. The Reserve Bank’s own surveys of small borrowal accounts found small accounts paying slightly less than large ones on average until 2008 (11.5% against 12.3% in March 2008), before the base-rate system of 2010 removed the ceiling on their pricing. The story’s 12.4% is a floor: the table’s top range is open and holds credit cards, which the Reserve Bank prices at 37%. Outside banks the same loans cost 22% to 24% (Sa-Dhan, 2026).
FACE (2026) counts 13.2 crore personal loans sanctioned in 2025-26 by the NBFCs that lend mainly through apps, 77% of all personal loans sanctioned by number and 19% by value, none of them in the bank returns. Gopal and Schnabl (2022) found US bank small-business lending in 2016 still 24% below its pre-crisis level, while total lending, once finance companies and fintech lenders were counted, grew no less in the counties that had depended most on banks. A bank-only count says where a fall happened at banks, not whether borrowers lost credit. The Guwahati entry shows how far one lender’s change of status can move that count.
THE NUMBERS BEHIND THE DOTS
The exact figures
The dots in the story round the figures off. These charts give them exactly, and each has a table view with every value.
Table view
| Deposit accounts | Loans | Deposit accounts per loan | |
|---|---|---|---|
| Villages | 85.8 crore | 9.5 crore | 9.0 |
| Small towns | 75.8 crore | 9.3 crore | 8.1 |
| Cities | 50.2 crore | 8.3 crore | 6.0 |
| Big cities | 62.5 crore | 13.2 crore | 4.7 |
From the Handbook of Statistics on the Indian Economy: deposits and credit by population group.
Table view
| Loan size | Average rate | Money at 13% or more | Loans at 13% or more |
|---|---|---|---|
| up to ₹25,000 | 12.4% | 64% | 75% |
| ₹25,000 to ₹2 lakh | 10.8% | 34% | 56% |
| ₹2 to 5 lakh | 10.3% | 19% | 35% |
| ₹5 to 10 lakh | 10.3% | 15% | 25% |
| ₹10 to 25 lakh | 9.7% | 7% | 13% |
| ₹25 to 50 lakh | 8.7% | 3% | 3% |
| ₹50 lakh to ₹1 crore | 8.3% | 2% | 2% |
| ₹1 to 4 crore | 8.5% | 1% | 1% |
| ₹4 to 6 crore | 8.6% | 1% | 2% |
| ₹6 to 10 crore | 8.5% | 1% | 2% |
| ₹10 to 25 crore | 8.5% | 1% | 3% |
| ₹25 to 100 crore | 8.4% | 1% | 3% |
| above ₹100 crore | 7.6% | 0% | 2% |
From the annual Basic Statistical Return, Table 2.1. Each rate is a weighted midpoint of DBIE’s rate ranges, taking 14% for “13% and above”. That top range has no upper end, and it holds 64% of the money in loans up to ₹25,000 and 91% of credit-card balances (Table 3.3). The rates shown for the small bands are therefore lower bounds, and the true slope is steeper. The Reserve Bank’s own weighted rate for credit cards (quarterly BSR-1, Table 1.9, March 2026) is 37.0%, and until December 2020 DBIE’s ranges ran to “20% and above”. The rates are the contracted rates banks report, before fees.
Table view
| Loan size | Loans | Share | Amount | Share |
|---|---|---|---|---|
| up to ₹25,000 | 5.5 crore | 13.6% | ₹0.3 lakh crore | 0.1% |
| ₹25,000 to ₹2 lakh | 20.7 crore | 51.5% | ₹12.5 lakh crore | 5.8% |
| ₹2 to 5 lakh | 7.2 crore | 17.8% | ₹15.7 lakh crore | 7.3% |
| ₹5 to 10 lakh | 3.6 crore | 9.0% | ₹17.4 lakh crore | 8.1% |
| ₹10 to 25 lakh | 2.1 crore | 5.3% | ₹24.4 lakh crore | 11.4% |
| ₹25 to 50 lakh | 70.4 lakh | 1.7% | ₹18.9 lakh crore | 8.8% |
| ₹50 lakh to ₹1 crore | 26.9 lakh | 0.7% | ₹14.2 lakh crore | 6.6% |
| ₹1 to 4 crore | 13.1 lakh | 0.3% | ₹17.8 lakh crore | 8.3% |
| ₹4 to 6 crore | 1.5 lakh | 0.0% | ₹5.3 lakh crore | 2.5% |
| ₹6 to 10 crore | 1.1 lakh | 0.0% | ₹6.0 lakh crore | 2.8% |
| ₹10 to 25 crore | 1.0 lakh | 0.0% | ₹9.6 lakh crore | 4.5% |
| ₹25 to 100 crore | 0.6 lakh | 0.0% | ₹14.5 lakh crore | 6.8% |
| above ₹100 crore | 0.2 lakh | 0.0% | ₹57.7 lakh crore | 26.9% |
From the annual Basic Statistical Return, Table 1.8: all 40.3 crore loans.
Table view
| Quarter | Small loans | Share of all loans | ₹ of every ₹100 |
|---|---|---|---|
| Jun 2017 | 11.05 crore | 73.2% | ₹7.2 |
| Sep 2017 | 11.69 crore | 73.6% | ₹7.2 |
| Dec 2017 | 12.37 crore | 74.0% | ₹7.2 |
| Mar 2018 | 12.71 crore | 73.9% | ₹7.1 |
| Jun 2018 | 12.98 crore | 73.8% | ₹7.2 |
| Sep 2018 | 13.52 crore | 74.0% | ₹7.1 |
| Dec 2018 | 14.42 crore | 74.4% | ₹7.2 |
| Mar 2019 | 15.50 crore | 75.0% | ₹7.3 |
| Jun 2019 | 16.79 crore | 76.1% | ₹7.6 |
| Sep 2019 | 17.64 crore | 76.3% | ₹7.7 |
| Dec 2019 | 18.44 crore | 76.4% | ₹7.8 |
| Mar 2020 | 18.83 crore | 76.2% | ₹7.6 |
| Jun 2020 | 18.88 crore | 76.2% | ₹7.7 |
| Sep 2020 | 19.33 crore | 75.8% | ₹7.9 |
| Dec 2020 | 20.07 crore | 75.8% | ₹7.9 |
| Mar 2021 | 20.53 crore | 75.6% | ₹7.9 |
| Jun 2021 | 20.27 crore | 75.3% | ₹7.8 |
| Sep 2021 | 20.73 crore | 75.2% | ₹8.2 |
| Dec 2021 | 21.35 crore | 75.1% | ₹7.8 |
| Mar 2022 | 21.98 crore | 75.0% | ₹7.7 |
| Jun 2022 | 22.63 crore | 74.9% | ₹7.6 |
| Sep 2022 | 23.55 crore | 74.9% | ₹7.6 |
| Dec 2022 | 23.93 crore | 74.5% | ₹7.5 |
| Mar 2023 | 24.64 crore | 74.3% | ₹7.4 |
| Jun 2023 | 25.39 crore | 74.2% | ₹7.4 |
| Sep 2023 | 26.37 crore | 73.6% | ₹7.1 |
| Dec 2023 | 27.00 crore | 73.3% | ₹7.0 |
| Mar 2024 | 27.22 crore | 72.9% | ₹7.0 |
| Jun 2024 | 27.18 crore | 72.3% | ₹7.0 |
| Sep 2024 | 26.85 crore | 71.4% | ₹6.8 |
| Dec 2024 | 27.28 crore | 71.1% | ₹6.6 |
| Mar 2025 | 26.49 crore | 69.8% | ₹6.3 |
| Jun 2025 | 25.88 crore | 68.7% | ₹6.1 |
| Sep 2025 | 24.93 crore | 67.2% | ₹5.8 |
| Dec 2025 | 24.44 crore | 65.9% | ₹5.5 |
| Mar 2026 | 24.02 crore | 64.6% | ₹5.2 |
| Jun 2026 | 23.09 crore | 63.2% | ₹5.0 |
From the quarterly Basic Statistical Return, Table 2.1, which leaves out the regional rural banks and counts each loan at the branch that made it. The December 2024 quarter includes about 0.7 crore accounts that entered when a fintech’s loans were merged into a small finance bank.
SOURCES
Reserve Bank of India, Basic Statistical Returns of scheduled commercial banks, as published on DBIE. The annual BSR-1 for March 2026, which includes the regional rural banks: Table 1.8 (credit by size of limit), Table 2.1 (size of limit by interest rate), Table 3.3 (interest rate by occupation), Table 1.13 (small borrowal accounts by occupation), Table 2.13 (states, by place of sanction), the district-wise return (by place of utilisation, every year from 2010) and Table 2.14 (districts by place of sanction). The quarterly BSR-1, March 2014 to June 2026, which leaves them out and counts each loan at the branch that made it: Table 2.1 (small borrowal accounts by population group and bank group), Table 1.7 (size of limit), Table 1.3 (states), Table 3.5a (small borrowal accounts by occupation), Table 1.9 (the Reserve Bank’s weighted average lending rate by occupation), Table 2.5 in its older format (size of limit by interest rate, with ranges to “20% and above”, to December 2020) and Table 1.10 (credit to individuals by district and gender). The Handbook of Statistics on the Indian Economy for deposit accounts and loans by population group, with credit by place of utilisation, and the net state domestic product series for population. The ₹2 lakh line dates from the March 1999 return, when the Reserve Bank raised the cut-off from ₹25,000 (RBI, Basic Statistical Returns, volume 28, March 1999, introduction; RBI Bulletin, Survey of small borrowal accounts: 2008, May 2011). The November 2023 change to risk weights on consumer credit is RBI notification RBI/2023-24/85; the circular of 25 February 2025 that took banks’ microfinance loans back out of it is RBI/2024-25/119, and RBI/2024-25/120, issued the same day, restored the risk weights on bank lending to NBFCs from 1 April 2025. The 10% yearly step-up in Kisan Credit Card limits is in the RBI’s master circular on the scheme (RBI/2017-18/4), and the 7% rate is the interest subvention scheme for 2025-26. The microfinance industry’s caps of ₹2 lakh per borrower and four, then three, lenders are MFIN’s directives of 8 July and 25 November 2024. The merger that put a fintech’s loans into the returns in December 2024 is recorded in slice Small Finance Bank’s annual report for 2024-25 (scheme of amalgamation effective 27 October 2024). Counts of people rather than loans, and of loans at lenders outside the banks, are from TransUnion CIBIL (8 April, 8 July and 30 July 2026), CRIF High Mark (MicroLend, August 2026, and How India Lends, September 2026), Sa-Dhan (quarterly report for January to March 2026), FACE (Digital personal loans, March 2026) and the World Bank’s Global Findex 2025. The population projections are those of the Technical Group on Population Projections (July 2020).
FURTHER READING
Demirgüç-Kunt, Beck and Honohan (2008). Finance for all? Policies and pitfalls in expanding access. World Bank.
Access against use, and how many people borrow against how much: the distinction the closing section rests on.
World Bank (2025). The Little Data Book on Financial Inclusion 2025 (Global Findex).
India’s page: 89% of adults with an account in 2024 and 15% having borrowed formally in the past year, the person-level counterpart to account counts.
Reserve Bank of India (2011). Survey of small borrowal accounts: 2008. RBI Bulletin, May 2011.
The Reserve Bank’s own sample of small accounts: what they were for, what they cost, and the history of the cut-off since 1972.
Rosenberg, Gonzalez and Narain (2009). The new moneylenders. CGAP Occasional Paper 15.
Why tiny loans carry high rates and how to judge whether a rate is exploitation, in plain terms (full title: The new moneylenders: are the poor being exploited by high microcredit interest rates?).
Chavan (2016). Inclusion of small borrowers by Indian banks. Reserve Bank of India.
The long series behind the ₹2 lakh line, when the line was set, and the caveat that comes with a cut-off fixed in rupees.
Sriram (2021). Banking system and financial inclusion. Inclusive Finance India Report 2021, chapter 3.
The same small borrowal account tables read year by year, with the question whether the growth is inclusion or the splitting of credit into ever smaller loans.
TransUnion CIBIL (2026). Credit growth in Uttar Pradesh, Madhya Pradesh, Bihar outpaced that of traditional strongholds.
What a credit bureau sees when it counts people rather than accounts: 28% of adults aged 18 to 60 credit-active in March 2026, up from 11% in March 2017.
Reserve Bank of India (2025). Financial Stability Report, December 2025, chapter I.
The Reserve Bank’s own account of the unsecured-retail slowdown and the microfinance contraction.
MFIN (2026). Micrometer for the June 2026 quarter.
The microfinance sector’s own numbers by lender type, with banks a quarter of the book and most of their year’s decline loans reclassified as retail.
slice Small Finance Bank Limited (2025). Annual Report 2024-25 (formerly North East Small Finance Bank).
The amalgamation, effective 27 October 2024, that put a fintech’s NBFC loan book into the bank returns in a single quarter. The bank publishes no count of its loan accounts, so the step is read from the DBIE tables.
Goldston and Lee (2020). Measurement of small business lending using Call Reports. FDIC Staff Study 2020-04.
What a loan-size line that nobody re-bases does to a series, and how to correct it.
Gopal and Schnabl (2022). The rise of finance companies and FinTech lenders in small business lending. Review of Financial Studies.
The cleanest case of a fall in bank lending that was not a fall in lending once non-banks were counted.
DOWNLOADS
METHOD
Loans, not people. Every count here is of loans, strictly loan accounts: one person can have several, and each credit card counts as one. The Reserve Bank publishes no count of distinct bank borrowers. India reports loan accounts, not borrowers, to the IMF’s Financial Access Survey, whose standard is one borrower per person however many accounts they hold.
What the bureaus count. Where the credit bureaus do count people, the gap is wide. In March 2026, 10.7 crore active credit cards were held by 5.2 crore people, and in June 2026, 10.3 crore microfinance loans by 6.6 crore borrowers. By TransUnion CIBIL’s count, 28% of adults aged 18 to 60 had a live retail loan in March 2026, against 11% in March 2017. On those ratios the 26.2 crore small loans belong to roughly 16 to 20 crore people, and the fall since 2024 is smaller in people than in loans.
The line. The ₹2 lakh line is the Reserve Bank’s own: a small borrowal account is a loan whose sanctioned limit is ₹2 lakh or less. The Bank drew the line in the March 1999 return, when it raised the cut-off from ₹25,000, and has not moved it since; the regional rural banks moved to the same line in March 2002. A limit is fixed when a loan is sanctioned. A loan crosses the line only when its limit is renewed higher, as a Kisan Credit Card limit is by 10% a year, or when a bigger loan replaces it.
Crossing the line. Some loans cross the line every year as their limits are renewed. That is why the share of small loans slipped from 76 in every hundred in March 2020 to 71 by December 2024 while their number was still rising. It does not explain the fall since then. Loans above the line grew by 2.3 crore in the eighteen months to June 2026, and grew by 2.2 crore in the eighteen months before. The number of loans of every size fell by 1.9 crore over the same eighteen months.
About the dots. At the start, each dot is ten lakh accounts, rounded so that the dots add up to the totals. When the stack is counted by money, each of the 403 loan dots stands for ₹0.25 of every ₹100 lent. In the scenes over time, each dot is a crore small loans, and a part-lit dot shows the fraction. The charts further down give the exact values.
Two sets of figures. The yearly figures for March 2026 include the regional rural banks. They give the stack of loans, what loans cost, and the districts. The quarterly figures leave those banks out: 2.2 crore of the 26.2 crore small loans and ₹2 lakh crore of their ₹12.8 lakh crore. They give everything that moves through time, the states, and the figures on women. The two are never mixed in one number.
Banks only. The Basic Statistical Returns are filed by scheduled commercial banks: public and private sector banks, foreign banks, the small finance banks and, in the yearly figures, the regional rural banks. They include the loans that microfinance companies make as banks’ agents. Finance companies, microfinance companies lending on their own books, lending apps, and cooperative banks and societies do not file them, so their loans appear nowhere in this story. That is a great deal of small lending. In June 2026 CRIF High Mark counted 10.3 crore microfinance loans across all lenders, and in 2025-26 the NBFCs that lend mainly through apps sanctioned 13.2 crore personal loans.
What the bank figures cannot tell. A fall in bank loans can mean borrowers moving to those lenders, or those lenders shrinking too, and the bank figures cannot tell the two apart. In the year to March 2026 microfinance shrank at every kind of lender, while in the year to June 2026 consumer durable loans grew across all lenders as they fell at banks. A lender can also join the returns: when a fintech’s loans were merged into a small finance bank in October 2024, about 0.7 crore accounts entered them in one quarter. Between December 2024 and June 2026, 1.4 crore consumer-durable accounts in the big cities left the return while the money outstanding on them hardly moved. That points to a reclassification rather than a repayment.
Breaks in the series. Accounts enter the quarterly figures when a lender becomes or merges into a bank, and leave when a bank closes or reclassifies them, without any loan being made or repaid. The small finance banks, most of them former microfinance lenders, report from September 2017, one quarter after the series here begins, and 25% of the rise since then is theirs. In December 2024 about 0.7 crore small app-based credit accounts, booked in Guwahati, entered the return in one quarter when a fintech’s loans were merged into a small finance bank. They are still on the books and have grown: Assam’s loan accounts stand 1.0 crore above their September 2024 count. They lift the December 2024 count, the urban count, and Assam’s and Kamrup Metropolitan’s figures from then on. Without them the count would not have risen above March 2024’s, and the fall since December 2024 would be somewhat larger than 4.2 crore.
Villages, towns and cities. These are the Reserve Bank’s four population groups. Rural centres have fewer than 10,000 people, semi-urban centres fewer than a lakh, urban centres fewer than ten lakh, and metropolitan centres ten lakh or more.
Where a loan is counted. The Reserve Bank records two places for every loan, the branch that sanctioned it and the place where the money is used, and publishes some tables on each basis. The district figures, and the deposit-and-loan counts by village and city at the start, count a loan where the bank reports the money is used. Everything else, the small loans by village and city over time, the states, and the figures on women, counts it at the branch that made it. Even by use, a loan to a company, a finance company or a government body is placed at the borrower’s office. That is why New Delhi district shows ₹12.4 lakh crore of credit.
Books that move. A lender that moves its loans to a central office moves them to that district, and across state lines. That is the likeliest reading of what happened in Mumbai between March and June 2026. Loans to individuals, men’s and women’s alike, left the books of branches in Mumbai Suburban while the all-India total kept rising. The same thing happens in reverse in Assam, where a fintech’s loans, made across the country through an app, are booked at the merged bank’s office in Guwahati. From December 2024 they appear as Assam’s in the state figures, as Kamrup Metropolitan’s in the district figures on women, and as urban in the village-to-city figures.
Group loans and agents. Banks also book their group loans to women at a few branches. The districts that lost the most loans to women, Chengalpattu, Vaishali, Wardha, Varanasi and Ranchi among them, are where those books are kept. Loans that microfinance institutions make as banks’ business correspondents are on the bank’s books (RBI circular RBI/2005-06/288) and enter the return through the bank branch that holds them. For a joint account, the gender is that of the first holder (RBI, Handbook of instructions for the Basic Statistical Returns, 2008).
Like with like. DBIE prints many districts on two rows, one carrying the older quarters and one the newer. Each state’s figure adds up every row, and each district its rows by name. Districts renamed or redrawn between March 2024 and June 2026, such as Ramanagara, now Bengaluru South, and Ahmadnagar, now Ahilyanagar, are left out of the district comparisons. Assam is left out altogether: a fintech’s loans booked in Guwahati entered its figures in December 2024 through a merger into a small finance bank, so its two dates would not cover the same lenders. The maps show the 28 states, Delhi, and Jammu & Kashmir; the smaller union territories are in the downloads.
Nothing typed by hand. Every number from the returns is read from the DBIE tables by a program that checks the totals and the shape of each table every time it runs. The sentences in the story use those same numbers. The interest rate for each loan size is a weighted midpoint of DBIE’s rate ranges. The only figures typed in are those cited from other publications, which are named where they appear.