Fintech Lenders: What Is Actually Different About Them
Open two lending applications on the same phone. Both ask for the same details, both come back inside a few minutes, and both put money into an account by the evening. From the borrower's side of the screen they are the same product. Behind the two screens can sit two completely different businesses, funded differently, holding different things, and failing in ways that have nothing to do with each other. Neither screen shows which is which, and that is not an oversight in the design. The screen is simply not where the answer lives.
A lender is classified by what it owes rather than by what it looks like. So the place to start is the far end from the interface, on the side of the business nobody borrowing ever sees.
Rukmini Finance Limited, invented, is a finance company that lends and takes no deposits. Across the one year on record it carries assets under managementHow large a lender's book of loans is. Purely a measure of size rather than of earnings, and here it is the denominator sitting underneath every percentage that follows. of Rs 18,000 crore, against which Rs 14,400 crore has been raised from parties in the market and Rs 3,600 crore is its own net worthWhatever remains for a business's own holders after everything it owes has been settled. No interest attaches to it, and that single fact does most of the work in what follows..
Suvarna Commercial Bank Limited, also invented, appears once and only for a comparison at the close. It reports total assets of Rs 2,40,000 crore, net worth of Rs 24,000 crore and profit after tax of Rs 2,250 crore for its own stated year.
Setu Payments Limited, invented, is named once where money has to move between two accounts and nowhere else. What exists here is a single year for each lender and not one line beyond it. There is no earlier year, no downturn and no failure of any kind. One year cannot serve as evidence about how well either way of running a lending business works.
Two lending applications look identical to somebody borrowing. How much does that reveal about the two businesses behind them?
What is a fintech lender, and what actually settles the question?
Three questions, asked in order about any business calling itself a lender, settle the matter before anything else is examined. Where does its money come from. Whose set of books is the loan sitting on. Who carries the loss if the money does not come back. Not one of those three is answered by the application, the approval time or the interface. They are answered on the funding side, in the loan agreements, and in whatever was signed between the parties involved.
Here is the everyday version. A vegetable seller and a supermarket both hand over a bag of tomatoes at the front counter. The counter is identical. Behind it, one of them bought the tomatoes this morning with money from a moneylender and carries the loss if they rot, and the other is selling them on behalf of somebody else for a cut. The difference cannot be seen from the queue, and the difference is the whole business.
The order matters as much as the questions. Trace the funding first. Everything downstream is a consequence of it: what the money costs, when it has to be given back, how much lending it supports, and what happens on the day it is not renewed. The front end is the part somebody borrowing sees and the part that tells an analyst least. The pairing is uncomfortable and reliable.
What did software genuinely change in a lending business?
Three things changed, and being strict about the list matters. A longer list starts smuggling in claims that no lending arithmetic supports. Software changed the cost of doing the work, the speed of doing it, and the reach of the business, and that is the whole list.
The cost first. An application that is taken, checked and decided without a branch, a queue, a physical file and somebody walking it between desks costs less to process than one that is not. The saving is not a matter of clever technology. The saving is a matter of how many hands touch the work.
The speed next. A decision that comes back in minutes reaches somebody whose need is this week and not the one after next, and for a broken shutter or a stock purchase before a festival, later is the same as never. Speed here is not a convenience feature. Speed changes which requests are worth making at all.
The reach last, and it is the one most often left out. A lender without branches is not confined to the places it has buildings. Ten shops in one shopping centre can each be reached by a lender that never opens an eleventh unit in that centre. The everyday version is a tailor who used to take orders only from people walking past the shop and now takes them from three towns over, without renting anything in those towns.
And what did it leave exactly where it was?
The unchanged half gets left out more often, and it deserves the same room. Three things did not move at all.
The funding did not change. Money still has to be raised from somebody, it still carries a rate, and it still has a date on which it has to be given back. No amount of software makes a rupee arrive without a lender on the other side of it. The cost of fundsThe annual price charged to a lender on money it has raised. Being a price it always comes with a period attached, and it applies to the borrowings alone rather than to the whole book. is a rate somebody else sets, and the negotiation is the same negotiation it always was.
The arithmetic did not change. A price still has to carry three things: what the money costs, what the work costs, and whatever comes back short of what was lent. Take any one of the three out of the price and the business loses money more slowly or more quickly, but it loses money.
And the capital a lender holds against what it lends did not change. The requirement itself, and the base it is struck on, are set by the Reserve Bank of India at rbi.org.in, and both move. The amount a lender holds behind the book is decided somewhere else entirely. A business that has made the work cheaper has not made itself lighter.
Name one thing software changed in a lending business and one thing it did not. Which pairing below is right?
Which three shapes can a fintech lender take?
The sorting runs on whose set of books the loan sits on, never on what the business calls itself. Ownership of the loan puts any real arrangement into one of three shapes, and the shape settles what the business carries.
Shape one: the business is the lender. It is registered to lend, it lends off its own accounts, it funds itself in the market, it holds the loan, and it carries the loss. Rukmini Finance Limited is this shape. Its arithmetic can therefore be worked all the way through.
Shape two: the business is not a lender at all. It finds the borrower, or it services the loan, or it collects on it, for somebody else who is the lender. The business holds no loan, so no loan appears on its books, and it carries no loss unless it has separately agreed to carry one. The party in this shape is called the lending service provider. The requirements on a party that sources, services or collects a loan for a lender are set by the Reserve Bank of India at rbi.org.in, and they move.
Shape three: the loan is shared. Two lenders each hold their own part of the same loan and each funds its own part. Both are lenders, both carry loss on their share, and both appear in the arithmetic.
Notice the four things the sorting ignores: the presence of an application, the age of the business, the word technology, and how quickly money arrives. Whose books is the loan on is the whole test, and every real arrangement answers it, whether or not it wants to.
One business runs an application, finds borrowers and chases repayments, and holds no loans at all. Which shape is it, and what does it carry?
How NBFCs and Fintech Lenders Make Money: which lines are the business?
Two revenue lines and three cost lines, and every one of the five is quoted here on the same base, assets under management of Rs 18,000 crore, for the same period, one year. The single base is not decoration. A common base and a common period are the only reason the lines can be added and subtracted at all.
The revenue lines are the spreadHow far the rate a lender charges on its lending sits above the rate it is charged on its borrowing, both measured over one period. Here both rates are annual, so the gap is annual too., being what lending earns less what borrowing costs, and fees, being what is charged for doing something. The cost lines are the cost of funds, the cost of doing the work, and whatever comes back short.
A fee is charged once and a spread is earned for as long as the loan runs, so the spread is the business and the fee is a supplement. A shop that charges a delivery fee once and then earns nothing more from that customer is in a different business from one that earns a little every month for three years. Both are real revenue. Only one of them compounds with the size of the book.
Now the whole of Rukmini Finance Limited, worked line by line. A yield of 14.50 per cent a year across the Rs 18,000 crore book brings in Rs 2,610 crore. Against that, the Rs 14,400 crore of borrowings carry 8.50 per cent a year, an interest bill of Rs 1,224 crore. Net interest incomeWhatever survives of a period's interest receipts once that same period's interest bill has been taken out. A rupee figure to begin with, and a percentage only once the divisor is stated. is therefore Rs 1,386 crore, and set against that same Rs 18,000 crore book it reads 7.70 per cent for the year.
| The year at Rukmini Finance Limited | Rs crore | On assets under management of Rs 18,000 crore |
|---|---|---|
| Interest earned, at 14.50 per cent a year on the book | 2,610 | 14.50 per cent |
| Interest paid, at 8.50 per cent a year on borrowings of Rs 14,400 crore | 1,224 | 6.80 per cent |
| Net interest income | 1,386 | 7.70 per cent |
| Less the cost of doing the work | 540 | 3.00 per cent |
| Less the charge for loans not repaid in full | 396 | 2.20 per cent |
| What is left before tax | 450 | 2.50 per cent |
| Tax, at 25.0 per cent of the amount above | 112.50 | 0.625 per cent |
| Profit after tax | 337.50 | 1.875 per cent |
Read the last column downwards and the whole business is one stack on one base. 7.70 less 3.00 less 2.20 is 2.50, and 2.50 per cent of Rs 18,000 crore is Rs 450 crore, so the stack ties in both directions. Every line sits on the same denominator and the same twelve months. The subtraction is honest rather than merely tidy for that reason alone.
The gap between what Rukmini Finance Limited earns on lending, 14.50 per cent a year, and what it pays on borrowing, 8.50 per cent a year, is 6.00 points. Is its net interest income more or less than 6.00 per cent of its book?
Why is the margin on the book wider than the gap between the two rates?
One step in the arithmetic catches almost everybody the first time. Rukmini Finance Limited earns 14.50 per cent a year and pays 8.50 per cent a year, a gap of 6.00 points, and yet its net interest income is 7.70 per cent of assets under management. The margin sits 1.70 points above the gap, and the reason is arithmetic rather than error.
Follow it in two lines. The 6.00 point gap is earned on the whole book: 6.00 per cent of Rs 18,000 crore is Rs 1,080 crore. But the lender does not borrow the whole book. The lender borrows Rs 14,400 crore and funds the other Rs 3,600 crore with its own net worth, on which it pays nothing: 8.50 per cent of Rs 3,600 crore is Rs 306 crore that never leaves. Rs 1,080 crore plus Rs 306 crore is Rs 1,386 crore, exactly the net interest income in the table above.
The lender's own money funds part of the book at no interest cost, and on this book that is worth 1.70 points of margin on assets under management. The household version. A scooter bought half on a loan and half out of savings carries interest only on the borrowed half. The rides are on the whole scooter. The saving is real, it is the size of the interest on the half that was not borrowed, and nobody sends a receipt for it.
Where does the cost of doing the work actually sit?
Operating expensesWhat it costs to run the business itself: people, systems, collections, rent and everything else that is not interest and not a charge for loans that went unpaid. The figure is quoted as a share of the book. at Rukmini Finance Limited are Rs 540 crore for the year, which is 3.00 per cent of assets under management of Rs 18,000 crore. Operating expenses are the line software is supposed to move, so the figure is worth putting beside its neighbours rather than admiring on its own.
On the same base and the same twelve months: net interest income 7.70 per cent, the cost of doing the work 3.00 per cent, the credit costThe charge a lender takes in its own accounts for a period, covering loans not repaid in full. The charge is a flow through the year rather than a stock sitting on the balance sheet. 2.20 per cent, and 2.50 per cent left before tax. The cost of doing the work is the larger of the two cost lines, so a change in it is material rather than cosmetic.
And here the arithmetic stops. Only one lender and one year sit behind these figures. There is no second lender to compare against and no earlier year for this one, so there is no saving to state. The size of the line can be given and the saving cannot, and the difference between those two sentences is the difference between arithmetic and a claim. A business that says its costs are lower is making a comparison, and a comparison needs two observations.
Rukmini Finance Limited's operating expenses are Rs 540 crore for the year. Expressed on the base used throughout, where does that rank among the cost lines?
Where does a claim on one set of books show up on another?
Here is a habit that catches errors nothing else catches. Almost every line in a lending business is one side of a pair. If Rukmini Finance Limited owes Rs 14,400 crore, somebody else is holding a claim of Rs 14,400 crore, and the two entries add to nothing. A claim that does not net to zero somewhere is either an error or it is not a claim, and knowing which is worth the ten seconds it takes to check.
| The line, for the stated year | On this lender's books | On the other side | Nets to |
|---|---|---|---|
| Borrowings of Rs 14,400 crore | owed by it | held by the parties that lent | zero |
| The book of loans, Rs 18,000 crore | held by it | owed by those who borrowed | zero |
| Net worth of Rs 3,600 crore | owed to its own holders | held by those holders | zero |
| Interest paid, Rs 1,224 crore | a cost to it | income to the parties that lent | zero |
| Interest earned, Rs 2,610 crore | income to it | a cost to those who borrowed | zero |
| A fee paid to a lending service provider, at any size at all | a cost to it | income to that party | zero |
| The charge for loans not repaid, Rs 396 crore | a charge in its own accounts | no matching entry anywhere | does not pair |
The last row is the interesting one. Every other line moves a claim from one place to another, so it appears twice with opposite signs. The charge for loans not repaid does not move anything. The charge is a claim marked down rather than a claim that travelled, and a marked-down claim has nobody on the other side of it. The fee row nets to zero at any size whatsoever, which is why the row can stand without a number in it.
What happens when the market stops lending to the lender?
Rukmini Finance Limited takes no deposits. Its Rs 14,400 crore of borrowings were lent to it by other parties, and each of those parties decides again, every time a facilityA borrowing arrangement with a stated size and a stated end date. When it ends, continuing is a fresh decision by the party lending rather than the old arrangement carrying on. comes up for renewal, whether to lend again. The renewal decision is not made by the lender and it is not made on the lender's timetable.
The trouble arrives on the funding side. The lending side is completely unchanged, and nothing needs to be wrong with anybody's repayments for this lender to be in difficulty. Every rupee due can arrive on the day it is due and the business can still be stuck. A funding failure is a different kind of vulnerability, not a larger one, and reading it as a bigger version of a credit problem gets both of them wrong.
The everyday version: a food stall that supplies one office canteen on a monthly contract. The food is good, the customers are happy, and the contract is not renewed by somebody in a meeting the stall was not invited to. Nothing about the food was the problem.
The order in which things stop is worth being precise about. A lender whose funding has stopped cannot make new loans, so originationThe act of making a new loan. Origination is a flow. It stops the moment new money stops arriving, and the existing book carries on running down on its own schedule. is the first thing to go. The existing book keeps running down on its own schedule. New lending stops before existing lending does. A funding problem therefore shows up as a business that has gone quiet rather than as a business that has gone wrong.
Every repayment due to Rukmini Finance Limited is arriving on time and the lender is still in difficulty. What has happened?
How would somebody lending to this lender actually read it?
Seen from the other side of the facility: a party deciding whether to lend Rukmini Finance Limited money again has exactly the figures above. What does that party look at?
Not the interface, and not the growth. First, what is left after everything: 2.50 per cent of the book before tax, or Rs 450 crore, against the Rs 14,400 crore of borrowings that party is being asked to keep in place. Then the cost of doing the work at 3.00 per cent, the only one of the three cost lines that can be cut in a bad year. Then how much of the book is funded by the lender's own money. Rs 3,600 crore of net worth absorbs a loss before it reaches the party that funded it. A party lending to a lender is not buying the growth story; it is buying the distance between a loss and itself.
An analyst does the same arithmetic for a different reason, and a household does a smaller version of it every time it lends money to a relative starting a business. The questions are what the business earns, what it spends, and how much of the relative's own money is in it. The third question is the one that changes the answer.
Why does return on equity fail to separate two lenders that are nothing alike?
Rukmini Finance Limited earns Rs 337.50 crore after tax on a book of Rs 18,000 crore. The return on assetsHow much each rupee of what a business holds produced after tax across the period, settled before any question of how that holding was paid for arises. is 1.875 per cent, and it runs assets of 5.0 times net worth. The two multiplied give 9.375 per cent on net worth.
Suvarna Commercial Bank Limited earns Rs 2,250 crore after tax on total assets of Rs 2,40,000 crore. Its return on assets is 0.9375 per cent, and it runs assets of 10.0 times net worth. The two multiplied also give 9.375 per cent on net worth of Rs 24,000 crore.
The equality is constructed rather than discovered. An identical number arriving from two unlike businesses reads like a copying error until both limbs are laid beside it. Half the return per rupee of assets, twice the leverageAssets divided by the business's own net worth. Leverage says how many rupees of holdings each rupee of the owners' money is carrying, and it multiplies whatever the assets earn., same answer. Two businesses that are not remotely alike arrive at one number from opposite limbs, and the number cannot tell which is which.
So the rule that follows is short. A return on equityProfit after tax divided by the business's own net worth, for the same period. The ratio is the product of two separate readings, and it hides both of them. quoted without both limbs beside it has said nothing usable. Ask for the return on assets and the leverage every time, and if only one of the three numbers is offered, the other two are the ones worth chasing.
And the limit, said plainly. Neither way of running a lending business is the better one on this evidence. Higher leverage is not recklessness and a wider margin is not skill. One year of each, with no downturn in it and no failure, cannot settle which way of running a lending business works.
Two lenders both return 9.375 per cent on net worth for the year. What has to be seen before that number means anything?
Why the reported figures refuse to multiply back
One more thing about the reported figures, and it is the kind of thing that quietly costs an hour. The figures usually reported for these two lenders are rounded: a return on assets of 1.88 per cent and leverage of 5.0 times at the finance company, and 0.94 per cent and 10.0 times at the bank, with a return on net worth of 9.38 per cent at each.
The reported figures multiply to 9.40 per cent, not 9.38 per cent, at both of them. Nobody made an arithmetic mistake: 1.88 and 0.94 are already rounded, and the product only lands on 9.38 when the exact limbs, 1.875 and 0.9375, are used. Rounded limbs and an unrounded answer will not multiply back, and the gap between them is rounding rather than miscalculation.
The difference is 0.02 points. On the axis used higher up, at 75 user units to the percentage point, that gap would be 1.5 user units wide, thinner than the line around every box in those figures.
The reported 1.88 per cent multiplied by the reported 5.0 times gives 9.40 per cent, but the reported answer is 9.38 per cent. Who made the mistake?
Which numbers are missing here, and what shape is each absence?
A missing number is not one thing. Some absences still allow the answer to be bounded from above, some from below, and some allow nothing to be said at all. Naming which of the three is in front of the analyst is an output, not an apology. The name tells whoever reads the working what they can and cannot do next. The four absences here follow, each with the proof of its shape.
Fee income is a ceiling, and the ceiling is nil. This lender's stack closes exactly: Rs 1,386 crore of net interest income less Rs 540 crore for the work and Rs 396 crore for loans not repaid is Rs 450 crore, which is the stated figure before tax with nothing left over. Any fee income at all would have to appear somewhere in that subtraction, and it does not. So within this record, fee income is bounded above at nil. In lending generally a fee is an ordinary line. The nil ceiling is a fact about one set of figures, not about the business.
The number of parties funding the Rs 14,400 crore is a floor, and the floor is one. The borrowings are a positive amount, so at least one party lent them. There is no ceiling anywhere in the record, so the count could be one or two hundred. The floor is the half that matters: if it is one, the entire renewal decision belongs to a single meeting, and the failure drawn two sections above needs only that meeting to go the other way.
The capital held against the book is a floor by kind, and its size is decided elsewhere. A capital requirement is a minimum, so its shape is known before its value is: whatever this lender holds, it is at or above the line. The line itself, and the base it is struck on, are set by the Reserve Bank of India at rbi.org.in, and both move. A row that names the authority instead of a value survives every revision.
The split between what this lender originated for its own books and what it originated for anybody else is unknowable here, and so is any saving on the cost of doing the work. The split first: every figure here is struck on the book held, Rs 18,000 crore, and nothing in the arithmetic moves when the split moves, so no endpoint is derivable in either direction. The saving next: a saving is the difference between two observations and this record contains one, one lender for one year. There is no floor and no ceiling to state, and the honest output is the absence itself.
A control here would ordinarily move the share of new lending kept on the lender's own books against the capital needed behind it, and it is missing for an arithmetic reason rather than an editorial one. No such split can be derived from the figures, as set out just above, so every position of that control would be an invented figure wearing the clothes of an illustration.
The distinction is worth holding on to. The same absence elsewhere rests on a different ground entirely: sliding a number across somebody's access to credit teaches nothing the arithmetic does not. One refusal is about what cannot be known and the other is about what should not be dramatised, and they are not the same refusal.
The figures for Rukmini Finance Limited carry no fee income line at all. What is the strongest honest statement available about fee income here?
The failure: reading a lender off its front end
The mistake is made by an analyst or a writer rather than by anybody inside the business, and it runs in both directions. The first direction: this lender has an application, therefore it is a fintech lender, therefore it is a different kind of business from a finance company. The second is just as common and gets noticed less: this lender has no application, therefore it must be a traditional one.
Both readings use the channel to answer questions the channel does not touch. Where the money comes from, whose books the loan sits on and who carries the loss are settled on the funding side and in the agreements, and two businesses with identical screens can sit in different shapes entirely. Somebody who cannot separate them cannot say what either would do under strain. Saying that is the only reason to sort them at all.
The cost of the mistake is a view of a lending business built entirely on the layer that can be rebuilt in a quarter. The funding structure that decides everything takes years to change. The fix fits on one line and is worth memorising: whose books is the loan on, and who lent that party the money.
| The conditions on which a finance company is registered and may lend at all | Reserve Bank of India, rbi.org.in |
| The requirements on a party that sources, services or collects a loan for a lender | Reserve Bank of India, rbi.org.in |
| The requirements on a lender that puts part of its lending process outside itself | Reserve Bank of India, rbi.org.in |
| The conditions on a guarantee against first losses in a lending partnership, and any limit on it | Reserve Bank of India, rbi.org.in |
| The route the money must take between the lender's account and the borrower's, where a system such as Setu Payments Limited moves it | Reserve Bank of India, rbi.org.in |
| The capital a finance company holds, and the base that requirement is struck on | Reserve Bank of India, rbi.org.in |
Not one of the six is filled in. Each is set by the authority printed inside its own row, each of them is revised, and a figure written into any of them would be wrong rather than merely old on the day it changed.
The criterion by criterion comparison against a finance company is covered separately. The end to end digital lending process, and the step by step route from application to disbursal, are each covered separately. So is the arrangement in which a partner guarantees the first losses, and so is the way a lender funds a pool of loans before selling it. The comparison of a lender with deposits against a lender without them is covered separately.
How a decisioning model is built, validated, monitored or governed is covered separately.
Registration, the requirements on a party that sources, services or collects a loan, putting part of a lending process outside the lender, the conditions on a guarantee against first losses, the route the money takes between accounts and the capital a finance company holds all belong to the Reserve Bank of India, and the six rows above carry the name and rbi.org.in in place of the value.
What was read to write this, and what was deliberately not?
The six rows below say where each requirement is decided and then stop at the doorway. Each was confirmed at the site named beside it on the date shown.
| What is decided elsewhere | Decided by | Site | Confirmed |
|---|---|---|---|
| The conditions on which a finance company is registered and may lend at all | Reserve Bank of India | rbi.org.in | 23 August 2026 |
| The requirements on a party that sources, services or collects a loan for a lender | Reserve Bank of India | rbi.org.in | 23 August 2026 |
| The requirements on a lender that puts part of its lending process outside itself | Reserve Bank of India | rbi.org.in | 23 August 2026 |
| The conditions on a guarantee against first losses in a lending partnership, and any limit on it | Reserve Bank of India | rbi.org.in | 23 August 2026 |
| The route the money must take between the lender's account and the borrower's | Reserve Bank of India | rbi.org.in | 23 August 2026 |
| The capital a finance company holds, and the base that requirement is struck on | Reserve Bank of India | rbi.org.in | 23 August 2026 |
Rukmini Finance Limited, Suvarna Commercial Bank Limited and Setu Payments Limited are invented.
Educational material. Not advice on any investment, tax, budget or market position.