How to Analyse an NBFC Stock (NIM, GNPA, Credit Cost)

In July 2024, the managing director of one of India's most conservative lenders told analysts, in plain words, that he saw "no downside risk to our growth guidance" — and reaffirmed the full-year credit-cost guide of 2.2–2.4% he had stood behind for months (CreditAccess Grameen Q1 FY25 concall, 19 July 2024). One quarter later, at the Q2 FY25 call (25 October 2024), he nudged that guide up to roughly 4.5–5.0%. He was not lying, and he was not careless — this is a genuinely well-run company. He underestimated the cycle. FY26 credit cost came in at 6.74% (Inve data, Q4 FY26), roughly three times the original figure, and within two quarters of that reassurance — Q3 FY25, December 2024 — the lender printed a net loss of about ₹100 crore (Inve data, Q3 FY25), against roughly ₹353 crore of quarterly profit a year earlier (Q3 FY24, December 2023). (Illustration of how to read the numbers, not a view on the stock.)

The point is not that he failed. It is that the numbers which would betray his confidence were already building when he spoke — in the delinquency trend and the GNPA line, not the headline profit. A non-banking finance company is, at its core, a spread business: it borrows at one rate and lends at a higher one, and almost everything that can go wrong shows up first in either the spread or the quality of what it lent. Across the thousands of management commitments Inve tracks, a large share are quietly revised, ghosted, or missed rather than delivered as first guided — and finance companies are well represented among the names that guide confidently on growth and credit cost, then go quiet when the cycle turns.

This is how to read an NBFC the way a credit analyst does: the four or five numbers that decide the outcome, the concall questions that separate a clean book from a cosmetic one, and the one red flag that has buried more retail investors than any other.

A note on the boundary first: you will not fully model a loan book from the outside. What you can do is read the direction of the key numbers and check whether management's account survives the Q&A.

Why is an NBFC a spread business and not a "growth" business?

Strip an NBFC down and it is an arbitrage on money: it raises funds at a cost and lends them at a yield, and the difference, after setting aside for loans that go bad, is the entire economic engine. No factory, no order book — just a balance sheet, the income statement a thin slice off the top.

That framing fixes two beginner errors. First, revenue growth means almost nothing on its own — an NBFC can grow its loan book 40% by lending to riskier borrowers at the same rate, loading the balance sheet with future defaults that won't surface for six to eight quarters. Second, leverage is the business model, not a warning sign — a well-run NBFC runs assets at six to eight times equity, which is how a thin spread becomes a respectable return on equity. The question is never "is it leveraged?" but "is the spread real, is the book clean, and is there capital to absorb the loans that will go bad?"

What does net interest margin (NIM) actually tell you?

Net interest margin is the spread, as a percentage of the assets that earn it.

A microfinance lender might run a NIM above 10% (high rates to underbanked borrowers); a housing finance company 3–4% (low-risk, competitively priced home loans). There is no universal "good" NIM — it is a function of what the NBFC lends against. Listed names worth studying across these segments include Bajaj Finance (diversified consumer and SME), Shriram Finance and Cholamandalam (vehicle finance), Muthoot Finance (gold loans) and SBI Cards (cards) — each with a very different spread and risk profile. So read NIM two ways: against the company's own history, and against direct peers in the same segment.

Read NIM together with the cost of borrowing, because that is where pressure shows up first. When the rate cycle turns, borrowing costs reprice faster than lending yields — bonds and bank loans reset quickly, while a five-year vehicle loan stays at its old rate until it runs off. A NIM holding steady while system rates rise deserves a hard look: either genuine pricing power, or reaching for higher-risk borrowers to defend the headline. So watch the gap between yield and cost of funds across six to eight quarters, not the level — a spread quietly compressing 20–30 basis points a quarter while management calls margins "stable" is the slow leak the profit hides for a year.

How do you read credit cost and GNPA together?

This decides whether the spread you admired is real. Two numbers matter, and only read against each other.

Gross NPA (GNPA) is the stock of bad loans already on the books — the share where borrowers have stopped paying. NBFCs increasingly report it via staging: stage-1 is performing, stage-2 is loans showing stress (typically 30–90 days overdue), and stage-3 is the impaired bucket, the GNPA equivalent under the expected-credit-loss framework.

Credit cost is the flow — the provisions charged to profit this period to cover expected losses, as a percentage of the loan book: the price of bad lending, recognised now.

Here is the relationship that matters. GNPA tells you about the past; credit cost tells you about the present; the gap tells you whether the future is being acknowledged or postponed. An NBFC can keep reported GNPA low by writing off or restructuring bad loans while its credit cost climbs because it is provisioning hard for stress ahead — that combination is honest. The dangerous inverse: GNPA creeping up and credit cost held artificially low to protect this quarter's profit. Losses are then recognised slower than they form, and the catch-up lands in one brutal quarter.

Watch the early-delinquency buckets especially; they are the leading indicator. A bulge there — loans not yet impaired but slipping — is where the next two quarters of stage-3 and credit cost already sit. The same instinct that distrusts a suppressed credit cost is what separates reported profit from real cash earnings in any business.

Watch how this actually unfolded at CreditAccess Grameen, a microfinance lender that runs an unusually clean, conservative book — which is precisely why it makes the better teaching case. Honest companies get the cycle wrong too, and the warning lived in the staging, not the profit (illustration, not a view on the stock; GNPA/staging figures are as reported in the company's quarterly concalls and are approximate — confirm against the source transcript before relying on any single number):

Read the first two rows together. Even as management held its growth confidence — "no downside risk" — and reiterated the 2.2–2.4% credit-cost guide at the July 2024 call, the early-delinquency line was already turning: 60+ DPD had ticked up to 1.46%. The headline profit still held; the leading indicator had begun to move. By the next quarter the guide was lifted to 4.5–5.0%, the quarter after that to nearly 7%, and that same quarter the company posted a roughly ₹100 crore loss. The figure to distrust was never the profit — it was the gap between a creeping GNPA and a growth-and-credit-cost story still leaning on the original guide.

Why does the borrowing mix decide whether an NBFC survives a shock?

The 2018–19 IL&FS and DHFL episode taught a hard lesson: an NBFC does not usually die of bad loans. It dies of a funding mismatch — and no profitability ratio captures it. An NBFC funding long-tenure loans (a vehicle or home loan) with short-tenure paper (rolled over every three months) is betting it can always refinance. But the moment lenders get nervous — a sector scare, a rating downgrade, one peer blowing up — the paper does not roll over, and the NBFC owes money it cannot raise against assets it cannot quickly sell. That is an asset-liability mismatch (ALM), and it can kill a profitable, well-provisioned lender in weeks. For a non-lender, the equivalent funding warning lives in how to spot a debt-trap stock.

So the borrowing mix is a survival question, read on three axes:

• Tenure match. Are long assets funded with long liabilities? The ALM table shows the buckets.
• Source diversity. A mix of bank lines, long-tenure bonds, and retail deposits is sturdier than one leaning heavily on commercial paper.
• Cost trend. Rising incremental cost of borrowing is the market pricing risk before the financials do.

A conservative NBFC talks about its ALM and liquidity buffer unprompted; one that changes the subject is telling you where the bodies are.

What capital adequacy number should you actually look at?

Capital adequacy (the capital-to-risk-weighted-assets ratio, or CRAR) is the cushion — the equity that absorbs losses before lenders take a hit. Per RBI norms, there is a regulatory floor (15% for most deposit-taking and systemically important NBFCs, though the exact minimum varies by NBFC category and layer), but the floor is not the interesting part. Two things are:

The trend and the buffer. An NBFC running at 16% capital while growing its book 30% a year is heading for a capital raise — consuming its cushion faster than it generates it. That raise will likely dilute shareholders, and its timing (forced, in a bad market, versus opportunistic) tells you a lot about management foresight.

Tier-1 quality. Headline CRAR can be propped up with Tier-2 (subordinated debt), but the number that truly absorbs losses is Tier-1 (mostly equity) — a company leaning on Tier-2 to flatter its ratio is thinner than the headline suggests.

Two real books, side by side

The cleanest way to feel the difference is to put two lenders next to each other through the same twelve months — the FY25 microfinance stress cycle — and watch what the staging did while one management held its nerve and the other genuinely held its book. CreditAccess Grameen is the microfinance lender from the table above; Home First Finance is an affordable-housing lender. Different segments, deliberately — the lesson is in the direction of the numbers, not their level (illustration, not a view on either stock; figures from Inve data unless noted):

Here is the uncomfortable part, and the reason this pairing teaches more than a fraud would: CreditAccess was not thinly capitalised, not aggressive on accounting, not evasive on its calls. It carried a 26% capital adequacy ratio — far above the regulatory floor — and a reputation for conservative provisioning. The cushion was never the problem. A well-run, well-capitalised lender still walked its credit-cost guidance from 2.2% to nearly 7% in three quarters, because the thing that broke was the book, and the book spoke through the GNPA line a full quarter before the guidance admitted it. Home First, lending against a house instead of a borrower's word, simply did not have that quarter. The craft is learning to find the first column's story uncomfortable while its profit and its capital ratio still looked fine.

The concall questions that actually matter

The financials give you the what; the concall tests whether management's account of them holds up — the same discipline you'd bring to reading any concall transcript, sharpened for a lender. The questions worth listening for — and the answers worth distrusting:

• "What is the incremental cost of borrowing, and how is it trending?" The headline cost of funds lags; the incremental (newest borrowing) cost is the real-time signal. A precise number is comfort; "broadly stable" buys time.
• "What's happening in stage-2, and what's the collection efficiency in the latest month?" The leading-indicator question. Vague reassurance here, quarter after quarter, is itself the answer.
• "What's the ALM position and on-balance-sheet liquidity?" A confident NBFC has these ready; hesitation is a flag.
• "What credit cost are you guiding to for the full year, and what's it based on?" A hard number with a stated assumption you can hold them to; "we expect it to normalise" you cannot.
• "Where is the growth coming from — new geographies, new products, or deeper into existing borrowers?" New-product, new-geography growth is where underwriting slips.

Track these across quarters: a single call tells you about a quarter; a sequence tells you about management. Go back to CreditAccess. The full-year credit-cost guidance it set in March 2024 — "2.2% to 2.4%" — and the matching "23–24%" growth guidance did not survive the year: in Inve's Promise Tracker the credit-cost guide was revised within two quarters (to 4.5–5.0% at the October 2024 call, then higher again), and the growth guidance was eventually marked ghosted (Inve data), never reaffirmed once reality overtook it. Each successive credit-cost guide it set was overtaken in turn — the year ultimately came in at 6.74%. Notice what this is and is not. It is not evidence of bad management — the company was candid once the cycle turned, raised its guidance in real time, and provisioned hard. It is evidence that the first, confident number deserved less weight than the early-delinquency trend sitting under it. A sequence of guidance that only ever moves one way — down on growth, up on credit cost — is the tell no single call gives you. This is the pattern Inve's Promise Tracker is built to surface: every forward commitment pinned to the quarter it was made, with a verdict as later calls come in. (Illustration, not a view on the stock — and a read on how management communicated through one cycle, not a lifetime verdict.)

The one red flag that buries NBFC investors

If you remember one thing: the most dangerous NBFC is not the one with high reported GNPA — it's the one whose GNPA stays suspiciously flat while every leading indicator deteriorates.

A high, honestly rising GNPA is a known problem the market is already pricing. The killer is the cosmetic clean book: GNPA held flat by aggressive write-offs and restructuring, credit cost suppressed to protect quarterly profit, stage-2 quietly bulging, and incremental borrowing costs climbing — all while management calls asset quality "stable and improving." Each piece is defensible alone; together, they are an NBFC recognising losses slower than it creates them, and the reckoning arrives as one catastrophic quarter that wipes out years of gains. The number you can trust least in isolation is the one the market watches most.

Where this lens can be wrong. The strongest case against everything above is that CreditAccess proves it. A reader who had distrusted that "no downside risk" call in July 2024 would have been right — but a reader who always distrusts confident guidance from conservative lenders will miss the best ones, because the best lenders also sound confident, right up until a sector-wide shock nobody could underwrite from a transcript. CreditAccess did not break because management hid anything; it broke because a whole microfinance cycle turned — borrower over-leverage, a bad monsoon, election-quarter collection gaps — and a clean, well-capitalised book took the hit alongside the dirty ones. No amount of concall-reading would have told you the timing of that turn; the GNPA line told you stress was building, not that a loss quarter was one print away. So the honest claim is narrower than it looks: reading the staging against the guidance tells you when a management's confidence has outrun its own book. It does not tell you when the cycle turns, and it cannot move you out of a bad neighbourhood — only lower your odds of owning the worst house in it.

A hard limit worth restating: judging an NBFC means judging credit risk, which is genuinely hard. A clean concall and a conservative funding profile do not immunise an NBFC against a system-wide credit shock or a liquidity freeze that takes good and bad lenders down together — the 2018 IL&FS–DHFL episode took some genuinely well-run names with it, and the FY25 microfinance cycle took a 26%-capitalised lender to a loss quarter.

A repeatable workflow

1. Anchor on the spread. NIM, yield, and cost of funds across six to eight quarters — watch the gap, against same-segment peers.
2. Test the book. GNPA / stage-3 against credit cost, plus stage-2. A rising book with falling provisions is the pattern to fear.
3. Check survival. The ALM table and funding mix — long assets on short money is the risk no profit ratio shows.
4. Check the cushion. Capital adequacy trend and Tier-1 quality, against the growth rate.
5. Audit the commentary. Check the credit-cost and asset-quality guidance against what happened next.

Inve's KPI Screener lines up NIM, GNPA, and other operational metrics across finance companies — value, YoY/QoQ trend, and a data-confidence flag per number — so step 1 takes minutes, not an afternoon. And the concall summaries pull every forward commitment into one guidance table per quarter, with speaker and quote.

Frequently asked questions

The discipline comes down to refusing to be impressed by the headline. The balance sheet, not the profit line, is the business — and it speaks through the spread, the staging, the funding mix, and the cushion. So invert the question you bring to a lender's results. Don't ask "is this a good quarter?" Ask: if this management were quietly underwriting tomorrow's losses to flatter today's profit, what would the numbers look like — and does this book rule that out? A creeping GNPA under a frozen credit-cost guide does not rule it out; it is the pattern itself.

And the owner's question, the one to sit with before you buy a single share of any finance company: what must I believe about the next downturn — not this quarter's profit — for this lender to still be standing, still funded, and still compounding on the other side of it? If the honest answer leans on management's confidence rather than the book's own numbers, you have read the headline, not the business.

Related reading

• Capacity Utilisation: Pre-Capex Signal for Stocks
• How to Analyse a Cement Company (India)
• How to Analyse a Pharma Stock in India