How to Analyse a Housing Finance Company (HFC)

In the March 2026 quarter, PNB Housing Finance reported a negative credit cost — minus 0.19% of assets — and a gross NPA that had just dropped below 1% for the first time, to 0.93% (PNB Housing Q4 FY26 investor presentation). Read quickly, that looks like the cleanest book in the sector. Read slowly, it tells you something else: the company recovered ₹332 crore during FY26 from a pool of loans it had already written off years earlier (PNB Housing Q4 FY26 investor presentation), so the provision line ran in reverse. The "profit" in that quarter's credit line was the ghost of a past mistake being clawed back, not a measure of how good the new lending is. (Illustration of how to read the numbers, not a view on the stock.)

That is the whole craft of analysing a housing finance company in one number. An HFC is a spread business, like any lender — it borrows at one rate and lends against a house at a higher one — but it is the safest-looking spread business in the market, and that safety is exactly what makes the numbers easy to misread. A home loan is secured, long-dated, and historically low-loss, so the headline asset quality almost always looks pristine. The questions that decide whether you're holding a compounder or a slow-bleeding spread are buried one layer down: in the gap between spread and NIM, in the cost of funds, in the borrowing mix, and in how the company funds a 15-year asset.

This is how to read an HFC the way a credit analyst does — the handful of numbers that decide the outcome, where each one actually hides (several are nowhere near the income statement), and the one red flag that has quietly cost patient HFC holders years of compounding.

A boundary first, said plainly: you will not model a mortgage book from the outside. What you can do is read the direction of the spread, the funding, and the early-delinquency line — and check whether management's account survives the next two calls.

How does a housing finance company actually make money?

Picture a shopkeeper who borrows from three lenders at 7.5% and lends it onward, secured against the borrower's house, at 12.5%. The five-point gap is the gross spread. From it he pays his staff and rent (operating cost), sets aside a little for the rare borrower who stops paying (credit cost), and what's left is the profit. Nothing else is the business. There is no factory, no order book — just a balance sheet, with the income statement a thin slice off the top of it.

Two consequences follow, and both trip up beginners. First, AUM growth on its own tells you almost nothing. A housing financier can grow its loan book 25% a year by lending against weaker collateral, to thinner-file borrowers, at the same headline rate — loading the book with defaults that won't surface for two or three years. Growth is only worth admiring once you know the spread held and the new loans were underwritten as carefully as the old. Second, leverage is the model, not a warning. HomeFirst Finance ran average assets at roughly 4.0 times equity in FY26 (HomeFirst Q4 FY26 investor presentation); that is how a five-point spread becomes a high-teens return on equity. The question is never "is it leveraged?" — it's "is the spread real, is the collateral genuinely safe, and can it keep borrowing cheaply enough to keep the spread?"

Spread or NIM — which number is the company hiding behind?

This is the single most important distinction in HFC analysis, and the one most retail screens get wrong because they only carry one of the two.

Spread is the clean number: portfolio yield minus cost of borrowing. It is the actual gap on the money lent. HomeFirst's overall spread in Q4 FY26 was 5.2% — portfolio yield 13.1% against cost of borrowing 7.9% (HomeFirst Q4 FY26 investor presentation). AAVAS Financiers, a rural affordable lender, ran a spread of about 5.3% in Q3 FY26 (AAVAS Q3 FY26 concall). PNB Housing, dominated by lower-risk prime salaried borrowers, runs a much thinner spread — 2.12% in Q4 FY26 (PNB Housing Q4 FY26 investor presentation). None of those is "good" or "bad" in isolation: a prime book should earn a thin spread, an affordable rural book a fat one. You compare a company only to its own history and to direct peers in the same segment. Other listed housing financiers worth lining up against their own segment include LIC Housing Finance, Bajaj Housing Finance, Can Fin Homes and Aadhar Housing Finance.

NIM (net interest margin) is the number that flatters. It is net interest income over assets, and for many HFCs it quietly folds in income that has nothing to do with the spread on loans held — chiefly the upfront gain booked when a loan is sold down (assignment / co-lending). AAVAS reports a NIM around 8.0% (AAVAS Q3 FY26 concall) while its spread is ~5.3%; the gap is not magic, it is income-recognition. PNB Housing shows the same wedge from the other side: a 2.12% spread but a 3.69% NIM (PNB Housing Q4 FY26 investor presentation). When NIM is comfortably above spread and the gap is widening, ask where the extra came from before you applaud it — a NIM propped up by ever-larger loan sell-downs is a margin you don't fully own.

Where to find them: spread, yield and cost-of-borrowing are almost never on the face of the income statement. They live in the investor presentation's "spreads and margins" slide and in the concall. Treat that slide as the real P&L of an HFC.

What does cost of funds tell you that profit can't?

For a lender, the cost of borrowing is half the business, and it moves before the profit does. Watch two versions of it.

The reported cost of funds is the blended rate on the whole borrowing book — it lags, because old cheap bonds are still on it. The number that talks to you in real time is the incremental (or marginal) cost of borrowing — the rate on the newest money raised. HomeFirst's reported cost of borrowing was 7.9% in Q4 FY26, but its marginal cost that quarter was 7.6% (HomeFirst Q4 FY26 investor presentation) — the falling marginal rate is the rate cycle reaching the book before it reaches the spread. PNB Housing improved its reported cost of borrowing to 7.35% in Q4 FY26 from 7.50% the prior quarter (PNB Housing Q4 FY26 investor presentation).

Here is why this matters more for an HFC than for almost any other lender: home-loan yields are sticky and competitive — you can't reprice a customer's mortgage at will without losing them to a refinancing rival — so when the cost of funds rises, the spread compresses fast, and there is little the company can do about it quickly. A rising incremental cost of borrowing, with portfolio yield flat, is the spread leaking out the bottom a quarter or two before the profit shows it. So read the cost-of-funds trend, not the level.

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

The 2018–19 IL&FS and DHFL episode taught the lesson the hard way: a housing financier rarely dies of bad home loans — those default slowly and are secured. It dies of a funding mismatch. An HFC funds a 12-to-15-year mortgage; if it funds that long asset with short paper it must keep rolling over, it is betting it can always refinance. The day the market gets nervous — a sector scare, a downgrade, one peer blowing up — the paper stops rolling, and the company owes money it cannot raise against an asset it cannot quickly sell. That is asset-liability mismatch, and no profitability ratio captures it.

So the borrowing mix is a survival question, and HFCs have one funding source no other lender has: National Housing Bank (NHB) refinance — long-tenure, low-cost money from the regulator, the sturdiest liability on the book. Read the mix on three axes:

• Source diversity and tenure. A blend of bank term loans, NCDs, NHB refinance, and deposits is far sturdier than one leaning on commercial paper. PNB Housing's mix at 31 March 2026 ran term loans ~39%, deposits ~25%, NHB refinance ~14%, NCDs ~11%, ECBs ~8%, and commercial paper just ~3% (PNB Housing Q4 FY26 investor presentation). HomeFirst is even more pointed: it states it carries zero borrowing through commercial paper, with funding spread across 31 banks and institutions, public-sector banks ~36%, private banks ~22%, NHB refinance ~15% and NCDs ~16% at March 2026 (HomeFirst Q4 FY26 investor presentation). No short-paper dependence is a survival feature, not a footnote.
• The liquidity buffer. PNB Housing maintained an average daily liquidity coverage ratio of 145% in Q4 FY26 (PNB Housing Q4 FY26 investor presentation) — a confident HFC keeps this ready and unprompted.
• The cost trend. Rising incremental cost of borrowing is the market pricing the company's risk before the financials admit it.

A housing financier that talks about its NHB lines, its ALM buckets and its liquidity buffer without being asked is telling you it has nothing to hide there. One that changes the subject is telling you where to look. For a non-lender, the same funding instinct lives in how to spot a debt-trap stock.

How do you read GNPA, stage-3 and LTV together?

These three decide whether the safe-looking collateral is genuinely safe.

GNPA / Gross Stage 3 is the stock of loans already gone bad. For HFCs it sits remarkably low — secured, owner-occupied homes default rarely. HomeFirst's gross stage-3 was 1.8% at March 2026 (HomeFirst Q4 FY26 investor presentation); AAVAS ran a GNPA of 1.19% in Q3 FY26 (AAVAS Q3 FY26 concall); PNB Housing's retail GNPA fell to 0.93% in Q4 FY26 (PNB Housing Q4 FY26 investor presentation). Because the level is always low, the direction and the early buckets matter far more than the headline. Watch the 1+ and 30+ DPD lines in the deck — they lead stage-3 by a couple of quarters. HomeFirst's own commentary makes the point: 30+ DPD improved 50bps QoQ to 3.2% and that drove GNPA down to 1.8% (HomeFirst Q4 FY26 investor presentation). The early bucket moved first; the headline followed.

LTV (loan-to-value) is the HFC-specific cushion that GNPA can't show you. It is the loan as a percentage of the home's value — the equity the borrower has in their own house, and therefore your protection if you ever have to sell the collateral. A book originated at conservative LTV can carry a higher GNPA safely, because each defaulted loan is over-collateralised. HomeFirst originates at low LTV and reports an effective LTV on its live book where roughly 48% of loans sit below 50% LTV (HomeFirst Q4 FY26 investor presentation) — meaning for most of the book, the house is worth more than twice the outstanding loan. AAVAS historically ran portfolio LTV around 51% (AAVAS FY19 investor presentation — a FY19-vintage figure that may have shifted since, included only as a directional reference). LTV almost never appears on the income statement; it lives in the investor deck's asset-quality slides. A book with low GNPA and low LTV is genuinely safe; low GNPA stacked on creeping LTV is safety borrowed against the next downturn in property prices.

Read the three together: GNPA is the past, the DPD buckets are the near future, and LTV is how much a bad future actually costs you. A clean GNPA on a high-LTV, fast-growing book is a different animal from a clean GNPA on a low-LTV, seasoned one — and the screen shows them as identical.

What about opex-to-AUM and disbursement growth?

Two operating numbers round out the picture, and they pull in opposite directions — which is the tension worth watching.

Opex-to-AUM (or opex-to-assets) is the efficiency of the lending machine. Affordable-housing lenders run higher cost ratios than prime ones — smaller-ticket loans in smaller towns cost more to source and collect. AAVAS runs opex-to-assets around 3.5% (AAVAS Q2 FY26 concall), with management guiding it down toward 3.0% over time; HomeFirst, with a more tech-led model, guides to a tighter 2.7–2.8% (HomeFirst concall guidance). The number to watch is the trend as the book scales: a genuine franchise sees opex-to-AUM fall as fixed costs spread over a bigger book. If it's flat or rising while AUM grows fast, the company is buying growth by adding cost — efficiency that never arrives.

Disbursement growth is the leading indicator of AUM growth two quarters out (today's disbursements become tomorrow's book, net of run-off). HomeFirst disbursed an all-time-high ₹1,572 crore in Q4 FY26, up 23.5% YoY (HomeFirst Q4 FY26 investor presentation). But disbursement growth is also where underwriting discipline is most easily abandoned — a sudden surge into new geographies or new products is precisely where the next cycle's bad loans get written. Fast disbursement growth is worth applauding only alongside a steady spread, a steady DPD line, and a stable LTV. Growth that comes with a softening on any of those three is the book being lent into a future problem.

How do you value a housing finance company?

You do not value an HFC on earnings multiples the way you would a manufacturer, for the same reason you don't value a bank that way: the balance sheet is the business, and a price-to-earnings ratio ignores the quality and the risk in that balance sheet. The right lens is price-to-book against return on equity — what you pay for a rupee of the lender's net worth, set against how much that net worth earns each year.

The logic is simple and worth holding onto: a lender that consistently earns, say, an 18% ROE deserves a higher multiple of its book value than one earning 12%, because each rupee of equity compounds faster and can fund more lending. So a 3x book multiple on a sustained-high-ROE, low-LTV, clean-funded HFC can be cheaper in substance than a 1.5x multiple on a lender whose ROE is propped up by aggressive leverage or assignment income that won't repeat. The multiple is the easy part; the work is deciding whether the ROE behind it is real and durable.

Two adjustments specific to HFCs. First, strip the one-offs out of the ROE before you capitalise it. PNB Housing's recent return on assets was lifted by recoveries from its written-off corporate pool — minus-credit-cost quarters (PNB Housing Q4 FY26 investor presentation) — and a one-off recovery is not a steady-state earning. Capitalise the through-cycle ROE, not the flattered one. Second, price the ALM and borrowing-mix risk into the multiple, not just the growth. Two HFCs growing at the same rate with the same ROE are not worth the same multiple if one funds long assets with diversified long money and zero commercial paper, and the other leans on short paper it must keep rolling. The market pays up for funding sturdiness because that is what survives the shock — and the cheapest-looking HFC is sometimes cheap precisely because its liabilities are fragile. The same discipline of separating reported profit from durable cash earnings runs through quality of earnings, and the lender's-eye view of the spread and the book carries over from how to analyse an NBFC.

A real book, read closely

Take HomeFirst Finance through its own FY26 numbers — an affordable-housing lender, chosen because it is candid and well-run, which makes it the better teaching case than a blow-up would. (Illustration, not a view on the stock; figures from the HomeFirst Q4 FY26 investor presentation.)

Now the said-versus-did, because a single quarter is a snapshot and a sequence is the management. In its Q4 FY26 letter, HomeFirst guided to "~25% YoY AUM growth" for FY27 and reaffirmed credit-cost guidance of "30–40bps even as we scale" (HomeFirst Q4 FY26 investor presentation). Those are exactly the two commitments to write down and check against the next four calls — not because you distrust them, but because the gap between a confident growth-and-credit-cost guide and the DPD line underneath it is where the signal always lives. Across the thousands of management commitments Inve tracks, a meaningful share are quietly revised or go silent rather than landing as first guided — and lenders are well represented among names that guide confidently, then change the subject when the cycle turns. Inve's Promise Tracker pins each forward commitment to the quarter and quote it was made in, so you watch the sequence rather than re-reading every transcript by hand. (Illustration, not a view on the stock — and a read on how management communicates now, not a lifetime verdict.)

The red flags specific to a housing financier

Each is defensible alone; the danger is when several arrive together.

• NIM rising while spread is flat or falling. The extra margin is coming from loan sell-downs (assignment / co-lending), not from the book you own. A widening NIM-over-spread gap is income you don't fully control dressed up as core profitability.
• Spread compressing while management calls margins "stable." Cost of funds reprices faster than sticky home-loan yields. A spread leaking 15–20bps a quarter under a "stable margins" commentary is the slow bleed the profit hides for a year.
• AUM and disbursement growth accelerating while LTV creeps up or DPD buckets bulge. Growth bought by easing underwriting. The clean GNPA is a lagging number; the LTV and the early buckets are the leading ones.
• A flattering ROE built on one-offs or thin Tier-2 capital. Recoveries from old write-offs, assignment gains, or a capital ratio leaning on subordinated debt rather than equity — capitalise the steady-state, not the flattered quarter.
• Commercial-paper-heavy funding and silence on ALM. The IL&FS lesson. Short paper funding a 15-year asset is the mismatch that kills profitable lenders in weeks; an HFC that won't discuss its ALM unprompted is the one to distrust.

Where this lens can be wrong. The strongest case against everything above is that a low-LTV, NHB-funded, clean-GNPA HFC can still underperform for years — not because the book broke, but because a thin-spread prime lender in a competitive market simply can't earn a high enough ROE to justify its multiple, however safe it is. Safety and returns are not the same thing, and the discipline that keeps you out of the fragile HFC can also anchor you to a sturdy-but-mediocre one. And the inverse: an analyst who always distrusts confident guidance from conservative HFCs will miss the best compounders, because the best ones also sound confident — right up until a property-price downturn or a funding freeze nobody could underwrite from a transcript takes good and bad lenders down together. Reading the spread against the funding tells you when a management's confidence has outrun its own balance sheet. It does not tell you when the housing 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 HFC means judging credit and funding risk, which is genuinely hard from the outside. A clean concall and a conservative funding profile do not immunise a housing financier against a system-wide liquidity freeze. The 2018 IL&FS–DHFL episode took genuinely well-run names down with the rest.

A repeatable workflow

1. Anchor on the spread, not the NIM. Portfolio yield minus cost of borrowing, across six to eight quarters, against same-segment peers. Then ask why NIM differs from spread.
2. Read the cost of funds in real time. Incremental cost of borrowing and its trend — the spread's pressure shows here first.
3. Test the collateral. GNPA and the early DPD buckets and LTV, together. A clean book on creeping LTV is borrowed safety.
4. Check survival. Borrowing mix, NHB lines, commercial-paper dependence and the liquidity buffer — long assets on short money is the risk no profit ratio shows.
5. Watch the operating leverage. Opex-to-AUM falling as the book scales; disbursement growth that doesn't come at the cost of spread, DPD or LTV.
6. Value on P/B vs a clean ROE. Strip one-offs and assignment income from the ROE before you capitalise it; price the funding risk into the multiple.
7. Audit the commentary. Check the spread, credit-cost and growth guidance against what actually happened next.

Inve's KPI Screener lines up spread, NIM, GNPA, cost of funds and opex-to-AUM across housing financiers — value, trend, and a data-confidence flag per number — so steps 1–5 take minutes, not an afternoon. 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 safest-looking line. A home loan is secured and slow to default, so the headline GNPA will almost always look pristine — which is exactly why the real signal sits one layer down, in the spread you actually own, the cost of funds heading toward it, the LTV cushion under the collateral, and the maturity of the money funding a 15-year asset. So invert the question you bring to an HFC's results. Don't ask "is this a clean book?" Ask: if this lender were quietly thinning its spread, easing its underwriting, or funding long assets with short money to keep growth going, what would the numbers look like — and does this balance sheet rule that out?

And the owner's question, the one to sit with before you buy a single share of any housing financier: what must I believe about the next funding squeeze and the next property cycle — not this quarter's clean GNPA — for this lender to still be funded, still earning its spread, and still compounding on the other side of it? If the honest answer leans on management's confidence rather than the balance sheet's own numbers, you've read the headline, not the business.

Related reading

• How to Analyse a Bank Stock (NIM, CASA, GNPA, ROA)
• How to Analyse a Building Products Company (Pipes, Tiles)
• How to Analyse a Cables and Wires Stock