Business

Bajaj Finance — Know the Business

Bajaj Finance is a spread-lender wrapped in a consumer-tech platform: it borrows AAA-rated in the wholesale market, lends across 32 retail credit lines, and reuses each customer four to six times through an in-house EMI card (94.4 MM cards in force), an 86.6 million-user app, and a ~242,000-point active distribution network. The market pays roughly 5x book because this franchise consistently prints ~18–20% ROE on ~4.3–4.6% ROA — about double the typical NBFC. The consensus-versus-reality gap is not the quality of the business but the price for it: at 31x P/E and a P/B above 5x, the stock is priced for the cross-sell flywheel to keep widening the lead it already has, leaving no margin of safety if asset quality, regulation, or competition compresses ROE by even 200–300 bps.

1. How This Business Actually Works

Takeaway. Bajaj Finance makes money the same way every NBFC does — by earning a spread between cheap wholesale funding and higher-yielding retail loans — but turns that arithmetic into roughly double the typical NBFC ROA through three engines: a low-cost-to-serve digital franchise, a self-feeding cross-sell loop that uses each acquired customer four to six times, and a diversified product book that smooths the credit cycle. It is closer in economics to a private-sector retail bank than to a traditional finance company.

The economic engine collapses into four-step arithmetic. The spread is what is earned; the cost-to-serve and cost-of-risk are what is given back; everything else is leverage.

The cross-sell flywheel is what turns this from "a good NBFC" into a franchise the market pays a quality premium for. Bajaj acquires a customer cheaply at the point of sale — an EMI on a refrigerator, a two-wheeler loan, a gold loan — captures their bureau data and bank-statement footprint inside its proprietary EMI card, and then over the next 24–36 months sells personal loans, business loans, insurance, fixed deposits, broking accounts, and consumer durables EMIs back to the same wallet. Of the 52.5 million loans booked in FY26, more than half went to existing customers, where credit cost runs materially below sourced-cold credit. Products-per-customer is the silent compounder behind the headline ROE.

Bargaining power runs in only one direction: the wholesale funders own the leash. Banks and the bond market set BFL's cost of funds, and they can withdraw in a quarter — IL&FS in 2018 made that vivid for the entire sector. BFL has neutralised most of that risk by being AAA-rated and structurally diversifying its funding (no single bucket above 48%, 16% as retail deposits that don't run), but it cannot be ignored. Borrowers in retail are price-takers within their band; dealers extract finance-sourcing rents in consumer durables and auto; the RBI sits above all of it.

2. The Playing Field

Takeaway. Bajaj Finance is in a peer group of one: it is the only listed NBFC where every metric — scale, asset quality, return, valuation — is simultaneously at the top of the cohort. Its market cap exceeds the next three peers combined (Shriram + Chola + Muthoot), and its 1.01% gross NPA is roughly one-quarter of the diversified-NBFC average. The right comparison set is not other NBFCs but the private-sector retail banks (HDFC Bank, ICICI, Kotak) whose cost of funds it cannot match and whose digital underwriting speed it routinely beats.

Source: peer_valuations.json (15-May-2026); BFL Q4 FY26 transcript; peer ratios from Screener.in snapshots. Tata Capital ROE pre-scale; not GNPA-comparable yet.

What the peer set reveals: Cholamandalam delivers a similar ROE to BFL and trades at a similar book multiple, validating that the rerating is earned, not idiosyncratic. Muthoot earns higher ROE but trades lower because gold is a single asset class with regulatory tail risk. Shriram and LTF earn solid ROEs but cannot match BFL on either growth or asset quality. The two real questions for the rest of the report are: (i) is the gap between BFL and its closest peer (Chola) about to widen or compress, and (ii) is a 5x book multiple defensible when the largest peer trades at 3.6x and earns within 3 ppt of BFL's ROE.

3. Is This Business Cyclical?

Takeaway. Bajaj Finance is moderately cyclical, but the cycle hits in three distinct places — funding access, asset quality on the unsecured book, and AUM growth — and almost never simultaneously. The company has navigated four sector tests in a decade (demonetisation, IL&FS, COVID, the RBI unsecured-risk-weight hike) without a single year of GAAP loss; ROE has dipped but never collapsed. The pattern: credit cost moves first and fastest, growth and margin lag by 2–3 quarters, then both reset to a higher base because the weaker NBFCs lose share to the stronger ones in every downturn.

Where the cycle actually hits, ranked by speed:

The lesson from FY21 — the only year ROE compressed to 13% in the past decade — is that BFL's downside cycle is provision-driven, not solvency-driven. The book mark went down, but capital adequacy never fell below 25%, the deposit base did not run, and the AAA rating held. Two years later FY23 ROE rebounded to 23.5% because the very same diversification that diluted upside in FY15–18 cushioned the downside in FY21 and accelerated the snapback. BFL's worst year is most NBFCs' average year.

4. The Metrics That Actually Matter

Takeaway. Reading an NBFC starts with five numbers, not the income statement. The order is: AUM growth (is the engine compounding?), credit cost (is it earning the spread or rebuying it back?), opex-to-NTI (is operating leverage real?), CRAR (is the capital base running ahead of risk?), and the ratio of NIM to cost of funds (is the spread defensible?). Headline EPS comes out the back end of these five; if all five are good, EPS takes care of itself.

Two of these are widely tracked but routinely misread.

Credit cost beats GNPA as the alpha indicator. GNPA is a stock measure that depends on classification policy and write-off practice; credit cost (provisions divided by average AUM) is a flow measure that captures forward-looking expected loss under Ind AS 109. A lender can flatter GNPA by aggressive write-offs while bleeding underlying credit, but cannot flatter credit cost for more than two quarters. The fact that BFL's credit cost compressed from 2.17% in FY25 to a guided 1.45–1.60% in FY27, while peers' credit costs are normalising upward, is the cleanest evidence that the underwriting franchise is earning its premium.

Opex/NTI is the closest thing to an operating-leverage lever. Banks measure this as cost-to-income and prize anything under 40%. BFL is at 33.8% and guiding 25–40 bps of further improvement annually as AI bots, India Stack, and digital servicing replace branch headcount. Every 100 bps of opex compression delivers roughly 30 bps of ROA — the silent compounder behind the headline ROE, since the spread itself has narrowed slightly over the past five years.

5. What Is This Business Worth?

Takeaway. Bajaj Finance is best valued as one economic engine — price-to-book pinned to through-cycle ROE — not as a sum of parts. Even though BHFL is separately listed and Bajaj Financial Securities exists as a wholly-owned subsidiary, both are consolidated, and the cross-sell flywheel that delivers the premium ROE is inseparable from the parent's customer franchise. The question reduces to a single judgment: at what multiple of book does 19–20% ROE compound to justify the price, and how durable is that ROE under three plausible pressures (regulation, competition, credit normalisation).

The first five rows justify the premium; the last three are why it cannot widen much further. The relationship between ROE and P/B in this peer set is roughly linear — at 19% ROE, the market is paying around 5x book; if ROE compressed to 16% the regression line implies ~4x; if it stayed at 19% and grew the book at 22% for three more years, the same multiple would compound to a 60% return without any rerating. The whole investment case sits inside that one regression.

Why this is not a sum-of-the-parts story. BHFL (88.75% owned, listed September 2024) and Bajaj Financial Securities (100% owned, ₹139 cr PAT in FY25) look like they invite SOTP, but the consolidated equity that the market is already valuing at 5.2x book already includes the 88.75% economic interest in BHFL — the 11.25% minority stake is netted out as minority interest. The cross-sell franchise that drives the parent's ROE is itself inseparable from BHFL's mortgage origination flow and from BFSL's broking distribution. Carving them out would double-count the franchise. SOTP becomes relevant only if regulation eventually forces a holdco-style restructuring under SBR — a tail risk worth tracking but not the base case.

6. What I'd Tell a Young Analyst

Watch credit cost, not GNPA. Credit cost moves first; GNPA confirms a quarter later. The FY27 guide of 1.45–1.60% is the single most important number in this stock — landing at 1.55% keeps ROA above 4.4%; drifting to 2.0% (i.e., MSME does not heal by Q2 FY27 as management expects, or unsecured re-escalates) strips ~60 bps of ROA and the 5x book multiple loses its anchor. Track Stage 2 and Stage 3 buckets disclosed quarterly; they lead the headline ratio.

The cross-sell flywheel is the real moat — and it is hard to verify from outside. Look for two proxies: products-per-customer (the company does not disclose it cleanly but you can infer it from loans booked divided by net new customers) and the ratio of repeat-customer disbursals to total disbursals. The day the proxy stops compounding is the day to recheck the multiple.

The biggest risk the market may be underestimating is regulatory drift, not credit. RBI's Upper-Layer framework is converging NBFCs toward bank-like rules. BHFL's mandatory IPO in 2024 was one consequence. Any future circular on co-lending economics, securitisation, or LTV could compress ROE by 100–200 bps without warning, and the multiple compression is non-linear. Read every RBI bi-monthly press release.

The bull case the market may be underestimating is operating leverage. Opex/NTI dropped from 58% to 33.8% over 15 years, and management is guiding a further 25–40 bps annually as AI agents and FINAI deployment replace bot-eligible work. Hitting 30% by FY29 alone delivers ~100 bps of ROA — enough to push ROE to 22%+ even with flat NIMs. This is the asymmetry that justifies a 5-year not a 1-year horizon.

What would genuinely change the thesis. (i) Two consecutive quarters of stage-2 deterioration in unsecured + MSME without offsetting recovery — the quality premium evaporates. (ii) An RBI circular that compresses cross-sell economics (e.g., capping co-branded card revenue share, or applying bank-style provisioning to retail BNPL). (iii) The cost-of-funds advantage closing as rate-cycle tailwinds fade and HDFC Bank / ICICI ramp small-ticket retail. (iv) Promoter-group restructuring or Bajaj Finserv-level capital action that affects the parent–subsidiary relationship. Anything else is noise around a high-quality compounder priced for that quality to persist.