FULL PUBLISHED RESEARCHEarnings and management commentary analysis
SAGE PRO Institutional Equity Research | 16 November 2026 | NSE: BAJFINANCE | Q2 FY2027, quarter ended 30 September 2026
Investment stance: CONSTRUCTIVE WATCH / valuation-sensitive. Research horizon: 12–24 months. Reference close: ₹1,117 on 13 November 2026 (Sage PRO canonical price history). Financial amounts are ₹ crore unless stated otherwise. Primary financial basis: consolidated issuer investor presentation; PAT in the headline panel is before minority interest, not earnings attributable to ordinary shareholders. Scenario figures are analyst illustrations, not guidance or targets. This is available-source institutional research; statutory equity/share-count and independently matched peer-valuation reconciliation remain open.
Chapter 1 — Investment decision: can a very large lender retain the economics of a specialist?
Key takeaways
- Bajaj Finance is no longer a small consumer lender whose growth can be judged by branch count or loan origination alone.
- The company combines a national acquisition network, granular underwriting, a growing cross-sell franchise, secured and unsecured credit and increasingly automated servicing.
- Its strategic advantage is the ability to reuse the customer relationship: transaction finance can create a low-cost acquisition channel, then risk data, repayment history and appropriate follow-on credit can increase lifetime value.
The same advantage becomes a liability if sales incentives encourage excessive borrower leverage or if rapid growth obscures vintage deterioration. Investors should buy sustainable risk-adjusted earnings per share, not gross loan volume.
The September quarter offers unusually strong evidence: AUM reached 584,624 versus 462,261 a year earlier (+26.5% calculated); quarterly net AUM addition was 37,680. Net interest income rose to 13,335 (+24%) and pre-provisioning operating profit to 10,699 (+23%). Reported pre-tax profit rose to 8,537 (+29%), and consolidated PAT before minority interest was 6,367 (+29%). Annualised ROA was 4.6% and ROE 20.6%, against 4.5% and 19.1% a year earlier. [A, panels 4–7]
Our differentiated view is that the *quality* of the next rupee of credit is now more important than the scale of the platform. The market should distinguish a genuine improvement in underwriting from a temporary credit-cost benefit when stressed legacy books run off.
The captive two-wheeler book is shrinking, MSME originations have tightened, and gold loans are expanding quickly. All three can improve consolidated earnings while changing risk composition. We favour the franchise if these shifts deliver durable returns without requiring a continuously higher capital base or softer credit standards.
Counter-thesis. A benign credit cycle, lower funding costs and fast secured-loan growth could inflate near-term profit and encourage an aggressive valuation. A return to more normal unsecured loss rates, weaker new vintages or a prolonged increase in capital could reduce earnings per share even as aggregate AUM expands. The proposed capital raise makes this distinction particularly material.
Decision: maintain a constructive research watch, do not infer a buy recommendation from headline growth alone. Require proof of sustained loss-adjusted profitability and share-level economics after the proposed capital issue. A high-quality company can still be a poor investment at an excessive entry multiple.
Chapter 2 — Financial diagnosis: earnings are outgrowing the book, but not without overlays
Quarter-on-quarter and year-on-year scorecard
Key takeaways
- Sources [A,C]. *Prior-year PBT and PAT inferred from rounded issuer growth, displayed approximately rather than represented as statutory exact figures.
- AUF = average assets under financing; AUM and balance-sheet loans are different measures.
Earnings bridge
The quarter's net total income of 15,974 less operating costs of roughly 5,271 (33% of net total income) produces about 10,703 pre-provision operating profit, close to reported 10,699. Less 2,166 of loan losses/provisions gives approximately 8,533 PBT versus 8,537 reported; minor difference reflects rounded presentation figures/other line items. This is a better analytical lens than EBITDA: interest cost is a core operating expense for a lender.
The 337 prudential macro overlay is central to interpreting the quarter. Excluding it, credit provisions would have been 1,829, 12% lower than the prior year's 2,080. PBT before overlay was 8,874 and PAT before overlay 6,619. The overlay represents about 3.8% of pre-overlay PBT. It is not cash already lost nor a guaranteed future profit reversal. An institutional model should retain a reasonable normalized loss assumption rather than add the overlay back indefinitely.
Net interest income expanded roughly 23.6% while AUM expanded 26.5%; growth in closing AUM does not translate mechanically into quarterly income because new disbursements arrive throughout the period. The more instructive measure is margin on average assets and product mix.
Operating expenses as a share of net total income remained virtually flat at 33.0%, compared with 33.1% a year earlier, despite investment in branches, staff and AI. There is no evidence yet of a step-change in the cost ratio; the business is producing scale without a large operating leverage dividend.
Analyst interpretation: earnings growth is currently supported by higher volumes and lower underlying loss intensity, rather than simply a dramatic efficiency gain. A more normalized credit environment would be a better test of the franchise's incremental earnings power.
Chapter 3 — The customer franchise: breadth, productivity and the risk of rapid cross-sell
Key takeaways
- The quarter ended with 128.85 million customers, including 82.36 million in the cross-sell franchise.
- The ratio is about 63.9%, not the proportion holding multiple loans.
- New customers added were 4.42 million and 13.45 million new loans were booked, compared with 12.17 million a year ago.
The issuer cautions that festival timing makes the annual volume comparison noisy. The company expects 60–62 million new loans and 18–20 million new customer additions in FY27. These are management targets, not achieved outcomes. [A, panel 5]
The economics of this network should be analysed in three stages: acquisition, underwriting and monetisation. Dealer or transaction-based acquisition makes it possible to observe customer repayment over time. Repeated use can improve acquisition efficiency and pricing accuracy, but a large cross-sell pool is not a promise of profitable incremental credit. The correct unit is cohort contribution after credit losses, funding, service costs and allocated capital.
Portfolio mix is the next source of alpha
Management reported strong growth in urban consumer finance (43%) and rural consumer finance (63%), rapid expansion in CV/tractor and MFI off small bases, and 42% growth in loans against securities. MSME grew only 3%, reflecting a more cautious stance, while the legacy captive book had reduced to around 2,400. Gold loans exceeded 24,000, roughly 4.1% of group AUM; the company added 321 gold-loan branches in the quarter and ended with 2,022. [B, pp. 2–3]
The expansion in gold loans matters for two reasons. First, it brings collateral-backed credit and a branch-based distribution network to a diversified lender; that may reduce exposure to unsecured consumption shocks. Second, low collateral loss does not imply zero operational risk. Fraud, inaccurate valuation, auction procedures, LTV compliance, interest servicing and borrower behaviour all affect the realized economics.
Gold and microfinance expansion also demand investment before cohorts mature. Management targets 2,700–2,800 gold branches and 520–550 MFI branches by March 2027. An investor should therefore track AUM per mature branch, branch-level productivity, operating expense versus contribution, and comparable vintage delinquencies. The network's scale alone is not evidence of returns.
FINAI: focus on measured outcomes rather than bot counts
The issuer reports AI-led service penetration of 75% of do-it-yourself customer servicing, compared with 65% a year earlier, and disbursements from voice/text AI leads of 2,744 versus 1,295. Its dedicated AI unit reached 328 people, and the presentation records 130 agents and 65 agentic deployments. [A, panels 9–12] These are adoption metrics. They are not evidence that AI has already created incremental net profit of a specific amount.
The important experiment disclosed on the call is a planned first risk-model cohort of approximately 500,000 B2B accounts, with performance judged at three months on book before broader rollout. Management's modeled 15–20% improvement in risk outcomes is a testable hypothesis, not realized experience.
[B, p. 4] The investor should demand matched approval rate, fraud loss, subsequent delinquency and contribution per approved borrower. A model which increases approval rates but produces weaker 6–9 month credit outcomes would destroy value even if customer conversion improves.
Chapter 4 — Credit quality: the gold-loan stage-2 shock and the hidden importance of denominator growth
Key takeaways
- Headline GNPA fell to 1.06% from 1.24% YoY; NNPA to 0.42% from 0.60%.
- Stage-3 provision coverage was 60%.
- These are favorable, but the quarter's stage-2 and stage-3 assets increased by 1,601, including 1,047 from gold loans.
Combined stage-2/3 assets increased from 1.87% to 2.02% of loans. Management explained that short-tenor schemes introduced following new rules created rollover friction and that the company stopped offering these from August. Adjusted for that transient gold-loan effect, issuer-reported combined stage-2/3 improved to 1.88% from 1.90%. [A, panel 6; B, pp. 3–6]
The key insight is that a technical delinquency can be a customer-experience problem before it becomes a loss. A borrower required to repay or roll over every three months can miss the operational deadline despite adequate gold collateral.
Longer-tenor bullet products with flexible interest servicing may improve regularization without changing the underlying credit demand. Management said approximately 550 crore had already returned to stage 1 by the call.
This observation supports the technical explanation, but it is not proof that the entire cohort has cured or that auctions and recovery costs are immaterial.
Two independent tests should follow: first, track the residual short-tenor cohort through the next quarter, including stage migration, auction incidence and realized loss; second, verify that the newly originated longer-tenor cohort does not create a maturity concentration twelve months later. The existence of collateral changes loss given default, not the probability of operational delinquency.
The Q&A also addressed MSME asset quality. Management attributed elevated GNPA partly to weak denominator growth and pointed to better 3/6/9/11-month vintages after origination changes. [B, pp. 6–7] That argument is plausible, not sufficient.
If a loan book is flat, an unchanged stock of old NPAs produces a higher ratio; rapid new originations can subsequently lower that ratio mechanically. Test the absolute stock of delinquent loans, cure and write-off flows, and comparable cohorts—not just percentage GNPA.
Management discussed cutting the share of originations to customers with multiple personal loans from 14% to 7%, with a target of 3.5% over six months. The strategic implication is more selective unsecured growth. The trade-off is between near-term volume and better loss-adjusted returns. We regard this as a more useful leading indicator than management tone alone.
Credit-cost sensitivity
A 25bp change in annual credit cost applied illustratively to September AUM of 584,624 would represent about 1,462 crore annual pre-tax sensitivity. This is an order-of-magnitude illustration, not a reported AUF calculation; exact sensitivity must use the correct average financed asset denominator. Relative to Q2 annualized PBT of 34,148, the illustrated amount is ~4.3%. The investment debate should therefore focus on credit-cost normalization, not only revenue growth.
Chapter 5 — Funding, leverage and capital: the market should not confuse liquidity with equity returns
Key takeaways
- The issuer reports liquidity buffer of 19,995, cost of funds 7.31% (9bp lower sequentially), deposits of 69,754 (14% of consolidated borrowings), capital adequacy 20.33%, Tier 1 19.43%, and consolidated leverage 5.1x.
- [A, panels 5,7] These indicators provide meaningful funding flexibility.
- Deposits diversify the liability book but are not identical to stable bank CASA funding, and they are sensitive to competing interest rates.
A prudent lender must manage maturity gaps, interest-rate repricing and committed liquidity, not just report a large absolute liquidity buffer.
On 1 October the board approved proposed capital raising up to 11,700 through a QIP and 5,800 through preferential warrants to Bajaj Finserv, subject to applicable approvals. The combined announced ceiling is 17,500.
[A, panel 8; B, p. 4] These are not assumed completed proceeds. The warrant conversion terms, timing, equity issue price and actual share count are material to per-share valuation. Management frames capital as growth and resilience capacity rather than an emergency response.
That is consistent with the reported capital ratios, but capital can be expensive to shareholders if new deployment earns below the existing franchise's return.
Capital allocation test: a 20.6% annualized current ROE does not guarantee that incremental equity earns the same return immediately. If capital is issued before deployment, denominator growth can dilute near-term ROE.
Over time, value creation requires incremental after-tax return on new equity above cost of equity; if a raised rupee earns 15% and the required return is 13%, it adds economic profit, but less than a rupee earning 20%. We need evidence of actual issue size, allocation, resulting Tier 1 and incremental profitability.
Bajaj Housing Finance illustrates portfolio diversification: management reported 25% AUM growth, 18% PAT growth to 757 and 12.7% ROE, versus 12.2% earlier. [A; B, p. 4] Housing loans offer different duration, collateral and yield economics than the parent.
The consolidated parent can increase scale through lower-yield secured mortgages while improving risk-adjusted capital stability. Evaluate the subsidiary separately, then reconcile ownership and minority interests rather than applying the parent P/E indiscriminately to subsidiary PAT.
Chapter 6 — Scenario model and valuation discipline
Operating model: FY28 illustrative run rate
Key takeaways
- The following model starts from Q2 annualized net total income (15,974 × 4 = 63,896) and applies explicit growth and operating/credit-cost assumptions.
- This is a normalized next-year run-rate sensitivity, not a statutory FY28 earnings forecast: seasonality, funding, provision timing, minority interest and capital issuance are not independently forecast.
- Operating expense and loan-loss ratios are expressed as percentages of net total income; these are not credit-cost percentages of AUM.
*Illustrative tax retention of 74.5%, close to the reported 6,367/8,537 relationship; not a tax forecast. Formula: net income × (1 − expense ratio − loss ratio), then × 74.5%. Base case assumes growing franchise and stable efficiency; bear case models weaker income growth and higher credit/operating costs; bull assumes lower loss intensity and scale economies.
The reported quarterly loan-loss ratio to net total income is 13.56%, including overlay; the base case is near this level. No overlay reversal is automatically credited.
The base case produces roughly 31,000 crore group PAT pre-minority on a run-rate basis, not ordinary equity attributable earnings. To calculate a fair per-share value, one must subtract minority interests and reconcile actual diluted shares following any QIP/warrants.
A headline group PAT growth rate is not an EPS growth rate. At the reference close of ₹1,117, the price implies a meaningful premium to many conventional lenders, but we do not assert a peer-relative multiple without a same-date, split-adjusted, attributable earnings and book-value audit.
A valuation framework that survives the credit cycle
For a lender, P/BV and sustainable ROE are related; use justified P/B = (normalized ROE − long-run growth)/(cost of equity − long-run growth), assuming clean surplus and steady-state growth below required return. This is a terminal economics framework, not a near-term target.
With 18% ROE, 13% cost of equity and 7% growth, justified terminal P/B is 1.83x; with 20%, 12% and 8% it is 3.0x; with 22%, 11.5% and 8% it is 4.0x. A fast-growth franchise can deserve higher than steady-state value during its high-growth phase; the excess value is the present value of that phase's economic profit.
Do not treat these illustrative terminal multiples as current fair P/B.
Peer set: Cholamandalam for scaled diversification, Shriram for secured vehicle underwriting, Tata Capital for broad lending, HDB Financial for customer distribution, and selected digital/consumer lenders for unsecured economics. Housing subsidiaries should be compared with housing financiers using their own capital and return ratios.
Gold lenders offer a useful collateral benchmark, not a consolidated valuation comp. A proper peer model should align date, consolidated scope, Stage 3/RBI NPA definitions, normalized loss rates, attributable earnings and capital structure.
We deliberately withhold precise relative cheapness or target price until that reconciliation is complete.
Chapter 7 — Investment playbook and monitoring dashboard
Key takeaways
- Variant view: the market may price a consumer-lending growth story; we view the next phase as a test of a multi-product risk architecture.
- If gold-loan operational issues cure, conservative MSME vintages scale and AI improves approval economics without weaker credit, the group could maintain returns while broadening secured lending.
- The alternative is capital-intensive distribution and credit-cycle normalization that merely preserves aggregate growth while diluting per-share earnings.
Near-term catalyst: December-quarter results can validate the gold-loan cure and whether festive origination produces comparable vintages; track capital transaction disclosures before assuming proceeds. Thesis breakers: sustained rise in comparable credit loss and early delinquency, rapid growth with weaker approval discipline, repeated unexplained provisions, or persistent post-raise ROE deterioration without higher sustainable per-share profit. A severe funding shock or regulatory restriction would also require immediate reassessment.
Portfolio decision: a quality watchlist candidate, not a blanket allocation recommendation. Add conviction only when the audited per-share valuation provides a margin of safety under normalized credit assumptions. Use risk limits appropriate to an individual portfolio; do not size on the basis of AUM growth alone.
References and evidence limitations
- [A] Bajaj Finance Q2 FY27 investor presentation, 2026-10-19: https://www.bseindia.com/xml-data/corpfiling/AttachHis/e7c0f169-e396-4a7d-a57d-b1f7c47f6290.pdf
- [B] Bajaj Finance Q2 FY27 earnings call transcript, 2026-10-19: https://www.bseindia.com/xml-data/corpfiling/AttachLive/7873abcb-7f97-487d-a474-028b0a26feea.pdf
- [C] Bajaj Finance Q1 FY27 earnings call transcript, 2026-07-30: https://www.bseindia.com/xml-data/corpfiling/AttachHis/b101e24e-b807-4c74-9025-90009f4b47a5.pdf
- Sage PRO canonical market price: BAJFINANCE, 13 November 2026 close ₹1,117. Internal price reference, not an issuer financial metric.
- Q1 financial cross-check: issuer Q1 call and Q2 comparative disclosures; exact prior-year rounded estimates clearly marked. Full statutory consolidated equity, cash-flow/ALM maturity schedule, minority-interest bridge, capital issue terms and matched peer valuation remain to be reconciled. No fabricated target, FinBERT-derived management claim or score is used.