Long-Term Thesis
The 5-to-10-Year Underwriting Question
The long-term thesis is that Bajaj Finance can compound book value at roughly 18-20% a year through FY2030-FY2035 by reinvesting earnings into an unbanked-and-under-credited India at a through-cycle ROE that never falls below 17%, and that the cross-sell flywheel — 119.3 million customers, 86.6 million app users, 94.4 million EMI cards in force, ~242,000 active distribution points — defends that ROE against a converging field of private-sector banks, fintechs, and a recently-listed Tata Capital. Three things would break it: (i) the funding-cost gap to private-sector banks widens past 100 bps while banks take meaningful share in small-ticket consumer credit; (ii) RBI Upper-Layer drift continues converging NBFC rules toward bank-style provisioning and unsecured concentration caps; or (iii) the post-Rajeev-Jain succession (term ends 31-Mar-2028) produces a second false start. The under-appreciated bull lever is opex/NTI compression from ~33% toward 30% via FinAI — roughly 100 bps of ROA even if the financing margin never recovers a basis point. Medium-conviction long, not high-conviction long, because the franchise quality is undeniable but the price (5.21x book) already discounts a favourable resolution on each test.
The single sentence: Bajaj Finance is a structurally compounding retail-lender franchise with a multi-decade reinvestment runway in Indian credit; the next five-to-ten years test whether the cross-sell moat compounds faster than three slow-moving structural erosions — bank competition on cost of funds, regulatory convergence toward bank-style rules, and the AUM-base law of large numbers — can erode it.
The 5-to-10-Year Underwriting Map
Driver #1 (through-cycle ROE above 17%) is the keystone — every other line feeds into it, and the cohort regression in upstream tabs implies each 200 bps of ROE compression strips roughly 1x off the price-to-book multiple. The thesis does not require BFL to reclaim the FY23 ROE peak of 23.5%; it requires the floor under ROE to sit above 17% in any normal year and above 14% in any cyclical trough. Drivers 4 and 5 (funding cost and operating leverage) are the most credible bridges back toward 20%+; driver 6 (regulator) is the single largest exogenous swing factor with no analyst-controllable visibility.
Compounding Path
A 5-to-10-year compounder is judged on whether revenue, book value, and earnings re-invest at returns above the cost of equity for a duration the market is not pricing. For Bajaj Finance the historical record is 26% book-value CAGR and 19% through-cycle ROE — the question is whether that re-investment opportunity persists at materially the same shape through FY2030-FY2035.
AUM compounded at roughly 26% a year over a decade. PAT compounded at 30% over the same window, but the recent two-year deceleration to 15% is the warning chart for the next decade: book-base law of large numbers is real even before any structural break. Book-value per share compounded at ~26% over the same window — the right anchor for a lender. A 16-20% future BV CAGR plus a stable multiple is the realistic underwriting math for the next five years.
Opex/NTI fell from 58% in FY08 to 32.8% in FY26 (comparable basis) — a 25-point structural step-down that explains most of the ROA expansion the headline ROE chart hides. The next decade's compounding math depends on whether management can deliver another 3-4 points of opex compression (toward 28-30%) via FinAI, which would alone add ~100 bps of ROA. The next 10 years are not a repeat of the last 10 (book-base too large), but a 16-20% book CAGR with stable-to-modestly-expanding ROE is achievable if drivers 4 and 5 in the underwriting map work.
The base case delivers a high-teens compound return without any multiple expansion — that is the value of buying a 16-20% book compounder at a stable multiple. The bull-bear span (3% IRR to 35% IRR over five years) is wider than usual for a quality compounder because the valuation already discounts most of the favourable resolution; the asymmetry sits in the regulator + bank-competition outcome, not in the customer-franchise quality.
Durability and Moat Tests
A 5-to-10-year compounder is judged less on what its moat looks like today and more on whether the moat survives three identifiable structural pressures. Each test below has a clear validating signal and a clear refuting signal — anything ambiguous in between leaves the verdict on watchlist.
Test #2 (financial) is the cleanest because it is measurable every quarter. Test #1 (competitive) is the slowest-moving but the most consequential — the bundle defence is fragile against a 5-year window in which two of the five legs (cost-of-funds gap and digital-onboarding speed) erode against banks. Test #3 (regulatory) is the single largest exogenous swing factor and the lowest-confidence row in the underwriting map — it can change ROE by 150 bps without warning and is not in any analyst's control.
Management and Capital Allocation Over a Cycle
Across the last decade the company has not done any of the things that destroy long-term lender value: no margin-eroding M&A, no offshore mis-adventures, no buyback timing engineering, no opaque off-balance-sheet vehicles. Every rupee of net profit either funded the loan book at the through-cycle ROE or was paid out as dividend. The dividend payout ratio has doubled from 10% in FY15 to ~21% in FY26 — a quiet signal that management sees diminishing reinvestment opportunities at the historic ROE, but not yet abandoned discipline. The two genuinely significant capital actions of the past three years (the September 2024 BHFL IPO that raised ₹3,000 cr OFS and the September 2025 4:1 bonus + 1:2 split) were structural: the BHFL listing was forced by RBI's three-year Upper-Layer mandatory-listing rule; the split was a liquidity move that preserved retail accessibility post the run-up to ₹9,300+ on pre-split basis.
The team is genuinely disciplined and has earned the benefit of the doubt on capital allocation, but the last 24 months have introduced three real reservations — the rising payout (signal of diminishing ROIC), the Saha reversal (signal of fragile succession), and the recurring "exceptional" provision labels (signal that management is editorially smoothing earnings even where the underlying credit is normalising). None of these breaks the thesis; together they reduce the premium an investor can defensibly pay for the franchise.
Failure Modes
The thesis breaks if any one of these failure modes turns from possible to actual. None is a near-term event-risk; all are slow-moving structural pressures that show up in trailing 8-quarter prints, not in any single announcement.
The two highest-severity failure modes are structural, not idiosyncratic. Bank competition on funding cost and RBI Upper-Layer drift on unsecured rules are both outside management's control and both compound on a 24-36 month rhythm. They sit at the top because they are the only failure modes that can repeatedly re-price the entire ROE corridor without a single discrete event.
What To Watch Over Years, Not Just Quarters
Five observable milestones that will update the long-term thesis. Each is multi-quarter or multi-year, not a single print — quarterly noise should not move the thesis unless it confirms one of these slow-moving patterns.
The long-term thesis changes most if the cost-of-funds gap to private-sector banks widens past 100 bps for eight consecutive quarters while BFL loses share in small-ticket consumer credit — a single multi-year pattern that would simultaneously confirm bank-side erosion of the moat's most material leg and refute the cross-sell flywheel's ability to compound at the rate the price has already paid for.