Five of six blocks middling; no single re-grade changes the answer.
Why:
- Half the reported profit is a paper mark the company itself influences. 48% of FY26 standalone profit is an unrealised gain on its own illiquid security receipts, valued on an unobservable input where a 5% move swings ₹209 cr — half a year's profit (T1). The marks are set under a rating-band framework the RBI publicly faulted at a peer (Edelweiss ARC, May 2024) for incorrect valuation, and the filing back-tests none of its past marks.
- The cash runs the other way from the profits. Cash collections fell 10.3% in FY26 to below FY24's level while the book grew 32%; billed fees grew just 7% in two years against a 65% bigger fee-earning book; cash tax fell while pre-tax profit rose a third (T1). The growth is in accruals and marks, not in money arriving.
- A pure exit at a so-so price in a pool that has stopped growing. Sellers take all of the ₹697-733 cr and the company receives nothing, while its own borrowings rose 8x in two years; the industry's stressed-asset pool is flat at about ₹1.33 lakh cr with corporate stress at a 15-year low; and the one investor to fully exit before this offer earned roughly 2.7% a year over 18 years (T1/T3).
Valuation at the band
| Floor ₹132 (T1) | Cap ₹139 (T1) | |
|---|---|---|
| Bid window | 9 to 11 September 2026 | |
| Bid lot | 107 shares (T3) | |
| Fresh shares | none — pure offer for sale | |
| Post-issue shares | 324,897,140 | 324,897,140 |
| Market capitalisation | ₹4,289 cr | ₹4,516 cr |
| P/E (consolidated, as reported) | 12.2x | 12.8x |
| P/E (illustrative, ex unrealised marks and the litigation write-back) | 17.5x | 18.4x |
| P/B (consolidated book value ₹90.96/share) | 1.45x | 1.53x |
| Promoter group holding after | 78.7% | 78.7% |
EV/EBITDA is not shown: for a financial company whose borrowings are working assets the measure is meaningless, so price-to-book stands in its place. The reported P/E looks cheap; strip the ₹195 cr of unrealised marks and the ₹23 cr one-time provision write-back from FY26 and the multiple is in the high teens on a book earning about an 8% return ex-marks. The price does not move the verdict either way; the six ratings judge the business, not the band.
The story
A buyer of this IPO is buying a fee annuity that is supposed to emerge: fee-earning assets grew 65% in two years, retail is now a third of that base, and fees land almost intact in profit because costs barely grow. The catch is that the fee engine has not yet shown up in billed cash — two-thirds of the reported fee growth is an accounting accrual swing — and meanwhile the profits shown today are mostly marks and run-off from an old corporate book that is going away.
What this business is
Arcil is India's first asset reconstruction company, founded in 2002. It buys bad loans from banks and finance companies at a discount — cumulatively it has paid ₹44,114 cr for ₹89,909 cr of defaulted principal — puts them into trusts it manages, and earns three ways: management fees of 0.25-5% on trust assets, a share of recoveries, and gains on the security receipts it holds itself. It manages ₹20,150 cr of stressed assets, roughly 15% of the industry pool, second behind the government-backed NARCL, with 206 employees across 13 offices.
The business has repositioned hard: retail loans are now a third of the fee-earning book (from a fifth), sourcing has shifted from banks to NBFCs, and 86% of FY26 purchases were structured deals that pay fees rather than tie up capital. FY26 standalone income was ₹785 cr with ₹408 cr of profit — but ₹195 cr of that income is an unrealised mark.
Easy or difficult business? A licence only five ARCs meaningfully clear (₹300 cr minimum capital against a ₹1,000 cr bar for bankruptcy-court work), statutory enforcement powers, and 25 years of recovery infrastructure — genuinely hard to enter. But the actual transaction is a price auction among 6-7 bidders for assets with no market price, so the difficulty protects participation, not margins.
Key numbers
| ₹ cr, standalone | FY2024 | FY2025 | FY2026 |
|---|---|---|---|
| Total income | 548.3 | 623.4 | 785.1 |
| of which unrealised fair-value marks | (25.8) | 118.4 | 195.2 |
| PAT | 305.2 | 355.4 | 407.7 |
| PAT (consolidated, attributable) | 330.5 | 329.5 | 351.7 |
| Cash collections | 3,678.1 | 3,882.7 | 3,484.4 |
| Assets under management | 15,230 | 16,853 | 20,150 |
Consolidated profit — the group including the trusts — is flat across three years; the growth lives in the standalone numbers, and within them in the marks. Collections fell in FY26 mostly because the old written-off book's recoveries are running out; collections on the live book rose about 7%.
| Revenue engine, ₹ cr standalone | FY2024 | FY2025 | FY2026 |
|---|---|---|---|
| Recurring fees (management + recovery + accruals) | 187.8 | 179.5 | 291.1 |
| Realised investment and recovery income | 94.9 | 198.4 | 200.9 |
| Recoveries of amounts written off earlier | 287.5 | 100.0 | 65.9 |
| Unrealised fair-value marks | (25.8) | 118.4 | 195.2 |
| Other income | 4.0 | 27.0 | 32.0 |
The FY24 column is dominated by a one-time harvest of previously written-off assets the filing itself calls exhausted; the FY26 column is dominated by marks. The recurring fee line in between is the business the story rests on — and about two-thirds of its FY26 jump is an unbilled accrual swing, not billed cash.
Scorecard
| Block | Rating | Why |
|---|---|---|
| Right to win | MARGINAL | The moat is real but defensive: the cheapest funding in the peer set, a capital-gated licence, and a delivered retail repositioning. It does not convert into superior economics — the realised fee yield fell on every measure, expenses ate 82% of the FY26 fee increase, apparent market-share gains are mostly the denominator shrinking, and pricing is set in auctions the company concedes are competitive. A licence that protects a seat at the table, not a margin. |
| Industry and TAM | MARGINAL | The pool is flat at about ₹1.33 lakh cr and the corporate stress that feeds the profitable end is at a 15-year low with the RBI expecting it to stay there. The fee-bearing slice is ₹30,000-43,000 cr — real headroom but only 2.6-3.7x — and the growing retail end is lower-recovery with the issuer's own report flagging margin compression. A consolidation story in a market that is not growing. |
| Financial momentum | MARGINAL | The operating engine ex-artifacts grew mid-teens and the scary-looking collection decline is mostly an old book running off. But the FY26 print rests on non-repeating and partly discretionary items larger than the year's whole profit increase, half of reported profit is an unrealised mark, the receivable tail grew to 43% of gross at falling coverage, and a ~₹70 cr finance-cost step-up lands in FY27. |
| Risks, governance, RPTs | MARGINAL | Four RBI supervisory letters in three years with substantive findings, including one on how revenue is recognised; a ₹56 cr tax demand under a suppression allegation plus ₹41 cr of penalties under appeal; dealings with promoter-shareholder SBI outside the related-party note with no year-wise sizing; and a database-level audit-trail gap. Everything disclosed, no enforcement ever concluded against it — but an untidy file at a regulated financial. |
| Promoter and cap table | PASS | As clean as an Indian cap table gets: no new share issued in 17 years, nothing convertible, nil pledges. The retail pivot was promised and delivered at scale, the CEO succession was designed with an 11-month overlap, and Avenue cut its own sell-down 64% between draft and final filing. |
| Offer structure | MARGINAL | A pure ₹697-733 cr exit at 1.45-1.53x book in which the company raises nothing while its borrowings rose 8x; the sole prior full exit made ~2.7% a year over 18 years; ~64% of the register unlocks at month six; and a further ~3.7% of the company must by law be sold within listing timelines. Clean mechanics, mediocre shape. |
Watch out for
- The marks framework has failed at a peer. The RBI's May-2024 action against Edelweiss ARC expressly cited incorrect security-receipt valuation under the same rating-band framework that produces half of Arcil's profit, and Arcil's filing contains no back-test of past marks against realised recoveries (T1/T3).
- A cluster of manager-influenced income in the final pre-filing year: ₹195 cr of unrealised marks, a ₹23 cr litigation provision released while the matters stay live, a ₹31 cr favourable accrual swing, and discretionary trust-deposit interest — together larger than the year's entire profit increase, with no quarterly stub to test any of it (T1). First post-listing results settle it.
- Supply is mandatory, not optional. Public holding is 21.3% against the 25% floor, so roughly ₹630 cr more stock must be sold within listing rules, on top of ~64% of the register unlocking six months after listing (T1).
- The tax matter carries a suppression allegation: ₹56 cr demanded under the extended limitation period plus ₹41 cr of penalties already imposed by order, under appeal — about 28% of a year's profit combined (T1).
The offer
- Raising ₹697-733 cr at the band — entirely an offer for sale of 52,731,946 shares; there is no fresh issue and the company receives nothing (T1).
- Selling: promoter Avenue Capital entities sell about 11% of their holding; SBI and Federal Bank trim broadly pro-rata; an 18-year investor (GIC/Lathe) exits in full.
- Implied valuation is in the table above: ₹4,289 to ₹4,516 cr at the band.
- Promoters (the Avenue Capital group) hold 78.7% after the offer, unchanged in economic terms by price since no new shares are issued — with further selling required within listing rules to reach the 25% public minimum.
Initial assessment from the RHP, with no outside verification beyond the price band and offer dates, which come from the BSE public records with press agreeing, and the peer and industry checks noted above. Numbers carry source tiers: (T1) the filing's audited sections, (T2) exchange or registry data, (T3) the issuer-commissioned industry chapter or press, UNVERIFIED where the filing does not support them. Not a recommendation. The valuation section states what the announced band implies and is not a view on whether that price is right.