Chapter 7

Figures converted from Indian rupees at historical FX rates — see data/company.json.fx_rates for the rate table. Ratios, margins, and multiples are unitless and unchanged.

Loan Book

Nuvama runs a $535 million lending book against securities and margin, ~99.7% secured, that carries an expected-credit-loss provision of just 0.29% and a stated non-performing balance of $1,700 — effectively zero. That near-absence of losses is real, but it is a model output resting on collateral the filings never quantify, in a book that has never met a genuine equity drawdown at this scale. For a reader who screens out anything that could go bankrupt, this is where the balance-sheet risk lives.

A short, secured book that lends against listed collateral

The lending sits inside Nuvama Wealth Finance Limited and shows up on the consolidated balance sheet as one line, "Loans." At 31 March 2025 it was $537 million gross — $431 million of retail credit (loan against securities) plus $107 million of margin trading facility [1]. Both are lent against financial collateral — equity shares, mutual-fund units, bonds and AIF units [2]. Of the $537 million, $535 million was secured; the unsecured slice — mostly legacy ESOP financing — was $1.6 million, about 0.3% of the book [3].

Two features matter for how this book behaves under stress. It is short: every dollar of the $535 million net loan balance matures within twelve months, and $111 million falls due inside three months [4]. And it is callable — the facilities carry a right for the group to cancel or reduce them at one day's notice, and margining collateral is valued daily [5]. A lender who can re-price collateral every day and pull the loan within a day is structurally better placed than a term lender when markets fall.

Gross Loan Book, FY25 ($ m)

537

Secured

99.7%

ECL Coverage, FY25

0.29%

Impaired Balance ($ '000)

1.7
Loading...

Source: FY2025 Annual Report, Note 14 (FY24–FY25) [6]; FY2023 Annual Report, Note 14 (FY22–FY23) [7].

One point of arithmetic worth flagging early: on this consolidated measure the book did not grow in FY25 — it fell 6%, from $587 million to $537 million [8]. The "+76%" growth cited elsewhere in the report (Wealth Economics) is the Nuvama Wealth segment's loan line — $327 million to $543 million between FY25 and FY26 [9]. The two are different scopes: the segment figure is the HNI/affluent book alone; the consolidated Note-14 figure adds Nuvama Private's lending and the group's margin trading facility. The surge, in other words, is a FY26 event that lands after the last audited credit disclosure.

The provision is a model output, not an observed loss rate

The book's headline credit quality is close to pristine. At FY25, $537 million of gross loans classified as Stage I "high grade," with Stage II nil and Stage III — the credit-impaired bucket — at $1,700 [10]. The expected-credit-loss allowance was $1.55 million, or 0.29% of the book [11]. That is not a one-year artefact: coverage has sat in a 0.26%–0.29% band every year since FY22, and the impaired balance has never exceeded $22,000 [12].

No Results

Source: FY2025 Annual Report, Note 14.A [13], [14]; FY2023 Annual Report, Note 14.A [15].

The reason the provision is so small is mechanical, and the annual report states it plainly. Because the business lends against securities, the expected-credit-loss model applies a Basel-style haircut to the collateral, treats only the exposure above the haircut collateral as at risk, and assigns loss-given-default of 65% to the unsecured portion alone [16]. On a book that is 99.7% secured, that construction all but guarantees a near-zero provision: the model assumes the haircut collateral will be there and can be realised. The 0.29% is therefore best read not as an observed loss experience but as a statement about how well-collateralised the book is assumed to be — and the assumption has only ever operated in the rising or sideways equity market of FY21 through FY26.

What the disclosure does not tell you

For a reader trying to size the tail, the gap in the record matters as much as the numbers in it. The filings name the collateral types and describe the haircut methodology, but they do not disclose the one thing that would let an outsider judge the cushion: the fair value of collateral held against the loan book, the distribution of loan-to-value ratios, or a concentration table by borrower or by pledged stock. Management asserts there is "no significant concentration with regard to single/group client and industry," but publishes no figures behind it [17]. Collateral, the notes confirm, is not recorded on the balance sheet unless repossessed [18]. Nor do the quarterly calls fill the gap: management's collateral commentary is about the custody and clearing business, not the lending book. The reader is left to take the coverage ratio on trust.

Collateral realisation is not automatic — the Anugrah matter

The record contains one concrete demonstration that liquidating collateral can create losses even when the loan ledger shows none. The FY2025 audit opinion carries an Emphasis of Matter — an item the auditor thought material enough to flag while leaving the opinion unmodified — pointing to two subsidiary disputes now before the Supreme Court over the liquidation of collateral [19].

The larger of the two runs through the clearing subsidiary, Nuvama Clearing Services. When a trading member, Anugrah, defaulted on its margin obligations in early 2020, the subsidiary liquidated $53.6 million of Anugrah's collateral to cover the shortfall. NSE Clearing later held that the liquidation swept up collateral beyond the margin shortfall — reaching securities belonging to Anugrah's end-clients — and ordered the securities reinstated; the Securities Appellate Tribunal upheld that order in December 2023, and the subsidiary's appeal to the Supreme Court is pending. No provision has been made [20].

Source: FY2025 Annual Report, Independent Auditor Report [21] and Note 56 [22]; equity per FY2025 balance sheet.

The read: a manageable book on an untested assumption

Weighing the two sides, the loan book is not, on the evidence, a plausible path to insolvency — the risk the value-investor reader screens for. It is 99.7% secured, callable at a day's notice, matures entirely within a year, and has been de-risked over time: the unsecured share fell from 3.7% of the book in FY23 to 0.3% in FY25 as legacy ESOP and IPO-funding exposures ran off [23]. Even through the FY23 correction it produced no meaningful impairment. Against Nuvama's $407 million of FY25 equity, a book this structure would have to lose an implausible fraction of its collateral value, on top of failed margin calls, to threaten solvency.

The honest qualification is that the comfort rests on an assumption that has not been tested. The near-zero provision is what the model produces when collateral is assumed realisable at haircut value; the Anugrah matter is a live, cited instance of collateral realisation going wrong; and the fastest-growing, least-seasoned vintage of the book — the FY26 surge — postdates the last audited staging. What would change the read is observable and specific: Stage II migration or a rising coverage ratio in the FY2026 annual report's Note 14.A; the first disclosure of loan-to-value or collateral-coverage figures on the enlarged book; an adverse Supreme Court outcome on the Anugrah appeal; or a genuine, sharp equity drawdown that lets an outsider watch, for the first time at this scale, whether the collateral cushion holds.