It started with one aged receivable
On 11 March 2024, a shareholder wrote to SEBI pointing at trade receivables that had stayed outstanding for over two years. That is the oldest test in this profession: real sales convert to cash; fictitious sales sit as perpetual receivables. An Investigating Authority was appointed in October 2024, a forensic auditor in December 2024 — and over the following eighteen months that single thread, per the order, pulled the group's reported numbers apart layer by layer.
I want to dwell on the origin, because it is the part most coverage skips. Cases of this scale are assumed to require sophisticated detection — algorithms, informants, exchange surveillance. This one allegedly required an ageing schedule. The working paper every articled trainee prepares in their first month is the same paper that opened what may become one of the most significant accounting investigations in Indian capital-market history. That should change how seriously you treat the boring schedules in your own audit file.
How a fraud examiner actually reads a file like this
Before walking through the substance, it is worth being honest about method, because it is what separates examination from commentary. When I sit across a management team in a forensic review, I am not asking whether the revenue is fake. That question is unanswerable on day one and it puts everyone on the defensive. The questions I am actually working through are narrower and colder:
- Can I verify this independently? Not from the ledger — from outside it. Bank statements, counterparty records, customs data, exchange data.
- Can I trace it to source documents? Invoice, contract, delivery evidence, approval. A number with no document chain is a claim, not a fact.
- Can I identify the counterparty? Real customers exist. They have GST registrations, bank accounts, their own audited books — and they answer confirmations.
- Can I reconcile the cash? Every genuine transaction eventually touches a bank statement. Cash is the one record management cannot journalise.
- Can I reproduce management's explanation from evidence? If their story is true, the documents should let me rebuild it without their help.
Read SEBI's order with that frame and you will notice something: the order is structured around exactly these failures. Not "the revenue was fake" — but could not be independently verified, no transaction-level documents, counterparty denies the relationship, ledger entries with no bank trail, explanation unsupported by any contemporaneous record. That is examiner's language. It is also why the order is more dangerous for the noticees than a sensational allegation would be: each finding is a statement about the absence of evidence, and absence is very hard to argue with.
The structure: where the revenue lived
The group's reported activity sat overwhelmingly offshore, in a four-layer chain:
| Entity | Role |
|---|---|
| Rajesh Exports Ltd (India) | Listed parent — bullion import, gold products, jewellery retail |
| REL Singapore Pte Ltd | Wholly-owned holding company; no day-to-day operations |
| Global Gold Refineries AG (GGR), Switzerland | Holding company of the refinery; recognised the gross value of gold transactions |
| Valcambi SA, Switzerland | The actual operating refinery |
Per the order, 97–99% of consolidated revenue was attributed to overseas subsidiaries and step-down subsidiaries — whose standalone financial statements were never uploaded to REL's website as Regulation 46(2)(s) of LODR and Section 136(1) of the Companies Act require, and whose primary records SEBI could not fully access. (Two percentages recur in this article and they measure different things: 97–99% is the share of consolidated revenue attributed to subsidiaries; 99.80% — the ₹15,15,385 crore — is the portion of consolidated revenue SEBI states, at para 219(VI), was prima facie misrepresented.) When I see revenue parked where the home regulator cannot easily reach the books, I don't treat it as suspicious by itself — plenty of legitimate groups are structured this way. I treat it as a concentration of verification risk: the entire group's credibility now rests on records one jurisdiction away from scrutiny. That is a fact pattern that demands more disclosure, not less. Here, per the order, it got less.
The numbers that frame the case
Table 1 of the order sets out the standalone-versus-consolidated divergence (₹ crore):
| Particulars (₹ cr) | FY21 | FY22 | FY23 | FY24 | FY25 | FY26 |
|---|---|---|---|---|---|---|
| Consolidated revenue (A) | 2,58,306 | 2,43,128 | 3,39,690 | 2,80,676 | 4,23,099 | 7,78,716 |
| Standalone revenue (B) | 2,060 | 6,237 | 5,762 | 5,401 | 7,027 | 9,189 |
| Revenue from subsidiaries (A−B) | 2,56,245 | 2,36,891 | 3,33,928 | 2,75,276 | 4,16,072 | 7,69,527 |
| Consolidated PAT (C) | 845 | 1,009 | 1,432 | 336 | 95 | 113 |
| Standalone PAT (D) | 99 | 23 | 30 | 17 | 24 | 32 |
Read it the way an examiner would. Standalone PAT never exceeds ₹99 crore while consolidated revenue scales into the lakhs of crores — nearly all reported revenue and profit originate in entities whose primary records were beyond reach. Then look at the trajectory: the divergence widens after scrutiny began. FY26 shows ₹7.78 lakh crore consolidated against ₹9,189 crore standalone. In my experience the trajectory matters more than the level — a structural gap that grows while questions are pending tells you how management responds to examination, which is itself evidence.
Pillar one: the offshore revenue gross-up
The decisive comparison in the order: Valcambi — the only real operating business in the chain — had audited standalone revenue of roughly ₹427–743 crore per year. GGR, its immediate holding company, reported revenue running into lakhs of crores.
REL's defence is a substance argument: Valcambi's standalone accounts reflect only processing and value-addition income, while GGR recognised the gross value of gold transactions. Let me be fair to the defence, because it deserves technical respect: gross-versus-net is a genuine question. The principal-versus-agent test under Ind AS 115 is one bullion refiners legitimately wrestle with, and I have seen both treatments defended competently. Grossing up metal throughput produces enormous "revenue" without changing economic substance by a rupee — which is exactly why the standard demands you evidence control, inventory risk, and pricing discretion before you book gross.
And that is where, according to SEBI's interim findings, the defence was not accepted — not on theory, on evidence. Per the order: no transaction-level proof, no accounting opinions, no ownership records, no reconciliations. In every gross-vs-net review I have worked on, the answer ultimately sits in a documents folder: purchase contracts, metal accounts, hedging records, title documents. If the folder is empty, the accounting debate is academic. On SEBI's account, the substance defence did not fail on Ind AS 115 — it failed on the missing audit trail.
Pillar two: the circular-trade pattern
Across FY22–FY24, REL booked sales of ₹11,487 crore and purchases of ₹11,488 crore with a single counterparty — roughly two-thirds of its standalone sales and purchases. Near-identical buy and sell values with the same party are the textbook signature of round-tripping: turnover with no margin and no apparent commercial purpose. When I see a matched pair like that in a trial balance, the working hypothesis writes itself; the job is to test it.
SEBI tested it the way you are supposed to: externally. The counterparty — a SEBI-registered broker whose own total revenue from operations was about ₹113 crore — confirmed on deposition that REL was never its client, and that it dealt only with the promoter in his personal capacity. Per the order, the real underlying activity was personal gold-derivatives trading across 102 trading days, with net payments of ₹7.45 crore. The order also flags two credit entries in REL's ledger for this counterparty — ₹1.18 crore and ₹45.16 lakh — that appear in no bank statement: prima facie plug entries forcing the books to tally.
Three numbers frame the allegation: ₹11,487 crore booked, ₹113 crore of counterparty capacity, ₹7.45 crore of actual underlying flows. I keep returning to the middle number. A counterparty's own audited scale is public, cheap to check, and devastating when it doesn't fit — you cannot transact eleven thousand crores through an entity that does a hundred. The scale test costs nothing and nobody runs it.
Pillar three: promoter fund flows without approvals
Independent of the revenue question, the order traces ₹338.90 crore transferred from the company to the promoter against ₹232.44 crore returned — a net outflow of over ₹106 crore with, per SEBI, no board approval, no audit-committee approval, and no related-party disclosure, in breach of Regulation 23(2) of LODR and Ind AS 24. The order also cites routing through a promoter-linked private company, including a ₹200 crore equity investment.
From a case-construction standpoint this is the limb I would least want to defend. Revenue recognition arguments have texture — standards, judgement, expert opinions. Fund flows have none: money either moved or it didn't, approvals either exist in the minutes or they don't. It is discrete, document-based, and provable entirely from bank statements and board records. When we run fund-flow tracing in our own engagements, this is precisely why we start there — it is the limb that survives every defence.
Pillar four: the asset SEBI said it could not verify
"Other Non-Current Investments" on a consolidated basis moved from ₹879.60 crore (March 2021) to ₹1,035.27 crore (March 2023) — and then to ₹10,547.72 crore (March 2024). Asked about the ₹1,035.27 crore, REL described it as an "Investment in Gold Mines in Africa". Per the order, no identifiable investment of that amount could be traced in REL's standalone financial statements, and no entity-wise breakup or reconciliation was furnished.
Alongside, the order finds trade receivables improperly offset against trade payables — contrary to the no-offset rules of Ind AS 1 (paras 32–33) and Ind AS 32 (para 42) and the disclosure requirements of Ind AS 107. Watch the loop close: improper netting masks aged balances, and aged balances were the original complaint. The alleged concealment mechanism and the detection trigger are the same line item. I have rarely seen the circle drawn that cleanly in a regulatory file.
The obstruction thread
Running through all four pillars is a documentation record the order quantifies precisely: ERP access, books, and the journal dump withheld; complete documentation furnished for only 2.03% of a ₹7,021.36 crore purchases sample and 35.07% of a ₹12,217.15 crore sales sample. The first forensic audit completed without the company's cooperation; the order directs appointment of a fresh forensic auditor.
Practitioners should study how SEBI used this, because it is a masterclass in documenting pendency. Non-cooperation was not treated as a procedural footnote — it was recorded sample-by-sample, value-by-value, and deployed to support adverse inference. I tell every team I work with the same thing: when you face denial of journal-level access, write down the sample value, the percentage substantiated, and the dates of every request. The 2.03% figure does more work in this order than any analytical inference — because it converts obstruction from a complaint into a measurement.
What the market saw
Per the order, REL's shareholder count grew from 22,472 (March 2020) to 2,06,942 (September 2025) (para 208). The share price peaked at ₹1,028.40 on 6 February 2023 — a market capitalisation of roughly ₹30,365 crore (para 209) — and had declined to ₹80.11 by 2 April 2026, about ₹2,365 crore (para 212). The order notes that the period of elevated prices substantially coincides with the period of the alleged misstatements; the erosion that followed — over 90% of peak value, on the order's own market-cap figures — fell on the nearly two lakh shareholders holding the stock by then. That is the investor-protection logic on which interim directions rest, and it is worth remembering who ultimately absorbs the cost of an unverifiable ledger.
The red-flag matrix
Every indicator in this case transposes to other engagements — this is the matrix I would lift into any fraud-risk diagnostic. The middle column is the only case-specific part:
| Red flag | As alleged in REL | The audit test |
|---|---|---|
| Aged trade receivables | Receivables outstanding >2 years — the original complaint | Ageing analysis; subsequent-receipts review; cash-conversion test |
| Revenue concentrated in opaque subsidiaries | 97–99% of consolidated revenue from overseas entities; subsidiary FS never published | Entity-wise revenue split; subsidiary FS filing compliance |
| Parent vs operating-entity mismatch | Refinery's standalone ~₹427–743 cr vs lakhs of crores reported up the chain | Reconcile operating entity standalone to consolidated; principal-vs-agent testing |
| Counterparty denial | Broker denied REL was ever its client | Direct independent confirmation of large/recurring counterparties |
| Near-identical buy = sell values | ₹11,487 cr sales ≈ ₹11,488 cr purchases, same party | Margin analysis; matched buy/sell timing tests; counterparty scale test |
| Promoter fund routing | ₹338.90 cr out, ₹232.44 cr back; no approvals | Fund-flow tracing against board/AC minutes and the RPT register |
| Plug entries | Ledger credits traceable to no bank statement | Journal-entry testing; ledger-to-bank matching |
| Unverifiable "other" assets | ₹1,035.27 cr "gold mines in Africa", untraceable | Existence and valuation testing on jumps in residual asset buckets |
| Improper netting | Receivables offset against payables (Ind AS 1/32/107) | Gross-up testing; offsetting-criteria review |
| Audit obstruction | ERP/journal dump withheld; 2.03% of sample documented | Document pendency precisely; treat as substantive, not procedural |
The provisions invoked
| Instrument | Provisions cited prima facie |
|---|---|
| SEBI Act, 1992 | Sections 12A(a),(b),(c); order under 11(1), 11(4), 11B; non-cooperation under 11(2)(ia), 11C(3) |
| PFUTP Regulations, 2003 | Reg 3(b),(c),(d); 4(1); 4(2)(e),(f),(k),(r) — misleading financial statements as a fraudulent trade practice |
| LODR Regulations, 2015 | Reg 4 (principles); 23(2) RPT approval; 33(1) & 48 financial results per Ind AS; 34(3) + Sch V disclosures; 46(2)(s) subsidiary FS on website |
| Companies Act, 2013 | Sec 128(1) true and fair books; Sec 136(1) subsidiary FS on website |
| Ind AS | 115 (revenue — principal vs agent); 110 (consolidation); 24 (related parties); 1, 32, 107 (offsetting and disclosure) |
Two procedural notes worth a practitioner's attention. First, SEBI expressly applied the Supreme Court's threshold for "fraud" under PFUTP laid down in Reliance Industries Ltd. & Ors. v. SEBI (2026 INSC 585) — requiring intent to defraud or manipulate to be made out even at the prima facie stage (applied at para 220 of the order) — and held the conduct met it. Second, the order has been forwarded to the NFRA for examination of the statutory audit. The interim directions: cooperation with the investigation within 30 days, the promoter restrained from dealing in the company's securities, true and fair restatement of disclosures, and a fresh forensic audit. The company, for its part, has publicly rejected the findings and stated it will submit a detailed response — which is its right, and which is why every finding above carries the prefix "alleged".
Five questions every founder should be able to answer
You do not need to run a listed bullion group for this order to apply to you. The same verification logic gets run — at smaller scale — in every funding due diligence, every statutory audit escalation, every investor dispute. Before any of those arrive, sit with your CFO and answer these five questions honestly:
- Can every related-party transaction be explained and documented? Not explained verbally — supported by an approval in the minutes, a disclosure in the RPT register, and an arm's-length rationale on file.
- Are all subsidiary financial statements current, audited, and accessible? If your group's value sits in entities whose accounts you cannot produce on request, your consolidated numbers are an assertion, not a record.
- Can your revenue be verified without your help? Customer confirmations, bank credits, GST trails. If the only evidence for a sale is your own ledger, you don't have evidence.
- Are board and audit-committee approvals traceable for every significant flow? Money moving before minutes exist is the single most common finding in our fund-flow work.
- Would a forensic examiner receive complete records within 48 hours? This is the real test. Not whether your books are clean — whether you can demonstrate they are clean, fast, from documents, without a single reconstruction.
If any answer is no, you don't necessarily have a fraud problem. You have a provability problem — and in a regulatory or diligence setting, the two are treated identically until the documents say otherwise.
The honest takeaway
Strip away the scale and what this file teaches is the same thing every forensic engagement teaches. Aged receivables remain the most reliable entry point into revenue fraud, because cash conversion is the test real sales always pass. Where revenue concentrates offshore, the gap between the operating entity's standalone accounts and the consolidated figure is the case. Counterparty confirmation is primary evidence, not a formality — one deposition collapsed two-thirds of standalone turnover here. Obstruction, measured precisely, becomes evidence in its own right. And fabricated value hides in the boring places — netting, "other" buckets — precisely because nobody glamorous looks there.
But the larger lesson is the one I would put in front of every board:
Markets rarely collapse because a regulator finds a problem. They collapse because the problem existed long before anyone asked the question — and because, when the question finally came, the company could not produce the one thing that ends questions: evidence.
Good governance is not the absence of wrongdoing. It is the ability to prove that nothing is wrong — within 48 hours, from documents, without a story.
Need a forensic review, fraud risk assessment, or investigation support?
Our forensics practice covers fund-flow tracing, related-party and round-tripping analysis, forensic audits under SEBI/IBC processes, and red-flag diagnostics for boards and audit committees — led by ACFE (USA) credentialed professionals.
Schedule consultationDisclaimer: This article is a professional commentary on SEBI's interim ex-parte order dated 3 June 2026 in the matter of Rajesh Exports Limited, prepared from the publicly available order. All findings described are SEBI's prima facie allegations at an interim stage — they are not final conclusions. The company has rejected the findings and stated it will submit a detailed response; the noticees are entitled to respond and be heard, and the matter will proceed to detailed investigation. Figures are as stated in the order. Nothing here constitutes legal or investment advice, or any comment on the eventual merits. For matter-specific advice, please consult a qualified professional.