Financial Shenanigans

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

The Forensic Verdict

Forensic Risk Score: 37/100 — Watch. There is no restatement, no auditor resignation, no SEBI enforcement, no qualified ICFR opinion, and five-year CFO/NI averages ~1.0×. The largest concern is an auditor-flagged Rule 11(g) deviation — the SAP/HANA audit-trail feature was not enabled at the database level throughout FY2025, and the Tally audit trail for one segment was only enabled from July 3, 2024. The second is a "kitchen-sink" FY2025 — $11.1M Romania standalone impairment, $3.2M consolidated goodwill impairment, $2.4M solar-EPC bad-debt write-off all in one year — followed by an "Adjusted EBITDA" framing that scrubbed the charges. None of this is fraud; it is concentrated bad news disclosed once, with management's preferred metric stripped of it. The data point that would most change the grade is whether FY2026's receivables build (revenue +16%, receivables +26%) reverses in FY2027 or hardens into a second round of bad-debt provisioning.

Shenanigans scorecard

Breeding Ground

The governance breeding ground is moderately concerning — not from any single failure, but because the structural pieces all lean the same way. Promoter group sits exactly at the 75% SEBI cap as of September 2025 after adding 0.39pp through Jun-Sep 2025 buying. Three of five executive directors are promoters (Rajendra Shah, Harish Rangwala, Vishal Rangwala); a fourth (Pilak Shah) is a Shah family member. Profit-linked commission now exceeds base salary for every whole-time director after the Oct 1, 2024 revision — Vishal Rangwala's pay ratio is $0.15M commission to $0.09M salary. Five independent directors balance five executives, but board composition matters less than committee independence, and the audit-committee chair is unnamed in the available file.

The cleanest part of the breeding ground is the absence of an expectation-beating streak: FY2025 earnings fell about 20% on disclosed exceptional items, and management did not try to keep the streak alive through aggressive non-GAAP framing. The most concerning part is the audit-trail Rule 11(g) deviation — Indian regulators introduced this requirement specifically to prevent retroactive ledger edits, and Harsha is one of the companies the auditor flagged.

Earnings Quality

Reported earnings are recognised in the right period — the income-statement-to-balance-sheet linkages do not look stressed. Debtor days have held in a stable 75-85 day range for five years (FY2025: 78 days), inventory days improved from 187 to 165 days FY23 to FY25, and there is no contract-asset, unbilled-receivable, or aggressive deferred-revenue release line item that would signal pulled-forward revenue. The quality concern is concentration of charges in FY2025 rather than spreading them.

The FY2026 gap is the one to underwrite: revenue grew 15.6% but receivables grew 25.9%, a +10.4 percentage-point gap that drove the working-capital build that crashed FY2026 CFO from $24.1M (FY25) to $7.2M. The same pattern showed up in FY2024 (+12.2pp gap) and was followed by the FY2025 bad-debt write-off in the solar segment. Whether the FY2026 build is real growth, channel-stuffing, or a precursor to further provisioning will show up in the FY2027 receivables and bad-debt line.

FY2025 absorbed $8.7M of consolidated charges (impairment of consolidated goodwill, bad-debt write-off, and provision for doubtful trade receivables), against $0.31M in FY2024 and zero in earlier years. Management presented FY2025 engineering-segment EBITDA two ways on the May 8 2025 call: reported $23.3M and "adjusted" $26.6M (adding back the $3.24M goodwill impairment). The adjustment is defensible — goodwill impairment is genuinely non-cash and event-driven — but the framing risks anchoring investors to the cleaner number while the underlying engineering EBITDA actually declined from $23.7M in FY2024.

The Solar-EPC business is the segment most at risk of a second forensic event: it swung from $0.21M positive EBITDA to $1.66M loss in one year, the bad-debt write-off came from "old sticky debts of two of our major customers" (transcript, May 8 2025), and management has guided to a smaller-project strategy that "does not involve major risk or significant capital allocation". The strategic reset is honest disclosure; the question is whether more sticky debt remains under the ~$10M solar-segment assets carrying value.

Cash Flow Quality

Cash flow tells a different story than the income statement, and the story is mixed. CFO to net income averages 1.15x over FY24-FY26 — superficially healthy — but the FY25 spike to 2.31x and the FY26 collapse to 0.44x are both driven by non-cash adjustments and working-capital swings, not by recurring conversion improvements.

Free cash flow has been negative or near-zero in 3 of the last 5 years. The FY2025 figure of negative $0.2M looks reasonable until you decompose it: CFO $24.1M was inflated by approximately $4.3M of non-cash impairment add-backs and a $2.5M working-capital release; capex was $24.3M including the Advantek greenfield. The FY2026 CFO collapse to $7.2M reflects an $8.4M receivables build and $4.2M inventory build — the working capital tailwind reversed exactly as expected.

The off-balance-sheet picture is the second leg of the cash-flow story. Letter of Credit / Corporate Guarantee / SBLC outstanding rose from $13.1M in FY2024 to $28.7M in FY2025 — a $15.6M increase. The bulk of this is SBLC extended to Citibank Romania, secured by first charge on inventory and receivables of the Romania subsidiary. The Romania subsidiary's carrying value was impaired by $11.1M in the same year. The arithmetic is uncomfortable: India parent guarantees the working-capital lender of a subsidiary that management has admitted needs "significant overhaul or resizing". This is not fraud, but it is a non-trivial contingent exposure that does not appear on the consolidated balance sheet.

Total contingent liabilities more than doubled in one year, from $19.0M to $38.5M. Income tax claims alone rose from $4.0M to $7.3M — the AR note attributes this to "interpretation of income tax law & rules" and quotes counsel that the issues "will not be sustainable in law", but the disclosure is light on specifics and the doubling is too large to ignore.

Metric Hygiene

Management's preferred metrics are reasonable but require investor adjustment. The most prominent reframings appear in transcripts and presentations, not in audited filings.

The single most useful adjustment for an investor is to net other income against reported operating income. FY2026's reported operating income of $25.6M includes $4.2M of other income (16% of opinc) — strip that out and core operating earnings are $21.4M, putting the trailing operating margin at 12.4% rather than the headline 14.7%. Other income comes principally from interest on the IPO proceeds (~$1.7M interest income in standalone FY25) and investment gains (~$1.7M profit on sale of investments). These are real but not from the bearings business.

The ratio fell to 6% in FY2025 because Q4FY25 had negative other income (a one-time charge to other income of ~$2.0M that the transcript does not fully explain — likely a foreign-exchange or investment mark). The reversion to 16% in FY2026 confirms the underlying pattern: 15-17% of reported operating income is genuinely non-operating.

What to Underwrite Next

The forensic risk is a position-sizing limiter, not a thesis breaker. Five concrete items to monitor:

1. Receivables in 1H FY2027. Receivables grew 26% in FY2026 against 16% revenue growth. If 1H FY2027 receivables build outpaces revenue by more than 5 percentage points, the working-capital release of FY2025 looks tactical, and a second round of solar-EPC provisioning becomes likely. If the gap narrows, the FY2026 build was genuine demand.

2. Romania subsidiary updates. The $11.1M standalone impairment was based on a fair-value calculation with WACC of 10.76% and terminal growth of 2.00%. Management's "long-term strategy for Romania" — possible overhaul or resizing — has not been quantified. Watch for either a second-round impairment (signal: customer offtake worsens further) or a write-back (signal: customer demand recovers). The $28.7M SBLC outstanding to Citibank Romania is the contingent exposure that grows if Romania needs more working capital.

3. Audit trail Rule 11(g) status in FY2026. Whether the FY2026 audit opinion still flags the SAP/HANA database-level audit trail as not enabled is the single highest-confidence forensic test. A clean opinion next year is a meaningful upgrade signal; a repeat finding is a downgrade signal.

4. AIA Engineering Ltd related-party sales growth. Sales jumped from $0 to $1.77M in FY2025 (about 1.1% of consolidated revenue). The common board link (Rajendra Shah chairs Harsha and is a Non-Executive Non-Independent Director at AIA Engineering) means this transaction should be scrutinised for pricing — confirm AIA's purchase price is in line with arm's-length bearings cage pricing, and watch whether the $1.77M scales into a larger channel.

5. Contingent-liability disclosures. Income tax claims rose to $7.3M from $4.0M. Specific dispute details are not in the AR. A favourable order would reduce the contingent; an adverse order over $3.5M would be material against $16.5M FY26 PAT.

Downgrade signals (would move score to Elevated/High): another year of audit-trail Rule 11(g) finding; a second Romania impairment; a fresh material write-off in the solar segment; receivables growth more than 10pp above revenue growth in FY27; or a change in auditor without clear cause.

Upgrade signals (would move score to Clean/Watch-low): clean Rule 11(g) compliance in FY2026 audit; Romania impairment write-back; solar segment returns to positive EBITDA without further provisioning; receivables normalise to the 75-day range.

Accounting risk here is a position-sizing limiter, not a thesis breaker or a valuation haircut on its own. Stable debtor days, the absence of revenue-recognition gimmicks, the small share of related-party revenue, the unmodified ICFR opinion, and visible (if uncomfortable) disclosure of the Romania problem all argue against a fundamental accounting concern. But the cluster — audit-trail gap, big-bath year, growing contingent liabilities, off-balance-sheet support of a value-impaired subsidiary, and the FY26 receivables build — is what calls for smaller-than-Kelly sizing and willingness to cut quickly if FY27 disclosures contain a second round of solar bad debt or another Romania charge.