sharma23yash-oss Forensic Red Flags (India)
Indian accounting fraud has a recognisable grammar: cash that exists on paper, receivables that never age out, a subsidiary web, a related-party supplier, a pledged promoter, and an auditor who leaves. This skill scores those patterns.
The output is a risk register with evidence, never a verdict. Say "these five disclosures are inconsistent with management's story", not "this is a fraud."
Part 1 — Quantitative screens
Run all of these. Each needs 5 years of consolidated data.
1. Cash conversion
Cumulative CFO / Cumulative PAT (5y) < 0.6 → flag
CFO / EBITDA (each year) < 0.5 → flag
Profits that never become cash are the base case for every accounting fraud.
2. Working capital drift
Debtor days = Trade receivables / Revenue × 365
Inventory days = Inventory / COGS × 365
Creditor days = Trade payables / COGS × 365
Cash conversion cycle = Debtor + Inventory − Creditor days
Flag when debtor days rise >20% over three years while revenue growth is flat or slowing. Then check the ageing schedule: receivables over 180 days rising as a share of the total, with provision not keeping pace, is the signature of revenue recognised on paper.
3. The Beneish M-Score (8 variable)
M = −4.84 + 0.920·DSRI + 0.528·GMI + 0.404·AQI + 0.892·SGI
+ 0.115·DEPI − 0.172·SGAI + 4.679·TATA − 0.327·LVGI
Where TATA = (Income from continuing operations − CFO) / Total assets.
M > −1.78 → elevated manipulation probability. Treat as a screen, not a conclusion. TATA and DSRI drive most of the signal; if the score trips, read those two components rather than the composite.
Calibration note for India: SGI runs high for genuinely fast-growing companies, so M-Score produces more false positives in Indian small and mid caps than in developed markets. Always pair with the qualitative section below.
4. Altman Z-Score (emerging-market variant)
Z(EM) = 3.25 + 6.56·X1 + 3.26·X2 + 6.72·X3 + 1.05·X4
X1 = Working capital / Total assets
X2 = Retained earnings / Total assets
X3 = EBIT / Total assets
X4 = Book value of equity / Total liabilities
Z < 5.5 → distress zone for an emerging-market non-manufacturer. Do not apply the classic manufacturing Z-Score to Indian banks, NBFCs or asset-light services companies — use bank-nbfc-analysis for lenders.
5. Piotroski F-Score (0–9)
Nine binary tests across profitability, leverage/liquidity and operating efficiency. F ≤ 3 on a statistically cheap stock is the classic value trap profile. F ≥ 7 with rising ROE is the opposite.
6. Accrual ratio
Accruals = (Net income − CFO) / Average total assets
Persistently above 10% is a flag. Rank against sector peers, not an absolute threshold.
7. Cash on books vs interest earned
Implied yield on cash = Other income (interest component) / Average cash & equivalents
If a company reports ₹2,000 cr of cash and earns an implied 1.5% on it while G-Secs yield ~6.8%, the cash is either restricted, encumbered, or not there. This single test has caught more Indian accounting frauds than any ratio.
Corollary: a company that holds large cash and carries expensive debt at the same time owes you an explanation.
8. Capital work-in-progress that never commissions
Track CWIP as a % of gross block over 5 years. CWIP that grows every year and never transfers to PPE is a classic parking place for diverted capex.
9. Tax rate reconciliation
Effective tax rate materially below the statutory rate (25.17% under the 22% concessional regime; ~34.94% under the default regime, both including surcharge and cess) with no disclosed reason — SEZ benefit, brought-forward losses, tax holiday — means reported profit and taxable profit disagree. Read the tax reconciliation note.
10. Auditor fee ratio
Non-audit fees / Total auditor fees
Above ~30% raises independence questions. Disclosed in the notes.
Part 2 — Disclosure and governance signals
These carry more weight than the ratios. Score each present/absent.
Severity 1 — investigate immediately
- Auditor resignation mid-term, or a Big-4/large-firm auditor replaced by a small unknown firm
- Qualified, adverse or disclaimer opinion
- CARO 3(ii) mismatch between quarterly stock statements filed with banks and audited books
- CARO 3(ix) default in repayment, or wilful defaulter tag
- CARO 3(xi) fraud reported by the auditor
- SEBI or NFRA order, adjudication, or show-cause notice against the company, promoters or directors
- Promoter pledge above 50% of promoter holding, or a sharp increase in pledge
- Independent director resigning with a substantive reason in the resignation letter (SEBI requires the full letter be disclosed)
- Delay in filing results, or repeated requests for extension
- Restatement of prior-period figures
Severity 2 — weight heavily
- CFO or Company Secretary churn (two or more in three years)
- Promoter stake sold quietly through the market while management guides for growth
- Frequent changes in accounting policy, estimate, or depreciation method
- A subsidiary web disproportionate to the size of the business, especially overseas
- Related-party royalty or brand fee rising faster than revenue
- Preferential allotment or warrants to promoters at a depressed price
- Repeated equity dilution without a matching increase in return on capital
- Pledged shares invoked by a lender
- Change of registered office or name that obscures a prior history
Severity 3 — note and monitor
- Very high receivable concentration in a single customer
- Loans and advances to entities with no disclosed commercial relationship
- Aggressive capitalisation of interest or development costs
- Segment reporting that is merged or restated to hide a declining line
- Contingent liabilities > 50% of net worth
- Consistently low dividend/buyback despite large reported cash
Part 3 — Where to verify
| Signal | Primary source |
|---|---|
| Filings, results, announcements | NSE and BSE corporate announcements |
| Shareholding, pledge | Quarterly shareholding pattern on NSE/BSE |
| Enforcement, orders, settlements | SEBI orders and adjudication pages |
| Audit quality actions | NFRA orders |
| Subsidiary financials, charges | MCA21 — company master data, index of charges, AOC-4 |
| Insolvency proceedings | IBBI public announcements |
| Defaults, credit view | Rating agency rationales (CRISIL, ICRA, CARE, India Ratings) — a rating downgrade rationale often says out loud what the AR does not |
| Litigation | Company's own contingent liability note; e-courts for larger matters |
Rating rationales are the most under-used forensic source in India. Agencies see monthly no-default statements, bank facility utilisation, and management interactions that public investors do not.
Part 4 — Scoring and output
Assign a composite:
- Severity 1 present → Red. Thesis is unsafe regardless of valuation.
- ≥3 Severity 2, or 1 Severity 2 + weak cash conversion → Amber. Requires a satisfactory management answer before sizing.
- Only Severity 3 → Watch. Note in the file, re-check each quarter.
# <Company> — Forensic Screen
**Composite: RED / AMBER / WATCH**
## Quantitative
| Test | Value | Threshold | Flag |
|---|---|---|---|
## Disclosure signals
| Signal | Severity | Evidence (source + date) |
|---|---|---|
## What would change this view
1. ...
## Questions requiring a management answer
1. ...
Hard rules
- Every flag cites a document. Filing name, date, page or note number. No flag without a source.
- Never state or imply that fraud has occurred. Describe the inconsistency and let it stand. Distinguish "the disclosures are inconsistent" from "management is dishonest" — you can evidence the first, never the second.
- Do not repeat allegations from social media, anonymous short-seller posts, or forums as fact. They can be a reason to look; they are not evidence.
- A high score is a reason to research, not to short. Say so.
- This skill produces research inputs. It is not investment advice, and the user should be told so in the output.