The reporting-quality spectrum, before the taxonomy
The word "shenanigan" implies a binary, honest or dishonest, and almost nothing you will meet is binary. Financial reporting quality runs along a spectrum, and putting a company on it is the first judgment you make, before any test fires. Six positions, from the top down:
| Position | What it looks like | Your response |
|---|---|---|
| 1. Decision-useful and sustainable | compliant reporting, and the earnings recur | value it; the accounts are an input you can use as given |
| 2. Decision-useful, not sustainable | compliant reporting of genuinely low-quality earnings: a cyclical peak, a one-off gain, a windfall | normalise the level before valuing anything (M2.05 §4.6) |
| 3. Biased choices within the rules | aggressive or conservative estimates and policies, disclosed, defensible one at a time | adjust to a common basis and count the direction of the bias |
| 4. Earnings management | the choices are made to hit a number: smoothing, timing, real-activities management such as cutting discretionary spend at quarter-end | adjust, and reprice management's character |
| 5. Departures from the rules | non-compliant accounting, restatement territory | the accounts are not an input; the position is a governance judgment |
| 6. Fictitious transactions | revenue, assets or counterparties that do not exist | no adjustment is possible; verification logic replaces accrual logic |
Two things follow from the shape of that list. Positions 1 and 2 are about the earnings, and positions 3 to 6 are about the reporting. A company can report impeccably on a business whose profits won't recur. That's still a quality problem, but you solve it with normalisation, and suspicion doesn't come into it.
The spectrum also descends by intent. Size doesn't set the order. A ₹50 crore fictitious sale sits below a ₹5,000 crore aggressive-but-disclosed depreciation policy. The first tells you what the management will do. The second tells you only what it prefers.
Advertisement
Which accounting choices can move reported profit?
Position 3 gets its own table. It's where you'll spend most of your working life, and the set of choices is finite. Each row below is a real, legal fork that management takes, and you've already met the mechanics of every one of them earlier in the course.
| The choice | Aggressive direction | Conservative direction | Taught in |
|---|---|---|---|
| Depreciation life and method | long lives, straight line, high residual value | short lives, accelerated or written-down-value, nil residual | M1.07 §5 and §6 |
| Capitalisation threshold | low threshold, development costs and borrowing costs and software capitalised | expensed as incurred | M1.07 §1 and §4 |
| Inventory cost flow (US filers) | FIFO while input prices rise | LIFO while input prices rise | M1.06 §4.7 |
| Receivable allowance and expected credit loss | low loss rates, long ageing buckets, optimistic forward-looking overlay | higher rates, tighter buckets | M1.06 §4.6 |
| Warranty and other provisions | low provision rate, late recognition, "not probable" | full expected-value provisioning, early recognition | M1.08 §4.3 |
| Discount rate on long-dated liabilities | high rate, so the pension or decommissioning liability shrinks | low rate, so it does not | M1.08 §4.3 and M1.12 |
| Revenue recognition timing and gross-versus-net | earlier, over-time, gross as principal | later, point-in-time, net as agent | M1.06 §§4.2 to 4.4 |
Read the table in both directions. An aggressive choice pulls profit forward and leaves a residue on the balance sheet. That's the master law of detection. A conservative choice pushes profit into the future and builds a reserve that can be released later, and that's how the cookie jar gets stocked in the first place.
Conservatism is still a bias. It's decision-useful only if it's consistent. A company that provisions heavily in good years and releases in bad ones is managing earnings just as surely as one that does the reverse. What you want is neutrality. What you can actually get is a company whose bias you can name, measure and reverse.
Why do managers manipulate earnings?
Start from an uncomfortable truth you already half-know from your first journal entries. Accrual accounting is a system of judgments. Revenue is recognized when performance obligations are "satisfied," which is a judgment. Depreciation spreads cost over a "useful life," which is an estimate. Provisions are booked when outflows are "probable," a probability call. Receivables are carried at what will "likely" be collected, a forecast. Every one of those judgments is made by the people whose bonuses, jobs, and reputations depend on the answer.
Most managers, most of the time, make them honestly. The minority who don't are the subject of everything that follows, and you can't tell whether you're holding one of them until you test the numbers.
The pressures, concretely:
- Compensation. Bonus plans pay on EPS, revenue growth, or EBITDA targets; option and RSU grants pay on the stock, which prices the reported numbers. Munger's rule, "show me the incentive and I will show you the outcome", is a forensic instruction, so treat it as one. Your first stop in any quality-of-earnings review is the proxy statement (US, DEF 14A) or the Board's Report / remuneration disclosures (India). Find the metrics that trigger management's pay, then watch those metrics for manipulation.
- Covenants. Loan agreements set tripwires: net debt/EBITDA below 3.5×, interest coverage above 2×, minimum net worth. Breach can mean default, repricing, or collateral calls. A company drifting toward a covenant has a quantifiable motive to inflate EBITDA this quarter. India adds a sharper version: working-capital limits are sanctioned against drawing power, a percentage of receivables and inventory declared monthly to the bank. Inflate receivables and you flatter the P&L, and you also raise your borrowing limit. Hold on to that one. The disclosure that polices it turns up in promoter related party transactions.
- Capital-raising windows. An IPO, QIP, rights issue, or big block sale prices off recent numbers. Filings covering the four to eight quarters before a raise deserve double suspicion. Luckin's fabricated 2019 quarters sat between its IPO and a follow-on offering.
- The meet-or-beat culture and the smoothness cult. Markets punish small misses savagely and reward metronomic delivery with premium multiples, so managements manage the path, not just the level. General Electric under Jack Welch was the model of uncanny quarterly regularity. In 2009 GE paid $50m to settle SEC charges that included end-of-quarter revenue acceleration and hedge-accounting misuse (settled without admission), and in 2020 it paid another $200m over disclosure failures in insurance and power. GE wasn't uniquely bad. The lesson is that suspiciously smooth is a red flag in itself. Real businesses are lumpy. A company that beats consensus by one paisa sixteen quarters running is telling you about its accounting flexibility, not its operations.
- India-specific: the promoter pledge spiral. When promoters have pledged shares as loan collateral (you've already learned to read the pledge disclosure in the shareholding pattern), a falling stock price triggers margin calls, which force share sales, which drop the price further. A pledged promoter has an existential incentive to keep both reported earnings and the stock aloft. High pledge percentage plus deteriorating cash conversion is one of the highest-yield red-flag combinations in Indian markets (Zee's January 2019 crash was a pledge unwind meeting governance questions).
The fraud triangle (Donald Cressey's framework, used by auditors everywhere): manipulation needs pressure (the incentives above), opportunity (weak boards, pliant auditors, complex group structures, dominant CEOs/promoters), and rationalization ("it's temporary," "everyone does it," "we'll catch up next quarter"). Satyam's Ramalinga Raju gave the canonical description of rationalization's endgame in his confession letter: "It was like riding a tiger, not knowing how to get off without being eaten." Almost no one sets out to commit a ₹5,000 crore fraud. They set out to paper over one bad quarter, and the accrual mathematics of financial shenanigans does the rest.
Earnings management vs fraud
The continuum matters for your language and your judgment. There is a ladder from (1) choice within GAAP (choosing FIFO, choosing a defensible 10-year life) to (2) aggressive estimates (the 10-year life becomes 15 while peers use 8) to (3) misclassification (a recurring gain dressed as operating revenue) to (4) fabrication (invoices for sales that never happened). Rungs 1–2 are earnings management. They're legal and common, and they still corrode the analysis. Rungs 3–4 shade into securities fraud.
As an analyst you rarely need to prove which rung you're on. Your job is to adjust the numbers to reality and to price the management's character. In writing, say "aggressive revenue recognition" or "low-quality earnings" unless a court, regulator, or the company itself has said "fraud." That habit protects your accuracy, and it protects you.
One more piece of mental furniture: economist J.K. Galbraith's "bezzle", the inventory of undiscovered embezzlement in an economy. It grows in booms (audits are lax, everything rises, nobody checks) and is discovered in busts. So fraud revelations cluster after market peaks (2001–02: Enron, WorldCom; 2008–09: Satyam, Madoff; 2018–20 in India: IL&FS, DHFL, Cox & Kings, CG Power). When the tide is high, raise your suspicion. You'll meet the same idea again in Buffett's swimming-trunks line.