Earnings Quality Checklist

Why the number on the bottom line can be wrong

Net income is built on accrual accounting, and accrual accounting runs on estimates. The IRS is not involved in the choices. Auditors review them, but they don't make them. Management makes them. And management has real financial reasons — stock price, executive compensation, debt covenants, analyst expectations — to push those estimates in a favorable direction.

The specific choices include: when to recognize revenue on a long-term contract, how much to reserve for expected bad debts, whether a software development cost gets expensed this quarter or capitalized and spread over five years, how quickly to depreciate a factory that might be worth less than the books say. None of these decisions is fraudulent by default. Every one involves a judgment call between two defensible positions. Accounting standards give management a range. Most managers fill it in honestly. A material minority fill it in strategically.

This is different from fraud. Fraud is filing numbers you know to be false. What this checklist tracks is legal earnings management — the deliberate use of accounting discretion to produce a better-looking income statement than the underlying cash flows would generate on their own. Richard Sloan's 1996 study at the University of Michigan found that companies with high accruals — large differences between reported income and cash flow — consistently underperformed companies with low accruals over the following year. The market was overpricing paper earnings. That pattern has held in subsequent academic research and in the analysis of nearly every major accounting scandal in the past 30 years: the cash flow statement saw it coming before the income statement gave anything away. The five signals below don't require an accounting degree. They require a 10-K and a calculator.

The five signals worth checking

1. Operating cash flow persistently below net income. The cash flow statement is the least manipulable of the three financial statements. Revenue recognition requires judgment calls — when is a contract "substantially complete"? How much of an up-front license fee belongs in this quarter? Operating cash flow bypasses most of those questions. Cash received from customers is cash received. Cash paid to suppliers is cash paid. The difference is what the business actually generated.

Divide operating cash flow by net income for each of the past three years. A ratio below 0.80 in any single year has explanations. Below 0.80 for two consecutive years is a pattern. Below 0.80 for three consecutive years is the most reliable signal on this list. GE's Power segment ran a cash conversion ratio below 0.70 for six consecutive quarters before the $22 billion goodwill impairment charge in 2018. The cash flow statement was publishing a warning the income statement wasn't. Six quarters is enough time to act on it. The direction matters more than any single reading. A company that reliably converts 92 cents of every dollar of net income into operating cash flow for five years and then drops to 55 cents in one quarter has a story to explain.

2. Receivables growing faster than revenue. Accounts receivable represents sales the company has recognized but hasn't collected yet. When receivables grow faster than revenue, the company is booking revenue before customers pay for it. That can happen for legitimate reasons — a large contract with net-60 payment terms, a shift toward enterprise buyers. It can also happen because sales are being pulled forward, terms are being loosened to get deals across the quarter-end line, or some of the "sales" will never convert to cash.

Sunbeam in 1997 is the textbook case. Under Al Dunlap, the company used "bill and hold" arrangements to ship goods to distributors and warehouses that had no immediate obligation to pay. Revenue was recognized at shipment. Receivables grew more than twice as fast as revenue. The stock fell 83% in five months when the scheme unraveled in 1998. The signal was on the balance sheet the whole time. The rule: take year-end accounts receivable and year-end revenue for both the current and prior year. If the receivables growth rate exceeds the revenue growth rate by more than 5 percentage points, find the explanation in the footnotes before you proceed.

3. Gross margin trend hidden by SG&A cuts. Gross margin reflects the basic economics of making and delivering the product. It's hard to manipulate without fabricating supplier invoices. But it can erode quietly over years while the company cuts sales, general, and administrative expenses to keep net income flat. The investor who watches only the bottom line sees stability. The investor who checks gross margin separately sees a deteriorating core business being papered over with cost-cutting that has a finite life.

This played out across several US department store chains between 2015 and 2019. Multiple companies reported reasonably steady net income while gross margins contracted 3–5 percentage points over three years, masked by continuous SG&A reduction programs. Eventually there was nothing left to cut. The gross margin deterioration hit the bottom line undisguised. Stocks fell 50–70% in a compressed period. Anyone tracking gross margin separately had 18 to 24 months of warning. Check gross margin and operating margin independently for each of the past four years. If gross margin is contracting while operating margin holds flat, find where the savings are coming from. Cuts to advertising and R&D eventually hit revenue too.

4. Recurring "one-time" charges. A restructuring charge is legitimate once. It signals that the business is adjusting its cost structure — closing plants, writing off inventory, settling litigation. The third one in four years signals something else: the accounting treatment has become load-bearing. The company is moving regular operating costs into the "non-recurring" category to protect the earnings number analysts model.

IBM reported restructuring charges in 18 of the 20 fiscal years from 2000 to 2019. Total restructuring expense over that period exceeded $10 billion. Whether or not each charge technically met the accounting standard for non-recurring treatment, the aggregate was plainly a permanent cost of running IBM. An investor who excluded those charges was using a number that was structurally too high, every year, for two decades. Count how many of the past five annual filings contain a material charge labeled as restructuring, impairment, special items, or non-recurring. Three or more is the flag.

5. Auditor changes or restatement history. A voluntary auditor change — particularly from a Big Four firm to a smaller one — is worth investigating. Most of the time the reason is mundane: cost, a new CFO, a change in audit committee relationships. Sometimes the prior auditor pushed back on accounting judgments that management didn't want to defend, and the company chose a different firm rather than changing the accounting. The prior auditor disclosure appears in the proxy statement (DEF 14A on EDGAR), not the 10-K. It takes two minutes to check. Restatements are even easier: search the company's EDGAR filings for 10-K/A or 10-Q/A amendments. Every restatement includes a management explanation of what changed and why.

GE amended its financials in 2009, 2018, and 2020, each covering different segments and time periods. Each restatement was presented as narrowly contained. A company whose accounting is sound doesn't need to retroactively re-describe what happened three years ago. The pattern of needing to re-file the same statements multiple times over a decade is diagnostic in itself.

Run it on a company

The checklist above takes five inputs from any company's most recent 10-K. Operating cash flow appears on the first page of the Statement of Cash Flows, labeled "net cash provided by operating activities." Net income is the bottom line of the income statement. Accounts receivable sits in the current assets section of the balance sheet. Revenue is the top line of the income statement. Gross margin requires two numbers from the income statement — net revenue and cost of goods sold. All of it is in the same document.

Pull three years of data where you can. SEC filings provide two years of comparative balance sheet data and three years of income statement and cash flow data in a single 10-K — no additional research required. One year gives you a snapshot. Three years tells you whether you're looking at a stable pattern or an anomaly. For the one-time charges question, scan four or five annual filings for any line labeled restructuring, impairment, litigation settlement, or special items. For the auditor question, open the DEF 14A proxy for the same company on EDGAR — the firm and fees are disclosed there.

What your score means

Zero or one flag is normal. Every company has at least one year where operating cash flow runs below net income, or where a genuine restructuring charge appears. One flag with a clear explanation in the MD&A is noise. It doesn't change your analysis unless subsequent quarters confirm it.

Two or three flags in the same direction is a pattern worth tracing to the footnotes. If the company is flagging on both cash conversion and receivables growth, those two signals often have a common cause — revenue being pulled forward, collection terms being loosened to get deals done before quarter end. Read the accounts receivable footnote and the revenue recognition policy note in that specific filing. If the MD&A addresses it clearly, you've found your explanation. If it doesn't, that silence is data.

Four or five flags is a pass-or-explain situation. Either find a clear structural reason the pattern is normal for this company's industry (covered below), or treat the income statement as unreliable until you can account for each flag using language from the actual filing. Some investors pass automatically at four flags. Others treat it as the beginning of a more intensive process — calling investor relations, reading every audit opinion for the past five years, comparing ratios to direct competitors. Either is defensible. Assuming the flags will resolve without looking is not.

When to adjust your expectations

Three categories of companies look alarming on this checklist for structural reasons that have nothing to do with earnings manipulation.

Early-stage companies burning cash deliberately will fail the cash conversion ratio every year by design. A company in year three of a sales-force buildout, investing $40 million per quarter to acquire customers who generate revenue over a five-year relationship, will show operating cash flow well below net income — and that is consistent with rational capital allocation, not accounting games. Apply the checklist to early-stage companies as a tracking tool to watch the trend over time, not as an absolute threshold. The flag becomes worth investigating when the gap is growing as the business scales rather than shrinking.

Insurance companies and banks have accrual structures that look alarming until you understand the accounting. An insurer's loss reserves are a management estimate with no equivalent in manufacturing. A reserve release in a light-claims year flows through income without any corresponding cash. A reserve build in a heavy-claims year reduces income without any cash leaving the building. The cash conversion ratio for a well-run insurer in a large-catastrophe year can drop below 0.50 and mean nothing beyond that year's claims experience. Compare insurance companies to insurance company peers over the same underwriting cycle — not to the generic 0.80 threshold.

Toll roads, water utilities, and pipelines have structurally low cash conversion because they service large debt loads and depreciate asset bases over multi-decade schedules. These businesses are often better analyzed on funds from operations or distributable cash flow than on raw operating cash flow to net income. A 0.70 conversion ratio for a regulated water utility might be its long-run norm. Before applying the threshold to any of these businesses, check what the company's own ratio has been over the past five years. If it has been consistently at 0.70 and management explains it clearly, the flag does not apply. If the ratio was 0.95 three years ago and is now 0.70, the flag applies regardless of industry.