Earnings Quality and Red Flags

The accrual gap, and how to measure it

Earnings quality is a relationship rather than a number: how much of the profit the company reports has become cash. Compute it as cash from operations over net income — the conversion rate — and read it over years rather than quarters. Near 100% over a full cycle is the ordinary signature of an honest reporting habit. Persistently well below that, while the balance sheet accumulates receivables and inventory, means the profit is a claim the future has to settle. Two adjustments make the measure honest. First, use cash from operations rather than free cash flow, because capex is a decision about the future rather than a quality problem in the past — a capital-intensive grower will look bad on free cash flow conversion and perfectly healthy on operating conversion. Second, watch the size of the accrual balance, not just the latest rate: the accumulated difference between profit and operating cash is a stock, and a stock that has grown for five years is the one that gets written off. The chart in this lesson draws exactly that, and the workbench measures it in dollars and in days. Days sales outstanding is the earliest hard evidence. Receivables over revenue, times 365, gives the average collection period, and it does not care what management says about the seasonality of its billing. When DSO rises while revenue grows, the company is either selling to customers who pay more slowly — which is a real business decision with a real cost — or it is recognising revenue earlier than the eventual cash supports. Neither is provable from one number, and both are the reason DSO appears first in most short-seller reports. Three years, and a $185m gap — Profit, years one to three: $280m, $340m, $420m · Cash from operations: $320m, $295m, $240m · Conversion by year three: 57% · Days sales outstanding: 57.0 days rising to 73.0 days ← · Cumulative profit not converted to cash: $185m A single year of low conversion is not a flag. A company financing a new product launch, or one whose customers are seasonal, will show it. The flag is a trend plus a mechanism you can name.

The other three flags

Adjustments. Every non-GAAP reconciliation starts with reported profit and adds back items the company considers unusual. Some are genuinely one-off: a settlement, a one-time charge for closing a plant. Others recur annually and are simply renamed — "restructuring" for the fourth year running, or "integration costs" three years after the acquisition closed. The test is mechanical rather than sceptical: look at the same adjustment four years in a row. If it appears every year, it is a cost of doing business and the adjusted figure is not a measure of the business. Share-based compensation. It is a genuine non-cash charge, which is why it is added back on the cash flow statement, and it is also a real cost, which is why the diluted share count rises every year to account for it. A company that excludes it from adjusted earnings while counting the shares it creates is reporting a cost it does not pay in cash and does pay in dilution. The size of the gap is worth reading every year, because it tells you how much of the employee bill is being settled in shareholders’ ownership. Revenue timing and the balance sheet places that hide it. Deferred revenue rising fast is usually good news — customers have paid for something not yet delivered — until it reverses, at which point the revenue that has been propping up the margin disappears. Capitalising costs that a peer expenses is the same technique in a different line: software development, customer acquisition, or interest, moved from the income statement to an asset, lifting profit now and creating an asset test in the future. When you compare two companies in the same industry, read the accounting policy note on capitalisation and revenue recognition before you compare their margins at all. Access to the qualitative evidence is what separates the two readers. The earnings call, the prior-year guidance and the insider transaction record are all public, and a company that missed its own guidance while the CFO sold shares is giving you a different signal from one that reiterated it.

The forensics, ranked by how early they appear

Earnings problems do not arrive without warning; they arrive while everyone is looking at the headline. What follows is the same list a short seller works through, ordered by how early each signal shows up relative to the collapse. **Earliest: the divergence between sales growth and receivables or inventory growth.** If receivables grow faster than revenue for two consecutive quarters, the company is either selling to customers who cannot pay or recognising revenue it has not collected. Inventory outgrowing sales says the same thing about the warehouse. Both are visible a year before the write-down, and both are in the 10-Q you already have. **Middle: the accrual gap and the adjustment habit.** Net income rising while operating cash flow falls is the mechanical signature of aggressive accruals. Alongside it, watch how often a company reports adjusted figures that exclude a cost which recurs every single quarter — a restructuring charge that never ends is a cost of doing business, not a one-off (F6). **Latest: the day-count and the guidance language.** Days sales outstanding creeping up, or a change in how a metric is described, are the last clean signals before the release that surprises everyone. • Receivables or inventory growing faster than sales, two quarters running • Net income up while operating cash flow falls — the accrual gap • Recurring one-offs in the adjusted numbers, quarter after quarter • Days sales outstanding creeping up with no change in the customer base • A metric redefined, or guidance moved from a number to a range • Insider selling accelerating into strength, which is context rather than proof None of these is proof on its own, and every one of them has an innocent explanation. What matters is the count: three of these together describe a company whose reported profit is travelling further from its cash with each quarter.

The signals that sit outside the statements

Everything in this lesson so far is computed from the numbers a company chose to publish, and the most reliable warning signs are often events rather than ratios. Three of them are worth memorising because they are public, dated and hard to explain innocently. The first is an **auditor change that is not a routine rotation**, especially a resignation rather than a re-tender, and especially one followed by a “disagreement” disclosure. The second is a **CFO departure** in the weeks around a reporting date, particularly when no successor is named and the reason is “to pursue other opportunities”. The third is a **late filing**: a company that misses its own regulatory deadline is telling you something about its own controls that no ratio will. A second family sits in the related-party section of the notes, which is the part most readers skip. Loans to executives, sales to a company owned by a director, consulting arrangements with entities connected to insiders, or acquisitions of businesses owned by the founding family are all legal and sometimes sensible. They are also the transactions where the interests of the people running the company and the interests of its shareholders can diverge without anyone breaking a rule, and their presence raises the standard of evidence you should require for everything else. The same is true of a **restatement**, which is not a red flag but a fact: the prior numbers were wrong, and the question is only whether the cause was a mistake or a choice. The practical use of all of this is to treat forensic analysis as a filter before a valuation rather than a substitute for one. A company with clean cash conversion, conservative recognition, stable auditors and no related-party entanglement can be valued with normal scepticism. A company with a widening accrual gap and a CFO who left in the last two quarters does not need a lower multiple; it needs a decision about whether to own it at all, because the analysis you can do from outside will never settle the question. • Auditor resignation, CFO departure near a reporting date, and a late filing are public, dated and hard to explain innocently. • Related-party transactions are legal and are where insider and shareholder interests can diverge quietly. • A restatement tells you the prior numbers were wrong; the question is only why. • Run forensics as a filter, not as a substitute for valuation.

The three day-counts that catch a stretched book

Accrual accounting is the subject, and the fastest useful test of it is not a complicated forensic score. It is three day-counts, each of them a ratio between a balance-sheet figure and an income-statement figure, and each of them read as a *trend* rather than as a level. **Days sales outstanding** is receivables divided by revenue, scaled for the length of the period. It measures how long the company waits to be paid. The level varies enormously by industry and by the terms a company grants, so the informative comparison is against the company’s own history and against revenue: receivables growing faster than sales means either collection is slowing or the sales recognised at the end of the period were more generous than the ones before them. A sharp rise in this count in a fourth quarter that also happens to be unusually strong is the classic shape of a pull-forward, and it is the earliest accrual warning available. **Days inventory outstanding** is inventory divided by cost of goods sold. It rises when goods sit longer, and it rises for two quite different reasons: demand has slowed, or production continues so that factory overhead stays absorbed in inventory rather than running through the income statement. The second is a decision rather than a slowdown, and it shows up first here. **Days payable outstanding** is payables divided by cost of goods sold, and it is the mirror. A rising number means the company is paying suppliers later, which is a genuine source of cash and a finite one — a supplier can be stretched once, and a company that has been funding itself this way has borrowed against a relationship rather than from a bank. Read together, the three produce the **cash conversion cycle**: the receivable days plus the inventory days minus the payable days. That number is how many days of funding the operating cycle consumes, and it converts the question “is this growth real?” into a question with an answer. A company growing revenue while its cycle lengthens is consuming cash to grow, and that combination — profits rising, cash falling, cycle stretching — is the most common reason a profitable and fast-growing business runs out of money. Two cross-checks keep the reading honest. Read the allowance and the write-offs beside the receivable days, because a company that stops adding to its allowance while collection slows will report flattering earnings in the near term and the opposite later. And remember that a change in revenue recognition, a shift in segment mix or an acquisition can move all three counts without anything deteriorating — which is why the counts are a reason to ask a question, not a verdict. • Read each day-count as a trend against sales, not as a level against a threshold. • Receivables growing faster than revenue is the earliest accrual flag available. • Inventory days rise either because demand slowed or because overhead is being absorbed. • The cash conversion cycle turns “is this growth real?” into a number that can move. Pull the three counts for the last eight quarters and plot them on one chart. The level of each is unremarkable; a step change in any of them, in the quarter before a miss, is the part worth remembering.

Sources

• Accruals and the empirical link between high accruals and lower subsequent returnsSloan (1996), "Do Stock Prices Fully Reflect Information in Accruals and Cash Flows About Future Earnings?"
• Revenue recognition under ASC 606 and the timing judgements it permitsFASB ASC 606; SEC Staff Accounting Bulletin 104
• Non-GAAP measures and the SEC’s conditions on presenting themSEC Regulation G; SEC Compliance and Disclosure Interpretations on non-GAAP measures
• The BENEISH M-score and other scoring approaches to earnings manipulationBeneish (1999), "The Detection of Earnings Manipulation"