Earnings Management and Accounting Fraud: Where the Line Runs

Why the line between earnings management and fraud matters to every owner

Shortly before the meeting with the bank, preliminary earnings are weaker than hoped. The tax adviser names options: set a provision a little lower, hold off on writing down a receivable, issue an invoice only in January. None of this sounds like fraud. Still, a question hangs in the air that every owner asks sooner or later.

Large accounting scandals rarely began with an open lie. They began with latitude that was used a little further each year. This article therefore pursues one question: Where does permitted earnings management end, where does accounting fraud begin, and how can an owner tell which side their own decisions are on?

To answer it, we first clarify the terms, then present the continuum by Dechow and Skinner as a thinking model, look at research findings and two readings, and finally derive a boundary test for your own financial statements.

What is earnings management, and what is accounting fraud?

Earnings management, in the German sense of Bilanzpolitik, is the purposeful shaping of the financial statements within the applicable rules. It uses three levers. Accounting options are explicitly permitted alternatives, such as whether internally generated intangible assets like in-house software are capitalized. Judgment arises because every valuation requires estimates: useful life, default risk, the size of a provision. Transaction structuring changes not the valuation but the business itself, for instance through the timing of an investment, factoring, or selling a property and leasing it back.

Research uses the term earnings management in a narrower sense. Healy and Wahlen (1999), as cited by Dechow and Skinner (2000), describe it as judgment in financial reporting and in structuring transactions that alters financial reports either to mislead stakeholders about the underlying performance or to influence contractual outcomes that depend on reported numbers. What matters in this definition is intent, not the tool.

Accounting fraud is clearly defined in German law. Under Section 331 of the German Commercial Code (HGB), a member of the management or supervisory board of a corporation commits a criminal offense if they misrepresent or obscure the company's position in the annual financial statements or management report. The penalty ranges up to three years in prison. For a false responsibility statement at capital-market-oriented companies, Section 331a HGB has provided for up to five years since the Financial Market Integrity Strengthening Act (FISG) of 2021.

The thinking model: the continuum by Dechow and Skinner

Patricia Dechow and Douglas Skinner (2000) arrange accounting choices along a line instead of splitting them into permitted and forbidden. On one side are decisions within the rules, on the other decisions that violate them.

The authors stress two points. Between the first three zones and the fourth there is a conceptual line: fraud violates the rules and intends to deceive. Within the rules, however, earnings management is hard to distinguish from the legitimate exercise of judgment as long as the intent is unknown. In addition, there are real decisions outside accounting, such as postponing advertising or maintenance to the next year. They break no rule, but change the picture just as much.

An analogy makes this tangible. Anyone photographing an apartment for sale may choose good light and a wide-angle lens. The picture is flattering but true. Anyone who retouches the crack in the wall shows an apartment that does not exist. In between, the question is: would the buyer feel deceived if they knew how the shot was framed?

Earnings management does not become fraud in a single day. It drifts, one year after another, from neutral to aggressive, until the result can no longer be sustained.

What research shows about earnings management

Dechow and Skinner (2000) describe a contradiction: regulators and practitioners consider earnings management widespread and harmful, while research long found little evidence, because its statistical methods are poor at identifying individual cases. The authors conclude that one should look less at measurement models and more at incentives, above all those of the capital market: avoiding losses, avoiding earnings declines, meeting analysts' expectations.

The strongest indication comes from the distribution of results. Small losses are unusually rare, small profits unusually common. From this, Burgstahler and Dichev (1997), as cited by Dechow and Skinner (2000), estimated how many companies just below a threshold lift their earnings above it.

These figures apply to listed US companies in the 1990s and are estimates derived from the distribution, not proven individual cases. But they show where the pressure lies: at thresholds that readers of the accounts take seriously. In the Mittelstand, these are loan conditions, bonus rules, or breaking even.

More recent work sharpens the distinction. Benzine (2025) compared an industrial and a trading company in Algeria using interviews with auditors. External pressure to enhance the company's image led there to creative accounting using the latitude in the standards. Internal pressure, above all from taxes, led to fraud with clear rule violations. The difference lay in management's intent. Two case studies are not statistics. Arifian and Chariri (2025) reviewed 74 journal articles on the gray area and note that research mostly examines either fraud cases or earnings management, but rarely the transition between them.

Two readings: information or a precursor to deception?

The judgment depends on what one takes financial statements to be for.

The first reading is supported by the point of Dechow and Skinner (2000) that accrual accounting makes earnings more informative precisely because it smooths. Smoothing is not suspicious in itself. The second reading is supported by experience from scandals. At Enron, it began with valuing long-term contracts at estimated market values, a permitted method that was filled with ever more optimistic assumptions. The shared patterns are discussed in Patterns of major accounting scandals.

I consider both readings correct, but for different readers. For an informed analyst, disclosed earnings management is information. For a bank that writes a ratio into a loan agreement, it is a risk, because the ratio loses its purpose. It follows that the line runs not only between permitted and forbidden, but also between disclosed and hidden.

Where the line runs in German law

The HGB sets the framework with a few principles that limit any earnings management.

Consistency is especially important for the Mittelstand, Germany's small and mid-sized companies. Using an option once is earnings management. Turning it every year toward wherever earnings are needed violates the principle and is a typical warning sign for banks and auditors. Companies subject to a statutory audit face these questions every year, as Statutory audits in the Mittelstand describes.

Even permitted structures can cross the line if their only purpose is to create a false picture. Shortly before quarter ends, Lehman Brothers booked short-term repurchase agreements as sales and thus lowered its reported leverage. How the case was assessed is described in Lehman Brothers 2008.

The boundary test: five questions for every accounting decision

From the continuum and the principles of the HGB, I derive a simple test that I call the boundary test. It does not replace a professional review, but it shows where a decision lies on the line.

1. Rule: is it permitted?

   Name the option or estimate on which the decision is based. If there is none, the question is answered.

2. Consistency: do I do it this way every year?

   Check whether the method and assumptions are the same as in the previous year. A change needs a factual reason, not an earnings target.

3. Openness: would I explain it to the bank?

   Imagine your bank asks about exactly this item. If the honest answer would be uncomfortable, the decision lies in the aggressive zone.

4. Intent: would I do it the same way without a target?

   Ask yourself whether you would make the same estimate if no loan agreement, no bonus, and no threshold depended on it.

5. Materiality: does it change the reader's decision?

   Calculate whether the sum of the decisions lifts a ratio above or below a threshold that the bank, buyers, or shareholders take seriously.

A worked example shows how quickly small steps add up. A business has preliminary earnings of 40,000 euros. In the loan agreement, the bank requires positive earnings of at least 100,000 euros.

The sum fails the boundary test, because the intent is obvious. How banks read such items is described in How banks decide on loans.

Transferability and limits

Earnings management is not a fringe phenomenon of large corporations. Every set of financial statements needs estimates, and every estimate can be pushed in one direction. The lesson from research is to watch incentives rather than methods: where a number decides on a loan, a bonus, or a purchase price, it will come under pressure. Anyone reading other people's figures, such as a customer's before a large order, should ask questions exactly there, as Reading a customer's balance sheet shows.

The limits of these statements are clear. Most studies examine listed companies, often in the US. For small companies reporting under the HGB, other incentives come first: taxes, the bank, and the family. And intent cannot be measured from the outside; often only hindsight shows whether an estimate was neutral.

Conclusion: the line runs at intent and openness

To the guiding question: permitted earnings management ends where the rules are violated or where latitude is used to create a picture that contradicts the actual situation. Legally, the line to fraud is clearly drawn. Economically, before it lies a broad zone of aggressive accounting that may be permissible and still costs trust as soon as a reader discovers it.

An owner recognizes their side not by the individual entry but by three features: whether the decision is consistent, whether they would explain it openly, and whether they would have made it without an earnings target. Anyone who answers these questions honestly stays in the neutral zone, even in a weak year.

• List the five largest estimates in the last financial statements with your tax adviser and compare them with the previous year.
• Go through the boundary test for each item and note the result.
• Check which ratios in loan agreements, bonus rules, or shareholder agreements depend on balance sheet figures.