Beyond Accruals: Why Traditional Earnings Quality Factors Miss Accounting Risk

The limits of a useful signal

Accruals have long occupied an important place in quantitative earnings-quality research. An accrual-based signal distinguishes the accounting component of earnings from the cash-flow component, offering a lens on the persistence and possible reliability of reported performance.

Richard Sloan's influential 1996 research examined whether prices fully reflected the different implications of accrual and cash-flow components for future earnings. Subsequent work by Richardson, Sloan, Soliman and Tuna connected differences in accrual reliability to differences in earnings persistence.

These are important foundations. But they do not mean that a single accrual ratio contains everything an institutional analyst needs to know about reporting quality.

A company’s economic exposures may be shaped by contract terms, revenue estimates, working-capital financing, contingencies and changes in accounting judgments. Such information may be disclosed in footnotes or management discussion yet remain poorly represented in a standardised factor dataset.

An accrual score is a screening variable, not an explanation of the business model.

What accrual-based earnings quality factors actually measure

A common starting point is the relationship between accounting earnings and operating cash flow. As a simplified illustration, analysts may examine:

Accrual component ≈ Net income − Cash flow from operations

The result is frequently scaled by average assets or another size measure. Research implementations differ. Some use balance-sheet-based accrual definitions, others cash-flow-based definitions, and still others disaggregate operating, non-operating or financing components.

Those choices matter: measurement and classification differences can change the signal and its interpretation.

Consider two hypothetical companies with identical net income and operating cash flow. Their headline aggregate accruals may match, but the source of those accruals can be economically very different.

One may have increased receivables from a rapidly expanding customer base. Another may have recognised contract assets under long-duration agreements involving significant performance estimates. A third may be affected by restructuring provisions or other accounting judgments.

A comparable factor value does not establish comparable underlying risk.

From accrual magnitude to accrual reliability

Research has shown that the reliability of the accounting components matters, not merely the overall amount of accruals.

Richardson and co-authors (2005) classified accrual components according to the reliability of their measurement and examined their relation to earnings persistence. The implication for investment research is that the composition of accruals can be more informative than an aggregate value alone.

The analysis therefore moves beyond asking whether accruals are high or low.

The institutional question becomes: which reported balances rely most heavily on estimates, and what evidence supports those estimates?

This is particularly relevant when historical factor models combine firms with materially different accounting structures.

Four accounting exposures that can sit beneath a factor score

1. Contract accounting and variable consideration

Companies with long-term contracts may recognise revenue using estimates about performance, variable consideration and contract outcomes. Under ASC 606, these assumptions can affect the timing and measurement of recognised revenue and contract assets.

A traditional accrual measure may capture the resulting balance-sheet changes without explaining which contractual judgments caused them.

A targeted disclosure review can ask whether management revised its estimates, whether contract assets grew faster than revenue, and whether the economic conversion to cash is consistent with the original assumptions.

2. Supplier finance and working-capital presentation

Reverse factoring and supplier finance arrangements can influence payment timing and reported working capital. Their classification depends on the terms and applicable accounting requirements; such programmes are not automatically debt.

An aggregate cash-conversion or accrual metric may not adequately distinguish an operating improvement from a change in financing arrangements.

Analysts should examine the programme disclosures, confirmed obligations, payment terms and liquidity exposure rather than assume that a favourable cash-conversion measure is self-explanatory.

3. Recoverability and impairment estimates

Goodwill, intangible assets, deferred tax assets and other carrying amounts can depend on assumptions concerning future profitability, discount rates or recoverability.

Some of these judgments may have little immediate impact on operating cash flow. Yet changes in assumptions can affect prospective valuation risk or the interpretation of earnings.

The relevant evidence may be in the impairment discussion or critical accounting estimates section, not in the factor calculation.

4. Disclosure and measurement changes

Two filings can present similar headline earnings-quality metrics while describing different estimation uncertainty, business commitments or accounting policies.

Narrative comparisons therefore have value when used carefully. A changed phrase does not prove deterioration, and unchanged text does not prove that economics are unchanged.

The point is to identify material questions worth reconciling with the numbers.

A practical example: identical factors, different exposures

Imagine two hypothetical issuers with the following simplified annual values:

At the level of this simplified metric, the firms appear identical.

Now introduce disclosure context.

Issuer A’s accrual movement reflects trade receivables arising from ordinary revenue growth, with broadly stable collection experience.

Issuer B’s movement reflects a combination of contract asset accumulation and revised estimates of performance under multi-year arrangements.

Neither fact alone establishes poor earnings quality. But the analysis needed for the two issuers is materially different.

Issuer A may warrant a collections and customer-mix review. Issuer B may warrant a deeper investigation into performance obligations, estimation sensitivity, contractual milestones and cash conversion.

The purpose of disclosure-aware research is to distinguish these situations before treating them as economically equivalent observations.

All figures and issuers in this example are hypothetical and are not Helmledgers findings.

The risk of forcing every sector into one accounting template

Accounting structures differ across industries. Contract assets have different implications for an engineering business than receivables have for a retailer. Customer deposits, deferred revenue, insurance reserves and credit-loss allowances also carry sector-specific meanings.

A factor can remain useful across a large universe while becoming less interpretable at the individual-security level.

For investment teams, this suggests separating two tasks:

• Cross-sectional screening: identifying securities or clusters that deviate from a model’s expectations.
• Mechanism-level investigation: determining which disclosed accounting characteristics explain those deviations.

The second task requires accounting knowledge, primary-source evidence and, often, specialised sector treatment.

What a disclosure-aware quantamental workflow looks like

A practical research framework can combine systematic breadth with targeted forensic depth.

Stage 1: Screen. Run established accrual, cash conversion and earnings-quality metrics across the eligible universe. Record data definitions and as-of dates.

Stage 2: Enrich. Identify material note disclosures, contractual arrangements, accounting estimates and changes between filings.

Stage 3: Classify. Link potential anomalies to mechanisms such as recognition timing, balance-sheet recoverability, financing dependence or estimation changes.

Stage 4: Escalate. Prioritise issuer-specific review based on materiality, evidentiary strength, portfolio exposure and whether the finding might alter the investment case.

Stage 5: Test. Where appropriate, assess whether disclosure-derived features add incremental explanatory or predictive information over the existing factor model.

The final step is essential. A narrative that sounds economically compelling is not automatically a validated investment signal.

Avoiding hindsight and information leakage

Quantamental teams should be especially rigorous about point-in-time availability.

A backtest using later disclosures can accidentally assign information to dates on which it was not public. Subsequent restatements, changed filing text and vendor revisions can also distort historical datasets.

A defensible test should preserve filing acceptance or publication dates, document versions, extraction timestamps and the original state of each signal.

It should distinguish information available before a decision from later events used to evaluate that decision.

Additional validation considerations include sector neutralisation, universe selection, transaction costs, multiple testing and out-of-sample performance.

These requirements apply whether the additional feature is numerical or derived from narrative evidence.

What forensic intelligence adds, and what it cannot promise

Structured accounting interpretation can potentially improve research triage by attaching relevant evidence to otherwise opaque factor results.

For example, a researcher might learn not just that an issuer’s cash conversion is changing, but that the change coincides with expanded supplier-finance obligations or a materially different revenue-estimation methodology.

Such findings can generate hypotheses, improve analyst questions and refine downside scenarios.

They do not establish that a trading strategy will outperform, nor do they eliminate the need for financial-model and accounting review.

The right standard is incremental decision usefulness, demonstrated through transparent evidence and appropriate testing.

A complementary intelligence layer for institutional research

Conventional factor research answers an important question: which financial characteristics are associated with patterns across a broad investment universe?

Evidence-driven forensic research asks a different question: what do the underlying disclosures reveal about the quality and economic interpretation of those characteristics?

Helmledgers is developing a source-linked forensic accounting intelligence approach intended to help analysts examine those underlying accounting mechanisms, track changes across reporting periods and focus investigations on material evidence.

The ambition is complementary: preserve quantitative coverage while introducing the context required for deeper fundamental interpretation.

For a quantamental research team, the opportunity is not simply to create another score. It is to bridge systematic signals and explainable, auditable accounting evidence.

Conclusion

Accruals remain an important component of earnings-quality research. Their value should not be dismissed merely because financial reporting contains additional layers of complexity.

But a factor score cannot by itself resolve the contractual, estimation and disclosure questions that influence reported earnings.

For institutional investors, a stronger workflow may combine broad screening with targeted analysis of the accounting mechanisms beneath the numbers.

The signal tells the analyst where to look. The disclosures help explain what the signal means.