Research Hub

The Academic Evidence Behind the Accruals Anomaly

You will find structured summaries of the peer-reviewed literature that established, tested, and debated the relationship between accounting accruals and cross-sectional return variation across decades of empirical finance research.

30+
Years of research
80+
Studies reviewed
40+
Markets examined

Built on Peer-Reviewed Academic Literature

Rigorous foundations, honest evidence

Every section of this research hub traces directly to published, peer-reviewed work in the finance and accounting literature. You will not find unsupported assertions about what accruals predict. You will find structured explanations of what specific studies tested, how they tested it, what their data showed, and where subsequent researchers challenged or extended their findings. The material is designed for practitioners and researchers who want to engage with the evidence on its own terms rather than receive a simplified digest of what the anomaly is supposed to mean.

These studies form the core of the accruals anomaly literature and remain the essential reference points for any serious engagement with the evidence.
Richard Sloan's 1996 paper in The Accounting Review established the foundational empirical case for the accruals anomaly. Sloan demonstrated that firms with high accounting accruals — where the non-cash component of earnings is large relative to total assets — subsequently underperformed firms with low accruals in cross-sectional return tests. His key insight was that investors appeared to price the persistence of earnings without adequately discounting the lower persistence of the accrual component relative to the cash flow component. The paper introduced the balance sheet decomposition method that became the standard approach for measuring operating accruals and remains widely used in replication studies to this day.

Hirshleifer, Hou, Teoh, and Zhang extended the foundational work by examining whether the accruals anomaly is best understood through the lens of investor inattention and limited processing capacity. Their research showed that the anomaly was concentrated among firms that received less analyst coverage and institutional attention, consistent with the hypothesis that the mispricing reflects a systematic failure to process publicly available accounting information rather than compensation for a missing risk factor. This behavioural interpretation placed the accruals anomaly within a broader literature on how cognitive limitations affect the incorporation of public information into market prices.

Dechow, Richardson, and Sloan contributed a detailed decomposition of the accruals construct itself, separating working capital accruals from long-term non-cash items and examining which components carry the most return-predictive information. Their work refined the measurement approach and showed that the return predictability is not uniform across all accrual types — some components carry stronger signals than others, and the composition of a firm's accruals matters as much as their aggregate magnitude. This granularity became important for subsequent researchers testing whether the anomaly survives more precise accounting decompositions.

International replication studies tested whether the accruals anomaly documented in US data generalises to equity markets operating under different accounting standards, institutional environments, and levels of market development. Researchers examined the anomaly across European, Asian, and emerging market contexts, finding mixed but broadly supportive evidence. The signal appeared in many markets but with variation in magnitude and persistence that researchers attributed to differences in accounting quality, analyst coverage density, and trading cost structures. These international tests are important because they help distinguish between explanations that depend on US-specific institutional features and those that reflect more general properties of accrual-based earnings measurement.
The debate about whether the accruals anomaly reflects genuine mispricing or compensation for a missing risk factor remains one of the most contested questions in the empirical asset pricing literature. Researchers on the risk-based side argue that high-accrual firms share characteristics associated with elevated fundamental risk — rapid growth, capital intensity, and uncertain future cash flows — and that the return differential reflects rational pricing of those risks rather than systematic error. Researchers on the mispricing side point to the concentration of the effect among less-followed firms, the anomaly's apparent predictability from public information, and the pattern of subsequent earnings disappointments as evidence that the signal reflects an exploitable accounting-based bias in how markets weight earnings components.

More recent work has examined whether the accruals anomaly has attenuated since its publication and widespread dissemination in the academic and practitioner communities. Some studies find that the return spread between low-accrual and high-accrual firms has narrowed in the post-publication period, consistent with the hypothesis that arbitrage activity and increased awareness have partially corrected the mispricing. Others argue that the anomaly persists after accounting for transaction costs and that the apparent attenuation reflects changes in the composition of listed firms rather than genuine elimination of the signal. This debate about anomaly decay is relevant for understanding both the current state of the literature and the appropriate interpretation of historical return evidence. Past performance documented in any period does not guarantee future results.

Practical Research Tips for Analysts Studying the Accruals Anomaly

How to compute accruals from financial statements correctly, which data sources researchers rely on, common backtesting pitfalls to avoid, and how to protect your analysis from look-ahead bias in accrual-based tests.

Compute Accruals Using the Balance Sheet Method Correctly

The original Sloan decomposition computes operating accruals as the change in non-cash working capital minus depreciation, scaled by average total assets. A common error is using net income minus operating cash flow from the cash flow statement directly without adjusting for the differences in scope between the two measures. The balance sheet method and the cash flow statement method produce different accrual estimates, and the choice matters for replication accuracy. Know which method the study you are replicating used before comparing your results.

Use Point-in-Time Data to Avoid Look-Ahead Bias

One of the most consequential errors in accruals research is using financial data that would not have been available to market participants at the time the return measurement begins. Annual financial statements are typically filed weeks or months after the fiscal year end. Any backtest that uses fiscal year-end accrual data without applying a filing date lag will inadvertently incorporate information that was not yet public, systematically overstating the anomaly's historical magnitude. Always align your accrual measurement date to the actual public availability of the financial data, not the fiscal year-end date.

Understand Which Databases Researchers Use and Why

The accruals anomaly literature relies primarily on Compustat for accounting data and CRSP for return data in the US context. The specific variable definitions in Compustat have changed over time, and some items used in early studies are no longer populated in the same way for more recent data. When replicating a study, check which data vintage and variable definitions the original authors used. International replications typically use Worldscope or Datastream, which have their own coverage gaps and standardisation differences that affect cross-country comparisons.

Avoid Survivorship Bias in Your Sample Construction

Research databases accessed today contain only firms that survived long enough to be included in the data vendor's historical archive. If you construct your accrual sample by pulling all firms currently available and then looking back at their historical financials, you will exclude firms that were delisted, went bankrupt, or were acquired — precisely the population of high-accrual firms most likely to have produced poor subsequent returns. Use the full historical universe available at each point in time, including delisted firms, to avoid overstating the anomaly's return differential in your own analysis.

Control for Known Risk Factors When Isolating the Accrual Signal

Raw return spreads between low-accrual and high-accrual portfolios confound the accrual effect with exposure to market, size, value, and momentum factors that are independently associated with cross-sectional return variation. Academic tests of the anomaly typically include Fama-French factor controls or characteristic-matched portfolio adjustments to isolate the component of the return differential attributable specifically to accrual intensity. If you are evaluating the anomaly's strength in a particular sample, risk-adjusting your return estimates is essential for drawing valid inferences about whether accruals carry independent return-predictive information.

Track How the Anomaly's Magnitude Has Changed Over Time

The accruals anomaly was documented in data from the 1960s through the early 1990s and published in 1996. Subsequent research has examined whether the anomaly persists in post-publication data, with mixed findings across different sample periods and geographies. When you test the anomaly yourself, splitting your sample into pre-publication and post-publication sub-periods is informative. A substantial reduction in the return spread after 1996 is consistent with the anomaly decay hypothesis, though researchers disagree about how much of any apparent reduction reflects genuine arbitrage and how much reflects structural changes in the listed firm population.

Latest Commentary on the Accruals Anomaly Literature

Recent updates covering new academic publications, evolving debates about anomaly persistence in algorithmic markets, post-IFRS accounting standard effects on accrual measurement, and practitioner implications of the latest empirical findings.

algorithmic trading screens showing financial data analysis

Does Algorithmic Trading Accelerate Accrual Mispricing Correction

New evidence from high-frequency data suggests that algorithmic traders respond faster to accrual signals than discretionary investors, raising questions about whether the anomaly continues to reflect delayed price adjustment or something more structural in how markets process accounting information.

Market Microstructure
international financial reports and accounting standards documentation

IFRS Adoption and Accrual Measurement Across International Markets

Research examining equity markets that transitioned to IFRS finds that standardised accounting rules affect the composition and magnitude of measured accruals, with implications for whether anomaly tests conducted before and after adoption are directly comparable across the same market.

Accounting Standards
emerging market financial data and equity analysis charts

Accruals Anomaly Persistence in Emerging Market Equities

A new cross-country study covering emerging market equities finds that the return spread between low-accrual and high-accrual firms remains statistically significant in markets with lower institutional ownership and analyst coverage density, consistent with the inattention hypothesis.

International Evidence
financial analyst reviewing forecast models and accounting data

Accrual Quality and Analyst Forecast Errors Revisited

Updated evidence on the relationship between accrual intensity and subsequent analyst forecast errors finds that the pattern documented in earlier studies persists in recent data, suggesting that professional analysts continue to underweight the cash component of earnings in their forward estimates.

Earnings Quality
machine learning models applied to financial data research

Machine Learning Methods Applied to Accrual Signal Extraction

Researchers applying machine learning techniques to financial statement data find that non-linear combinations of accrual components outperform the simple aggregate measure in predicting cross-sectional return variation, though questions about overfitting and economic interpretation remain open in the literature.

Quantitative Methods