The Accruals Anomaly: Why Earnings Quality Predicts Stock Returns

A Gap Between Reported Earnings and Cash

Between 1962 and 1991, NYSE and AMEX firms in the highest decile of balance-sheet accruals earned 4.9% less per year than firms in the lowest decile. The pattern was not subtle: companies whose reported earnings significantly exceeded their operating cash flows went on to deliver strikingly poor stock performance, while companies whose profits were firmly grounded in cash generation outperformed. This observation, documented by Richard Sloan in his 1996 paper "Do Stock Prices Fully Reflect Information in Accruals and Cash Flows About Future Earnings?", exposed a systematic failure of the market to distinguish between the durable and fragile components of corporate earnings.

The accruals anomaly has since become one of the most studied phenomena at the intersection of accounting research and empirical finance. It connects to a broader family of return predictors rooted in earnings quality, and its persistence across three decades of scrutiny makes it a critical lens through which to understand how financial statement information flows into asset prices.

Decomposing Earnings: Cash Versus Accruals

At the heart of Sloan's insight lies a simple accounting identity. Net income can be separated into two components: cash flow from operations and total accruals. Cash flow from operations represents actual money collected and disbursed. Accruals capture the difference, the portion of reported earnings that exists on paper but has not yet materialized as cash. This includes items like changes in accounts receivable, inventory buildups, deferred revenue adjustments, and depreciation allocations.

Sloan's central empirical contribution was to show that these two components differ sharply in their ability to predict future profitability. The cash flow component of current earnings is highly persistent: firms generating strong cash flows today tend to generate strong cash flows next year. The accrual component, by contrast, is far less persistent. A dollar of accrual-based earnings carries substantially less information about future earnings than a dollar of cash-based earnings.

The gap may appear modest in isolation, but when compounded across portfolios, it generates large return differentials. Sloan demonstrated this by sorting stocks into deciles based on the magnitude of their accruals (scaled by total assets). A hedge strategy of buying the lowest-accrual decile and shorting the highest produced abnormal returns of approximately 10.4% per year over his sample period.

Why Markets Misprice Accruals

The central puzzle is not that accruals are less persistent than cash flows, which is a well-understood accounting phenomenon, but that stock prices behave as if investors cannot tell the difference. Sloan's evidence showed that the market prices the accrual and cash flow components of earnings as though they carry identical implications for future profitability. When subsequent earnings reveal that high-accrual firms were overstating their economic performance, prices adjust downward. When low-accrual firms demonstrate that their earnings were genuinely sustainable, prices adjust upward. This delayed correction is what creates the return spread.

Several mechanisms have been proposed to explain this mispricing:

Fixation on the bottom line: Many investors, including sophisticated institutional participants, anchor to reported earnings per share without decomposing the number into its constituent parts. Quarterly earnings calls, analyst estimates, and media coverage focus overwhelmingly on whether a company "beat" or "missed" the consensus EPS figure. The composition of that earnings number, specifically how much represents collected cash versus accounting adjustments, receives far less attention.

Limited attention and processing costs: Even investors aware of accrual quality may find it costly to perform the decomposition systematically across hundreds or thousands of stocks. Dechow and Dichev (2002) showed that accrual estimation errors are widespread and vary significantly in magnitude across firms and industries, making any simple rule of thumb unreliable. This complexity acts as a barrier to arbitrage.

Managerial incentives to inflate accruals: Executives compensated on earnings-based metrics have incentives to use discretionary accruals to inflate reported profits. Revenue recognition timing, reserve adjustments, and capitalization decisions can all shift earnings from future periods into the present. Richardson, Sloan, Soliman, and Tuna (2005) extended Sloan's framework by disaggregating accruals by reliability, showing that the least reliable accrual categories, those most susceptible to managerial manipulation, generate the largest return predictability.

Connecting Accruals to Broader Earnings Quality

The accruals anomaly is one pillar of a broader research program on earnings quality. The quality factor in asset pricing encompasses multiple dimensions, including profitability level, earnings stability, and the degree to which reported profits are backed by cash. The accruals signal specifically targets the cash-versus-paper dimension of quality, making it conceptually distinct from, though correlated with, measures like gross profitability.

Dechow, Ge, and Schrand (2010) provided a comprehensive taxonomy of earnings quality proxies, concluding that accrual-based measures capture information about future performance that is not fully reflected in price-based or profitability-based metrics alone. Their review synthesized over a decade of follow-up research to Sloan's original study and confirmed that the cash-accrual decomposition remained among the most powerful signals in the earnings quality toolkit.

The relationship between accruals and post-earnings announcement drift is also noteworthy. Both anomalies arise from the market's incomplete processing of earnings information, but they operate on different aspects. PEAD reflects underreaction to the level of earnings surprise; the accruals anomaly reflects underreaction to the composition of earnings. Empirically, the two signals have low correlation and generate additive alpha when combined in portfolio construction.

The Replication Record

Like many accounting-based anomalies, the accruals effect has been subjected to extensive out-of-sample testing. The results paint a nuanced picture.

Green, Hand, and Zhang (2017) evaluated accruals alongside 94 other firm characteristics in a comprehensive horse race and found that it retained independent predictive power for returns even after controlling for dozens of competing signals. However, the magnitude of the premium was smaller than Sloan's original estimate, consistent with the general pattern of post-publication attenuation documented across many anomalies.

Allen, Larson, and Sloan (2013) examined the mechanism of accrual reversals directly, confirming that extreme accruals do reverse in future periods and that this reversal is accompanied by stock return corrections. Their work strengthened the causal narrative: the return predictability is not merely a statistical correlation but reflects an identifiable economic process where overstated earnings are corrected by subsequent cash flow realizations.

International evidence has been mixed. The anomaly appears robustly in several developed markets, particularly those with accrual-based accounting regimes similar to US GAAP. In markets where cash-basis accounting is more prevalent or where institutional investors dominate, the effect tends to be weaker. This cross-country variation supports the behavioral explanation: the anomaly is strongest where the gap between reported earnings and cash reality is largest and where retail participation creates more opportunity for fixation on headline numbers.

Measuring Accruals in Practice

Researchers and practitioners employ several approaches to measure accruals, each with distinct trade-offs.

The balance-sheet approach, used in Sloan's original study, calculates total accruals as the change in non-cash current assets minus the change in current liabilities (excluding short-term debt) minus depreciation expense. This method is straightforward and relies only on publicly available financial statements.

The cash-flow-statement approach measures accruals as the difference between net income and cash flow from operations. Since the adoption of SFAS 95 in 1988 required firms to report cash flow statements, this method has become the preferred approach for post-1988 data because it avoids measurement noise introduced by non-operating balance-sheet changes such as acquisitions and divestitures.

Richardson, Sloan, Soliman, and Tuna (2005) showed that disaggregating accruals by type and reliability produces a sharper signal. Working capital accruals tied to receivables and inventory changes carry the most predictive power, while depreciation and amortization (non-discretionary accruals) carry almost none. This finding is consistent with the managerial manipulation channel: the accrual categories most susceptible to timing and estimation discretion are precisely those that predict future underperformance.

Has the Anomaly Survived?

The accruals anomaly's post-publication trajectory follows a pattern familiar from the broader factor investing literature. The raw magnitude of the long-short spread has declined since Sloan's original 1962-1991 sample, particularly among large-cap stocks where institutional attention and quantitative screening have increased. Several studies estimate that the premium has roughly halved since the mid-1990s, consistent with the ~30% post-publication decay that McLean and Pontiff (2016) documented as the average across published anomalies.

However, the signal has not disappeared. In the segment of the market populated by smaller firms with limited sell-side attention and low institutional participation, accrual-based return predictability has not faded to irrelevance. The anomaly also retains power as a component of composite quality scores, where it contributes information that is orthogonal to profitability and leverage measures.

The survival of the accruals signal in smaller stocks aligns with limits-to-arbitrage explanations. High-accrual stocks that are candidates for shorting tend to be precisely the kind of firms that are expensive to borrow, have wide bid-ask spreads, and exhibit high idiosyncratic volatility. These frictions prevent the sophisticated capital that recognizes the mispricing from fully eliminating it.

Where the Evidence Points

Sloan's 1996 paper established a durable empirical regularity: the market treats accrual earnings and cash earnings as interchangeable when they are not. Subsequent research has refined the measurement, mapped the cross-country boundaries, and documented the expected post-publication decay without overturning the core finding. The accruals anomaly occupies a distinctive position in the factor landscape because it derives from a specific, identifiable failure of financial statement analysis rather than from diffuse behavioral biases or risk-based channels.

For investors constructing quality-oriented portfolios, the cash-accrual decomposition provides a complementary screening dimension. Stocks with earnings firmly supported by operating cash flow have historically been more likely to sustain their profitability and less likely to experience the negative earnings revisions that drive price declines. Whether deployed as a standalone signal or integrated into a multi-factor framework, the insight that not all reported earnings are created equal remains one of accounting research's most consequential contributions to investment practice.