The accruals anomaly is the documented tendency for firms with high accounting accruals (earnings far in excess of the cash they generate) to subsequently underperform the market, while firms with low or negative accruals (cash-rich earnings) outperform. First documented by Richard Sloan in 1996, it rests on a simple decomposition: reported earnings = cash flow + accruals. The cash component is more persistent — it repeats next year — whereas the accrual component (driven by accounting estimates, revenue timing, and inventory/receivable build-ups) reverses. The core tension is behavioral: investors appear to "fixate" on the bottom-line earnings number and fail to discount the lower-quality accrual portion, so high-accrual firms are systematically over-priced and low-accrual firms under-priced until the mispricing corrects.

How it's formed and measured

Accruals are the non-cash adjustments that bridge cash flow and accrual-basis earnings. Two standard measurement approaches exist:

Balance-sheet (Sloan 1996) method — the original definition, computed as the change in non-cash working capital minus depreciation:

Accruals = (ΔCurrent Assets − ΔCash) − (ΔCurrent Liabilities − ΔShort-term Debt − ΔTaxes Payable) − Depreciation & Amortization

This is then scaled by average total assets to produce a comparable ratio. Rising receivables and inventories inflate accruals; cash collections lower them.

Cash-flow-statement method — accruals ≈ net income minus cash flow from operations. Hribar and Collins (2002) showed this is cleaner because the balance-sheet method is contaminated by mergers, acquisitions, and divestitures that change working-capital balances without affecting earnings quality.

Richardson, Sloan, Soliman & Tuna (2005) extended the definition to a broad measure of total accruals covering all non-cash, non-equity changes (including non-current operating and financial assets), and ranked accrual categories by reliability — less reliable accruals (those involving more subjective estimation) predicted both lower earnings persistence and larger return reversals.

The trading construction is a classic cross-sectional hedge: rank the universe by scaled accruals each year, go long the lowest decile and short the highest decile, and rebalance annually after the fiscal-year filing window (Quantpedia's reference implementation rebalances in May).

How it's used in practice

In its pure form the anomaly is a quant equity long-short factor, deployed by quantitative funds as one of a family of "earnings-quality" or "accruals quality" signals alongside asset growth and net operating assets. In fundamental and discretionary work the same logic is used as a red-flag screen: a company whose net income consistently exceeds operating cash flow — a large positive accrual — is flagged as having lower-quality, potentially unsustainable earnings, often because of aggressive revenue recognition or channel stuffing. Many earnings-manipulation detectors (e.g. the Beneish M-score's accruals component) and "Sloan ratio" screens used by retail analysts derive directly from this work. The practical takeaway most analysts retain is durable even if the standalone trading edge is not: prefer cash-backed earnings; distrust earnings the cash flow statement does not corroborate.

Adoption, debate & evidence

The anomaly is one of the most heavily studied results in accounting, with hundreds of follow-up papers. Sloan (1996) originally reported a hedge return of roughly 10–12% annually on the long-short portfolio over 1962–1991. Quantpedia's indicative 1966–2003 backtest cites about 7.5% annual return, Sharpe ≈ 0.34, volatility ≈ 10%, with a ~36% maximum drawdown — useful but far from a free lunch.

The central debate is whether the edge still exists. Green, Hand & Soliman (2011, Management Science, "Going, Going, Gone? The Apparent Demise of the Accruals Anomaly") extended the sample past 2003 and found hedge returns had decayed to roughly zero, and turned negative out-of-sample. Their attributed cause is self-arbitrage: as the result became famous, capital flooded into accrual-based strategies — the paper measures this via rising hedge-fund assets under management and trading volume in extreme-accrual decile firms — and competed the premium away (they also note a secondary decline in the size of the mispricing signal itself). It is widely observed that accounting academics moved into quant asset management over this period (e.g. at Barclays Global Investors), though that specific channel is commentary rather than a measured result in the paper. Quantpedia accordingly grades the live signal "Weak."

A competing camp argues the premium never fully arbitraged away because it couldn't be. Mashruwala, Rajgopal & Shevlin (2006) showed the anomaly concentrates in stocks with high idiosyncratic volatility (no close substitute to hedge against) and in low-price, low-volume names where transaction costs are punishing — classic limits-to-arbitrage. A risk-based school (e.g. Wu, Zhang & Zhang on the "optimal investment hypothesis") argues accruals proxy for investment and discount-rate variation, meaning part of the return may be compensation for risk, not mispricing. There is no full consensus, and reported post-2003 results vary with definition (cash-flow vs balance-sheet), universe, and whether micro-caps are included.

Strengths & limitations

When it works: the economic intuition is robust and survives the trading debate — accrual reversal is a genuine, mechanical accounting property, and earnings-quality screening remains valuable for fundamental due diligence. The anomaly historically delivered relatively low market correlation, making it attractive as a diversifying quant factor.

When it fails: as a standalone tradable edge, the evidence that it has substantially decayed in U.S. large/mid caps since the mid-2000s is strong. What survives is concentrated in exactly the names that are hardest and costliest to trade (illiquid micro-caps, high idiosyncratic risk), so realized net-of-cost returns can be much lower than gross backtests imply, with deep drawdowns.

The #1 misuse: treating raw accruals as a high-conviction standalone signal on liquid stocks today, or assuming a high-accrual firm is committing fraud. High accruals are common and often benign — fast-growing firms legitimately build working capital. The signal is a probabilistic quality tilt and a flag for further investigation, not a verdict.

Sources

• Sloan, R. (1996), "Do Stock Prices Fully Reflect Information in Accruals and Cash Flows About Future Earnings?", The Accounting Review — original anomaly, ~10–12% hedge return.
• Richardson, Sloan, Soliman & Tuna (2005), "Accrual Reliability, Earnings Persistence and Stock Prices," J. of Accounting & Economics 39:437–485 — broad accrual definition + reliability. SSRN
• Green, Hand & Soliman (2011), "Going, Going, Gone? The Apparent Demise of the Accruals Anomaly," Management Science. SSRN
• Mashruwala, Rajgopal & Shevlin (2006), "Why Is the Accrual Anomaly Not Arbitraged Away?", J. of Accounting & Economics 42:3–33 — idiosyncratic risk / transaction costs. ScienceDirect
• Hribar & Collins (2002) — cash-flow vs balance-sheet accrual measurement.
• Quantpedia — Accrual Anomaly — strategy construction, backtest stats, "Weak" current grade.

Disputed: whether the residual return is mispricing vs risk-compensation; the degree of post-2003 decay (varies by definition and universe).