The Accruals Anomaly
As a tradeable strategy, not just an accounting curiosity: short the highest-accrual decile of stocks, go long the lowest, and the spread has historically paid off — though implementing it profitably runs into short-selling frictions that the academic long-short spread ignores.
Prerequisites: The Sloan Accruals Anomaly
The Sloan Accruals Anomaly describes why high-accrual earnings are lower quality: they're built from receivables and inventory rather than cash, and that lower quality predicts weaker future earnings. As a trading strategy, that finding turns into a simple rule: rank stocks each quarter by their accrual ratio, buy the decile with the lowest accruals, short the decile with the highest, and hold until the next ranking. The rule is easy to state; running it profitably is where the real work is.
From accounting fact to portfolio
The accrual ratio used for ranking is typically balance-sheet accruals scaled by average total assets:
In words: how much of the change in a company's working-capital accounts (receivables, inventory, payables) shows up as reported earnings once depreciation is backed out, relative to the size of the company. A high ratio means a large share of this period's earnings came from accruals rather than cash.
Worked example. A universe of 1,000 mid- and large-cap stocks is ranked quarterly into deciles by accrual ratio. Historically the bottom decile (lowest accruals, most cash-backed earnings) has outperformed the top decile (highest accruals) by an average of roughly 8–10% annualized in the original Sloan sample, before costs. A fund running $500m against this signal, long $250m in the bottom decile and short $250m in the top decile, would in a representative year capture something like $20–25m of gross spread — but the short leg is disproportionately full of small- and mid-cap names with high accruals (often growth companies aggressively booking revenue), which tend to have thin float and elevated borrow fees. If those shorts average 150 basis points of annual borrow cost on $250m, that's $3.75m taken directly off the gross spread before considering market-impact costs from quarterly rebalancing 1,000 names.
The accruals signal itself is straightforward to compute; the strategy's implementation cost is concentrated almost entirely in the short leg, because high-accrual names are disproportionately the small, hard-to-borrow ones.
What erodes it
Three things compress this edge relative to the textbook spread: borrow costs on the concentrated short leg, as above; crowding, since accruals is now a standard input in quantitative "quality" and "earnings quality" factor models run at scale, meaning the cheapest, easiest-to-implement version of the signal is already arbitraged into many multi-factor books; and data lag — accrual ratios require a full balance sheet, so the signal updates only quarterly with a reporting lag, giving competing funds time to trade on the same filing before a slower fund can rebalance.
Don't confuse the accruals anomaly with the separate, faster-moving post-earnings-announcement drift (PEAD) — accruals is a slow, quarterly balance-sheet signal about earnings quality, while PEAD trades the market's underreaction to an earnings surprise over weeks. They sometimes get bundled into the same "earnings-based" bucket but decay from different mechanisms.
Related concepts
Practice in interviews
Further reading
- Sloan, Do Stock Prices Fully Reflect Information in Accruals and Cash Flows About Future Earnings? (1996)