The Accruals Framework – History and Overview
The Accruals Anomaly – History and Overview
The Accruals anomaly, as it has come to be known, was made famous by Richard Sloan in his landmark 1996 research paper “Do Stock Prices Fully Reflect Information in Accruals and Cash Flows About Future Earnings?”¹. In this paper, Sloan hypothesised that because investors tend to “fixate” on reported earnings instead of distinguishing the cash flow and accrual components of earnings, this represented an investment opportunity. His proof involved two steps.
Firstly, he measured the persistence of earnings – how much of next years earnings can be predicted by this years earnings. For earnings in general this figure was 84c for each dollar. He then repeated this measurement, but separated the earnings into two components, a cash flow component and an accruals component. His results were clear:
- For the cash flow component, around 86c in every dollar of cash persisted into next year; and
- For the accruals component, around 77c in every dollar persisted into next year.
The persistence of earnings in this table are measured by the Coefficients – 0.841 (representing 84c in the dollar for earnings alone) and 0.765 and 0.865 when decomposed into Accrual and Cash Earnings respectively.
Source: Dechow and Schrand, Earnings Quality, Research Foundation of CFA Institute
Viewed another way, Sloan plotted the persistence of accruals and cash flows over time, splitting each into high and low baskets. The lower persistence of accrual earnings was highlighted by the fact that they mean reverted much more quickly than cash flow earnings. In short, cash earnings were more statistically significant than accrual earnings.
His second test was to see whether this lower persistence of accrual based earnings was understood and accurately priced by investors. Again the results were clear. Investors did not accurately price accruals, suggesting opportunities for significant gain.
Sloane ranked companies into deciles based on their accruals. He then constructed theoretical portfolios that went long the top decile (i.e. lowest accruals) and short the bottom decile (highest accruals). The annual average abnormal returns from this strategy was roughly 10% p.a. with only two negative years. Note that these portfolios were priced 4 months after fiscal year end, so do not capture price movements on the release of the annual results.
Considering the basic measure that Sloan used², this was a stunning result and prompted a wave of subsequent research which examined this finding in more detail. Some of the points made were:
1. The strategy worked for both growth and value stocks
Low accruals outperformed high accruals by ~9% pa. over 26 years, with similar results for both value and growth portfolios (based on price/book).
Mahedy, Bernstein Investment Research
2. Value driven by avoiding high accruals
Houge & Loughran, 2000, confirmed the result was independent of style and size using the Fama three factor model, whilst in addition showing that much of the value came from avoiding the very poor accrual stocks.
3. Measuring accruals relative to earnings rather than the balance sheet improved results.
Hafzalla, Lundholm &Winkle, 2010, compared the results when accruals were scaled as a % of earnings than as a % change in the balance sheet. At all levels scaling relative to earnings gave better results, but particularly where cash flow accruals were low. i.e. value add from owning stocks where cash flow >= earnings.
However, over time it appeas that the ability to profit from a simple measure of accruals has diminished.
Green, Hand and Soliman 2009 tracked performance of accruals portfolio’s over time and found the effect has diminished. This seems to reflect:
- A decline in the size of the signal – ie. less extreme accruals and differences between the persistence of accruals and earnings; and
- An increase in money chasing the strategy. (I.e. the market has adapted).
- The discovery of the accruals anomaly was a major finding.
- Because the market is adaptive (both investors and company accountants), the strength of the anomaly has been arbitraged away at a macro level;
- This does not mean accruals are insignificant. Although in aggregate the accruals signal is diminishing, it will still be important in individual cases.
Therefore, we need to understand in detail which accruals are important. To do this we need to consider:
Having done this, we then need to understand:
- How to measure accruals; and
- When the accrual results are important to our investment framework.