The Accruals Anomaly and Company Size

Research has shown that a trading strategy based on publicly available accounting accrual information can earn abnormal returns of approximately 10 percent in the year after it is applied. This article reports a study of whether this “accruals anomaly” is sensitive to company size. The empirical results suggest that the interaction between company size and accruals provides incremental information about future returns and that the accruals anomaly is not independent of company size. The negative abnormal returns when an accruals-anomaly strategy is applied come primarily from the larger companies, and the positive abnormal returns come from the smaller companies.In the study reported, we investigated the relationship between “accruals-anomaly” abnormal returns, which have been found by previous research, and company size. The accruals anomaly is the existence of a negative relationship between current accruals and future abnormal returns. If evidence were to show the anomaly to be limited to small companies, for which stock transaction costs are greater than for large companies, this evidence would provide an explanation as to why the anomaly has not been arbitraged away.Research has shown that a portfolio strategy of buying companies belonging to the bottom accrual decile (that is, companies with the highest accruals) and selling short companies in the top accrual decile (companies with the lowest accruals) generated significantly positive abnormal returns during the 1962–91 period. Results based on recent data also indicate that the accruals anomaly exists. The existence of the accruals anomaly for such an extended period is puzzling. One explanation is that transaction costs restrain investors from exploiting the accruals anomaly. To investigate this possibility, we examine how, in an accrual-based trading strategy, abnormal returns to long and short positions vary in relation to company size.The sample in this study includes all companies traded in the NYSE, Amex, and NASDAQ exchanges for which data are available in the CRSP and Compustat files for the fiscal years 1971 through 2003. Closed-end funds, investment trusts, foreign companies, and financial companies were excluded, and we eliminated shares priced below $5 at the beginning of the holding period to ensure that the results would not be driven primarily by very small, illiquid stocks or by the bid–ask bounce.To explore the relationship between the accruals anomaly and company size, we sorted all companies in our sample into accrual deciles and then sorted the companies in each decile into size quartiles. This procedure yielded four portfolios per accrual decile that had approximately the same level of accruals but varying market values. Through the double sorting, we constructed 40 portfolios in which we isolated the effect of size on returns and kept accruals stable.To measure abnormal portfolio returns, we used several risk-adjustment methods, including estimating Jensen’s alphas based on three-factor, four-factor, and liquidity-augmented models and computing size-adjusted buy-and-hold abnormal returns by using size benchmark returns.Our key results indicate that the positive abnormal returns previously documented come primarily from small companies in the bottom accrual decile and the negative abnormal returns come from large companies in the top accrual decile. We found results that were similar for all the methods we used to estimate abnormal returns. In addition, through regression analysis, we analyzed the relationship between company size and size-adjusted returns and obtained consistent results.In contrast to findings in which accruals-anomaly returns were confined to small and illiquid companies, our results for part of the sample indicate that abnormal returns increase as company size (and liquidity) increases. These results are not fully consistent with the view that transaction costs restrain investors from exploiting the accruals anomaly and point to the need to look for other factors that explain why the accruals anomaly might not be arbitraged away.Furthermore, the documented asymmetrical relationship between company size and abnormal returns calls for an accrual-based trading strategy that is slightly different from the standard strategy. The standard accrual-based strategy is to go long all bottom-accrual-decile companies and short all top-accrual-decile companies. In the adjusted strategy, the investor is to purchase companies in the smallest size quartile of the bottom accrual decile and sell short companies in the largest size quartile of the top accrual decile. Such a modified strategy, because of the interaction of the size and accrual terms, generated higher abnormal returns than the standard accrual-based strategy in our tests. In addition, because it is less costly to sell short large-company stocks than to sell short small-company stocks, the modified accrual trading strategy is likely to incur lower transaction costs and, therefore, generate higher abnormal returns net of transaction costs.In summary, our empirical results suggest that accruals-anomaly returns are correlated with company size and that returns to an accrual-based strategy can be increased by assigning long positions to small companies and short positions to large companies.