Purpose For most companies, growth measures such as asset growth are positively correlated with accrual measures. Just like investment in fixed assets, current accrual represents one form of investment and is an integral part of a firm’s business growth. This makes it difficult to distinguish between the growth-based and earnings quality-based interpretations of the accrual effects, because high accruals can represent both high growth and inflated earnings. The purpose of this paper is to add to the literature by examining an issue that has not received much attention: the correlation between asset growth and accruals and its implication on stock return predictability. The authors address the issue using Fama and Macbeth’s (1973) cross-sectional regressions that are conditional on the correlations between the two variables. Design/methodology/approach The authors partition firms based on whether the correlation between current accrual and asset growth in the past five years is positive (ρ+) or negative (ρ−). The authors refer to these two types of firms such as “positive correlation” and “negative correlation” groups. For both groups, the authors examine whether firms with higher asset growth and higher accruals are associated with lower future stock returns. The authors implement Fama and MacBeth’s cross-sectional regressions incorporating the effect of correlations between growth and accrual measures. In addition, the authors regress hedge portfolio returns on Fama and French (1993) three-factor and Fama and French (2015) five-factor models to see if the intercepts (a’s) from these regressions are significantly different from 0. Findings For each year, the authors partition firms based on whether the correlation between asset growth and current accrual is positive or negative. For both the “positive correlation” and “negative correlation” firms, the authors examine the association between accruals and future stock returns. The authors find that accruals remain strong in predicting future stock returns for both groups. The accrual effects from the “negative correlation” group cannot be attributed to the growth-based hypothesis because for these firms, when accruals are high, growth measures tend to be relatively low and vice versa. The effects are most likely driven by the alternative hypothesis that investors overvalue the accrual part of earnings. Research limitations/implications There exist a few issues when investors actually implement these strategies. These include liquidity costs, institutional holdings and short sale constraints. Lesmond (2008) concludes that the bulk of the trading profits is derived from the short side of the trade, but that this position suffers from high liquidity costs that reduces institutional holdings with consequent short sale constraints. The net gains after taking into account these issues remain an open question be addressed in the future. Practical implications The empirical results indicate that investors can do an implementable portfolio strategy of going long for a year on an initially equally weighted lowest asset growth (accrual) decile portfolio and going short for a year on an initially equally weighted highest asset growth (accrual) decile portfolio, which produces significant abnormal returns. The results further show that these abnormal returns can be improved if investors classify stocks into “the positive correlation” and “negative correlation” groups and implement trading similar trading strategies. Originality/value The empirical evidence finds that firm-year observations that exhibit a negative correlation between growth and accrual measures represents a significant 30 percent of the total firm-year observations during the sample period from July 1974 to June 2017. This highlights the necessity to undertake a detailed analysis on the issue. The authors continue to find accrual effects among these groups of firms. Therefore, the accrual effect cannot be attributed to the diminishing marginal return hypothesis. This is the main contribution of the paper.

中文翻译：

---

增长和应计指标负相关时的股票收益可预测性

目的对于大多数公司而言，诸如资产增长之类的增长指标与权责发生制正相关。就像固定资产投资一样，当期应计项目是一种投资形式，并且是公司业务增长不可或缺的一部分。这使得很难在权责发生制的基于增长的解释和基于收益质量的解释之间进行区分，因为高应计制可以代表高增长和虚假收入。本文的目的是通过研究一个尚未引起广泛关注的问题来增加文献资料：资产增长与应计之间的相关性及其对股票收益可预测性的影响。作者使用Fama和Macbeth（1973）的横截面回归来解决这个问题，该回归以两个变量之间的相关性为条件。设计/方法/方法作者根据过去五年当前应计收入与资产增长之间的相关是正（ρ+）还是负（ρ-）来划分公司。作者提到这两种类型的公司，例如“正相关”和“负相关”组。对于这两组，作者研究了资产增长较高和应计利润较高的公司是否与较低的未来股票收益率相关。作者实现了Fama和MacBeth的横截面回归，其中纳入了增长和权责发生制之间的相关性。此外，作者使用Fama和French（1993）三因子模型以及Fama and French（2015）五因子模型对套期保值投资组合进行回归，以查看这些回归的截距（a's）是否与0显着不同。年，作者根据资产增长与当期应计额之间的相关性是正还是负来对公司进行划分。对于“正相关”和“负相关”的公司，作者研究了应计利润和未来股票收益之间的关系。作者发现，应计利润在预测两组未来的股票收益方面仍然很强劲。“负相关”组的应计结果不能归因于基于增长的假设，因为对于这些公司，当应计额较高时，增长指标往往相对较低，反之亦然。这种影响很可能是由另一种假设所驱动的，即投资者高估了收益的应计部分。研究局限/含义投资者实际实施这些策略时，存在一些问题。其中包括流动资金成本，机构持股和卖空限制。莱斯蒙德（Lesmond，2008年）得出的结论是，大部分交易利润都来自交易的空头头寸，但是该头寸的流动性成本高昂，从而降低了机构持仓量，从而限制了卖空行为。考虑到这些问题后的净收益仍然是一个未解决的问题，将来需要解决。实际意义经验结果表明，投资者可以制定可行的投资组合策略，即对最初平均加权最低资产增长（应计）的十分位数组合投资多头一年，而对最初加权平均资产最高增长（应计）的空头组合一年。十分位数的投资组合，产生明显的异常收益。结果进一步表明，如果投资者将股票分为“正相关”和“负相关”两类并实施类似的交易策略，则可以改善这些异常收益。独创性/价值经验证据发现，在1974年7月至2017年6月的样本期内，增长与应计指标之间呈负相关的公司年观测值占公司年总观测值的30％。对此问题进行详细分析。作者继续在这些公司集团中发现权责发生制。因此，应计效应不能归因于边际收益假说的减少。这是本文的主要贡献。