Essays on the accounting accruals anomaly
MetadataShow full item record
This dissertation focuses on two issues related to the accounting accrual anomaly documented by Sloan (1996). In the first essay, I study the industry effect of the accrual anomaly. Contrary to the belief demonstrated in prior literature, I find industry membership only has a weak effect in explaining accruals of individual firms. The effect of industry membership is stronger when analysis is conducted on a group of homogenous industries categorized by 4-digit SIC code. Nevertheless, even for these homogenous industries, industry accruals can only explain less than 1/3 of total variation of accruals of individual firms. Although industry accruals seem to be able to explain a small portion of accruals of individual firms, both portfolio- and regression-based results indicate that information embedded in accounting accruals is primarily contained in the industry-neutral portion of accruals, or firm-specific accruals. In the second essay, I study the relationship between the accounting accrual effect and the momentum effect documented by Jegadeesh and Titman (1993). I provide evidence to suggest that the two effects do not subsume each other contemporaneously. However, accruals do seem to affect the subsequent return behaviors of the momentum portfolios in the post-holding period due to investors’ cognitive errors regarding the persistence level of the accrual component in reported earnings. Using the absolute value of accruals as a proxy for the ex-ante probability of future earnings surprises, I show that the ex-ante probability of future earnings surprises is related to the return reversals of the momentum portfolios. When this probability is sufficiently controlled, return reversals of the momentum portfolios become less significant. Although it becomes less significant, prior return momentum is still significantly related to the reversals of momentum returns. Nevertheless, compared to accruals the effect of prior return momentum as a predictor of stock returns in the post-holding period is much weaker. In addition, this paper also provides evidence that suggests besides reflecting the quality of earnings phenomenon as suggested by Sloan (1996), and acting as a proxy for the growth or glamour effect as advocated by Fairfield, Whisenant and Yohn (2002), accounting accruals also serve as proxies for other unknown effects.