Trends in accrual quality and real activity-based earnings management in the pre and post Sarbanes-Oxley eras

An increase in the prevalence of earnings restatements and cases of financial statement fraud in the early 21st century led to a significant loss of market capitalization and investor confidence in the attestation process. In an effort to restore such confidence, Congress passed the Sarbanes-Oxley Act (SOX) in July of 2002. The Act significantly increased the penalties for engaging in accrual activities aimed at either misleading users of the financial statements concerning the underlying economic condition of the firm or influencing contractual outcomes. Recent literature separates earnings management into accrual and real activities. Accrual activities include the management of accounts that have not yet been realized in cash, such as receivables and payables. Real activities include the management of actions that deviate from normal business practices, such as price discounts aimed at temporarily increasing sales, excessive inventory production aimed at lowering the cost of goods sold, and aggressively reducing discretionary expenditures such as R&D to improve profit margins. As a result of the increased penalties for engaging in accrual activities, one would expect a relative shift from accrual activities to real activities to facilitate earnings management in the post-SOX period. As with most academic social disciplines, the test employed in my dissertation is a joint test of the sensitivity of the tools available to detect management activities, the research design, and the presence and strength of the effect for which I am searching. This dissertation is the first to test for changes in both accrual quality and real activity-based earnings management in the post-SOX period. In order to test for a change in accrual quality in the post-SOX period, I utilize a model developed by Dechow and Dichev in 2002. The Dechow and Dichev (2002) model of accrual quality is an appropriate measure of accrual information risk, and may therefore be superior to the use of discretionary accrual models to test for an economic effect (Francis et al. 2004). I also utilize three empirical measures of real activity-based earnings management developed by Roychowdhury (2006) to document a change in real earnings management in the post-SOX period. The findings of the study empirically support a change in earnings management techniques in the post-SOX period compared to the pre-SOX period. Specifically, the quality of accruals incorporated into the accounting earnings figure have significantly increased in the post-SOX period. However, instances of earnings management using real activities have also significantly increased in the post-Sox period. These findings inform academics about the power of the tools used in academic accounting research and the overall quality of the argument. They inform users of financial statements about where to direct their attention in reading and evaluating the financials. Finally, they inform regulators, practitioners and policy makers of the effectiveness of the law at improving the quality of accruals, and bring to their attention a potential substitution in the techniques used to manage earnings