Friday, March 31, 4:00 p.m. to 5:40 p.m.
Concurrent session 4C - Cash Flow, Accruals & Earnings Management (Financial Accounting and Reporting)
Title: An Alternative Measure of Discretionary Accruals to Detect Intentional Earnings Management
ABSTRACT: Models that use aggregate accruals in the detection of earnings management have been shown to contain measurement error, which can be significant at times and can lead to biased results. Current models can be improved upon by examining how different components of aggregate accruals are managed. In this paper, it is shown that managers who wish to manipulate earnings will do so using one or more components of accruals in a way that would affect earnings in the same direction. Given this expectation, a measure is devised that captures the signs and magnitudes of the components of discretionary accruals. The measure is used in conjunction with the familiar discretionary accruals to provide a less noisy measure of earnings management behavior, a corrected measure of discretionary accruals.
Following the methodology in Dechow et al. (1995), the corrected measure of discretionary accruals is tested and is found to improve the power of the discretionary accruals models in the detection of earnings management. This is tested in samples that have artificially added manipulation of accruals, as well as in a sample of firms that were targeted by the Securities and Exchange Commission (SEC) for alleged accounting fraud. The corrected measure of discretionary accruals shows only slight improvement with respect to the specificity of detection. This is tested in random samples with no expectation of earnings management, as well as in random samples chosen from extreme earnings and cash from operations sub-samples.