Journal of Business Finance & Accounting, 37(3) & (4), 356–368, April/May 2010, 0306-686X doi: 10.1111/j.1468-5957.2010.02205.x Discussion of How Do Restatements Begin? Evidence of Earnings Management Preceding Restated Financial Reports JAN BARTON ∗ Ettredge, Scholz, Smith and Sun (2010) examine the extent to which ‘within-GAAP’ earnings management precedes periods of earnings reported ‘without-GAAP.’ Specif- ically, they model the probability of a firm restating its financial statements (because of either an error or an irregularity) as a function of the magnitude of the firm’s working capital immediately prior to the restated period and a set of controls. They find a consistent and significant positive association between working capital and the likelihood a firm restates irregularities, a weaker and less consistent association with the likelihood a firm restates errors in core operating accounts, and essentially no association with the likelihood of restating errors in noncore accounts. Ettredge et al. interpret these finding as suggesting that managers intentionally overstate earnings without violating GAAP and, once they exhaust flexibility within GAAP, they manage earnings upward by violating GAAP. They conclude, ‘a pattern of aggressive accounting can be detected several years prior to clearly identified frauds’ (p. 4). Despite decades of ongoing research on the causes and consequences of earnings management (see reviews by Fields et al., 2001; Dechow and Schrand, 2004; Ronen and Yaari, 2008; and Dechow et al., 2010), we know surprisingly little about how and why managers begin to manipulate earnings, and how and why they cross the line between acceptable and unacceptable earnings management practices. I would like to highlight in my discussion some aspects of Ettredge et al.’s study that I believe will help assess their contribution to knowledge in this area. 1. WHAT IS THE QUESTION, REALLY? The research question defines the study, sets boundaries, and provides direction on every aspect of the study — from the literature review and hypothesis development, the empirical design and analysis, to ultimately the contributions of the study. Titling ∗ The author is Associate Professor of Accounting at Emory University’s Goizueta Business School. He thanks Peter Pope (editor) for the invitation to write this discussion; participants (especially Scott Richardson) at the JBFA 2009 Capital Markets Conference held in Venice, Italy, for their comments on the original draft of the discussed paper; and that paper’s authors for feedback on this discussion. Address for correspondence: Jan Barton, Emory University, 1300 Clifton Road, Atlanta, GA 30322, USA. e-mail: jbarton@emory.edu C  2010 Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA. 356 Journal of Business Finance & Accounting