4분기 이익조정과 이익지속성

This study examines the relation between earnings persistence and earnings management through fourth-quarter earnings, depending on operating performance over the first three quarters of the same fiscal year. Since 2000, Korean GAAP has required public firms to disclose quarterly reports to improve timeliness of accounting information. This disclosure requirement is expected to make information users evaluate distinct role of fourth-quarter earnings relative to cumulative earnings over the first three quarters for the same fiscal year. Previous studies have documented that managers tend to increase earnings to avoid net loss or earnings decreases. Such earnings management in turn affects earnings persistence to the extent that managed earnings distort time-series property of earnings over time. In this study we investigate the effect of earnings management through the fourth-quarter earnings, conditional on earnings-increasing incentives, on earnings persistence. When a manager faces poor performance during the first three quarters, s/he has incentives for inflating fourth-quarter earnings to report annual earnings at a target level. Although various means to earnings management can be taken, we focus on a case where a manager makes premature recognition of future revenue or fictitious revenue. However, inference below does not change for cases of omitted or delayed recognition of expenses. Such earnings management causes either the covariance between current-period fourth-quarter earnings and next-period annual earnings to decrease or the variance of current-period fourth-quarter earnings to increase. Then the current-period fourth-quarter earnings persistence, which is defined as the covariance between the current-period fourth-quarter earnings and the next-period annual earnings divided by the variance of the current-period fourth-quarter earnings, becomes smaller for firms with smaller-covariance earnings or larger-variance earnings than for those with larger-covariance earnings or smaller-variance earnings. Thus, we hypothesize that earnings are less persistent for firms with worse operating performance for the first three quarters than for firms with better operating performance for the first three quarters. We partition sample firms into two groups based on incentives for earnings management in two ways. First, the sample firms are partitioned on the basis of whether they report net loss or profit for the first three quarters. The net loss- reporting firms are further classified into two groups: (i) net losses on the first three quarterly reports and net loss on their annual reports and (ii) net losses on the first three quarterly reports but net income on their annual reports. Second, they are partitioned based on whether they report earnings decreases for the first three quarters of the same year. The earnings-decrease firms are further grouped into two: (i) earnings decreases on the first three quarterly reports and earnings decreases on annual reports (ii) earnings decreases on the first three quarterly reports but earnings increases on annual reports. We adopt total accruals and discretionary accruals as a means to earnings management and discretionary accruals are estimated from the cross- sectional modified Jones model by two-digit industry, year and quarter. The sample consists of non-banking firms (1,017 firm-years) with December fiscal years traded over Korean Stock Exchange during 2002-2005. Empirical findings generally support our hypothesis. Fourth-quarter earnings are significantly less persistent for firms that report net losses for the first three quarters than for firms that report net income for the first three quarters. Further, lower persistence of the fourth-quarter earnings is more prominent for firms that report net losses for the first three quarters but net income annually than for firms that report net losses for the first three quarters and net loss annually. The lower persistence of the fourth-quarter earnings is mainly attributable only to discretionary accruals; the persistence of nondiscretionary accruals is not significantly different between two groups. On the other hand, when the sample firms are classified into two groups based on increases or decreases of earnings, the results are qualitatively different from those discussed above. Specifically, the persistence of the fourth-quarter earnings is not significantly different between earnings-increase firms and earnings-decrease firms for the first three quarters of the same year. Further, whether firms report earnings increases or decreases do not significantly affect the persistence of the fourth-quarter earnings and their components for the earnings-decrease firms. These results suggest that there exists incentives for earnings management through fourth-quarter earnings, conditional upon operating performance for the first three quarters and annually. And to the extent that such earnings management introduces noise into the fourth-quarter earnings, it causes either the covariance between current-period fourth-quarter earnings and next-period annual earnings to be smaller or the variance of earnings to be larger so that it affects persistence of the fourth-quarter earnings. However, inconsistent results for earnings management of earnings increases and decreases suggest further exploration for which incentive for earnings management better describes reality.