This study examines whether firms releasing earnings reports later than expected engage in earnings management. Extensive research has identified various motives for earnings manipulation. Previous research has documented that early earnings announcements are associated with good news and that reporting delays are associated with market anticipation of bad news. For example, Givoly and Palmon (1982) find that price reactions are more pronounced for early announcements than for late announcements. Further, Chambers and Penman (1984) report that, on average, firms publishing earnings reports later (earlier) than expected display stock price declines (rises). Several possible explanations for announcement delays are offered in the literature. For instance, managers may be attempting to effect planned stock sales or negotiate contracts in the best possible light prior to the disclosure of unexpected bad news. Trueman (1990) analyses two other managerial motives for delaying bad news. First, extra time is required to undo the bad news through accruals manipulation. Second, management might deliberately delay bad news until other industry-wide bad news is released in order to justify the potential reputational and litigation costs. Moreover, firms subject to high degree of political pressure may also delay their announcements. Han and Wang (1998) document that petroleum refining firms delayed reporting earnings and selected income-decreasing accruals during the 1990 Gulf crises when they made extraordinarily huge profits. Most important, their findings are inconsistent with Chambers and Penman (1984). Apparently, the incentive to report `good news' early is outweighed by political costs. Given the market anticipation of bad news, we hypothesize that late announcers depress reported earnings because future expected earnings and bonuses would be improved. Besides, it becomes more advantageous for them to engage in contractual renegotiations (DeAngelo et al., 1994).

Earnings management has generally been tested on accounting accruals. Managers favor using accruals to manipulate earnings since it is difficult for outsiders to undo the effects of discretionary accruals given limited or unavailable information (Schipper, 1989). Our sample consists of 2,045 late reporters during the four-year period from 1991 to 1994. Non-discretionary accruals are measured using a cross-sectional modified Jones model (Dechow et al., 1995). Many recent studies (DeFond and Jiambalvo, 1994; Becker et al., 1998; and DeFond and Subramanyam, 1998) adopt this approach as it allows analysis of the entire sample and thus overcomes the survivorship bias problem in the time-series counterpart. We also perform multivariate tests that control for several confounding factors. Our results are consistent with the hypothesis that late reporters exhibit negative abnormal accruals in the year of late announcements.

Ideally, financial reporting conveys information about firms' performance and facilitates the resource allocation efficiency on a timely basis. Unfortunately, market forces alone are inadequate to bring the disclosure of all material information. Evidence on the pervasiveness of earnings management sheds light on the interaction between financial reporting discretion and announcement timing. Finally, our study contributes to the earnings management literature by understanding the implications of management's motives and the possible impact of 'big bath' accounting on resource allocation in the economy.

The remainder of this paper is organized as follows: Section 2 develops the research hypothesis, Section 3 describes the data and method, Section 4 reports the empirical results, and the final section contains our summary and conclusions.