اثر اهداف جایگزین بر مطالعات مدیریت درآمد : بررسی معیار درآمدها

Firms’ management manages earnings because they have incentives or goals to do so. Earnings management studies have to account for these different goals as tests of earnings management can be compromised by the effect of conflicting goals. I illustrate this in the setting of Dechow et al. (2003). Their study examines whether firms with small profits and firms with small losses (loss-avoidance benchmark) have differing levels of discretionary accruals. Dechow et al. (2003) find that firms just above the loss-avoidance benchmark do not have discretionary accruals that are significantly different than firms just below the benchmark. However, they do not consider firms just below the loss-avoidance benchmark that might be using discretionary accruals to avoid missing an alternative benchmark. I find that after I consider these alternate earnings benchmark goals, firms just above the benchmark have significantly higher discretionary accruals. This provides direct evidence that the ‘kink’ in the distribution of earnings arises from earnings management. I find similar results for the earnings changes benchmark. These findings highlight the need to consider alternative earnings benchmark goals when examining firms immediately around benchmarks.

Earnings management studies are common in accounting literature.1 In many earnings management studies, researchers select firms to be included in a sample because the firms’ management appear to be responding to an incentive or goal.2 However the selected firms may be responding to entirely different goals than the one of interest. These alternative goals may confound research results. In this study I illustrate this issue by examining firms around the loss-avoidance (zero earnings level) benchmark. Prior research shows that firms do not appear to be managing earnings, as measured by discretionary accruals, to meet or beat the loss-avoidance benchmark (Dechow et al., 2003). I find that this lack of results appear to be driven by firms’ management responding to alternative earnings benchmark goals. Extensive research documents that firms manage earnings to beat benchmarks, such as zero earnings level, earnings changes, and analysts’ consensus forecasts (Ronen and Yaari, 2008).3 While the literature recognizes the potential conflict among these benchmarks (e.g. Einhorn, 2007 and Graham et al., 2005), to date, research has looked at each benchmark in isolation. In this study, I fill this lacuna by examining the properties of discretionary accruals for small loss firms. I hypothesize that firms with small losses use discretionary accruals to maintain (or establish) positive earnings changes (earnings improvement benchmark), or to meet or exceed analysts’ forecasts (analyst forecast benchmark), even if positive earnings are unattainable. Thus firms missing one benchmark may use discretionary accruals to meet an alternative earnings benchmark. After controlling for alternative benchmarks, I find that firms with small profits have discretionary accruals that are significantly higher than the resulting sample of firms with small losses. These results provide an explanation for the findings of Dechow et al. (2003). In particular, I show that their result seems to be driven by firms from the small loss sample that may have manipulated discretionary accruals upward because of opportunities to beat an alternative benchmark. Prior research (Ayers et al., 2006) finds that firms do not appear to be managing earnings to meet or beat the earnings improvement (earnings changes) benchmark. I also hypothesize that firms with small negative earnings changes use discretionary accruals to maintain positive earnings (loss-avoidance benchmark) or to meet or exceed analysts’ forecasts (analyst forecast benchmark). I examine whether firms that report small increases in earnings (as compared to the previous year’s earnings) have higher levels of discretionary accruals than firms with small decreases in earnings. For the full sample of firms with small earnings changes, I do not find that firms with small earnings increases have discretionary accruals that are higher than firms with small earnings decreases. When I control for firms that may be responding to alternative benchmarks (loss-avoidance and analyst forecast), I find that firms with small increases in earnings have higher discretionary accruals than firms with small decreases in earnings. My study contributes to the literature that attempts to explain whether the kink in the distribution of earnings (i.e. Burgstahler and Dichev, 1997) provides evidence of earnings management (Ayers et al., 2006, Beaver et al., 2007, Dechow et al., 2003 and Durtschi and Easton, 2005). The initial explanation was based on the conjecture that unmanaged earnings should have a normal distribution. Hence, an observed deviation from a normal distribution is consistent with earnings management. Mine is the first study to provide direct evidence that the kink arises from firms indeed managing earnings. Previous studies either did not find earnings management (e.g. Dechow et al., 2003), found it for a specific accrual (e.g. Phillips et al., 2003), provided circumstantial evidence (e.g. Jacob and Jorgensen, 2007), or provided alternative explanations that refutes earnings management (e.g. Beaver et al., 2007).4 A second contribution of my study is that it points out that care needs to be taken by researchers in earning management studies when firms have different incentives or goals than the one of interest—these alternative incentives or goals may affect the measures of earnings management. Ayers et al. (2006) find that forward-looking discretionary accruals are higher for firms just above the loss-avoidance benchmark than for firms just below and that these results are more pronounced than pseudo-benchmarks along the earnings level distribution. They do not reconcile their results to Dechow et al. (2003). Ayers et al. (2006) use a different time period than Dechow et al. (2003). I investigate differences in abnormal accruals between the two different time periods and find that discretionary accruals appear more pronounced just above the loss-avoidance benchmark in recent years (i.e. 2001–2002). I also find that in recent years, firms just below the benchmark do not appear to be responding to opportunities to meet alternative benchmarks like they did from 1988–2000. Consistent with Ayers et al. (2006), I control for performance as an alternative explanation for my findings. I delete high performance firms (firms that meet or beat multiple benchmarks) from just above the benchmarks and again compare the discretionary accrual levels just above and below the benchmarks. Results from these additional tests are consistent with initial findings for the loss-avoidance and earnings improvement benchmark. These results suggest that firms above the loss-avoidance benchmark that meet multiple earnings benchmarks may be smoothing earnings. Finally, I run OLS regressions with abnormal accruals as my dependent variable and alternative benchmarks as my independent variables for samples that cover the entire earnings levels and changes distributions. These results confirm the univariate tests and highlight firms, other than those immediately around the earnings benchmarks, which appear to manage earnings upward or discretion-save (save abnormal accruals for future periods) when they may have responded to opportunities to meet or beat alternative benchmarks. In summary, I contribute to the earnings management literature in the following ways: (1) I find an explanation for the results of Dechow et al. (2003), where discretionary accruals for firms just above the loss-avoidance benchmark are not significantly different than the firms just below. I find that alternative benchmarks (earnings improvement and/or analyst forecast) impact observed discretionary accruals just below the loss-avoidance benchmark; (2) Similarly, I also find that the loss-avoidance benchmark affects observed discretionary accruals just below the earnings improvement benchmark. After controlling for these impacts below the loss-avoidance and earnings improvement benchmarks, evidence suggests that firms record discretionary accruals in a manner consistent with earnings management to beat benchmarks. Mine is the first study to provide direct evidence that the kink in the distribution of earnings arises from earnings management; (3) I do not consider each benchmark in isolation. I highlight the need for researchers and practitioners to consider that firms that miss one earnings benchmark may be managing earnings because of an opportunity to meet or beat an alternative earnings benchmark or goal; (4) I bridge the results of Ayers et al., 2006 and Dechow et al., 2003. I find that the significant difference in abnormal accruals for firms immediately around the loss-avoidance benchmark (Ayers et al., 2006) can be explained by the decline in firms just below the loss-avoidance benchmark using discretionary accruals to meet alternative benchmarks in more recent years (i.e. 2001–2002); and finally (5) I provide evidence that firms throughout the earnings level and earnings changes distributions (other than those immediately around the benchmarks) sometimes use discretionary accruals in a manner consistent with earnings management when certain alternative benchmarks have been met. These firms use discretionary accruals in a manner consistent with discretion-saving when other alternative benchmarks have been met. In their conclusion, Dechow et al. (2003, 24) state: “another possibility [to explain lack of results for discretionary accruals of small profit firms as compared to small loss firms] is that we are measuring discretionary accruals correctly, but that the earnings management story is more complex than the one we test… We leave the examination of more complex earnings management stories to future research.” I provide a more complete story for discretionary accruals around the loss-avoidance and earnings improvement benchmarks—where firms just below benchmarks use discretionary accruals to meet or beat an alternative earnings benchmark goal and inflate the overall level of discretionary accruals for firms just below benchmarks. The remainder of the paper proceeds as follows. Section 2 gives background and outlines the relevant literature. Section 3 presents the research design. Section 4 reports the results. Section 5 summarizes and concludes the paper.

The results of this study suggest that the ‘loss avoidance story’ put forth in Dechow et al. (2003) appears to be a portion of a more complex story where firms close to meeting one earnings benchmark may have used their discretion to meet different earnings benchmark goals. Also, the findings highlight the need to for researchers to consider alternative goals or objectives when conducting earnings management research. These findings may be especially helpful for researchers that do not find results consistent with earnings management when such results are predicted by theory or incentives. First, using a sample consistent with Dechow et al. (2003), I find that discretionary accruals for firms just above the loss-avoidance benchmark are not significantly higher than firms just below. Using a more recent sample, I also find similar results for the earnings improvement benchmark. Next, I identify firms with small losses that have met or beat the earnings improvement benchmark and/or the analyst forecast benchmark. These firms may have managed earnings (using discretionary accruals) to meet or beat the alternative earnings benchmark goals. When I remove firms from the small loss sample that may have responded to incentives or goals to meet or beat the earnings improvement benchmark and/or analyst forecast benchmark, firms with small profits report higher discretionary accruals than firms with small losses. In essence, I create switches for the three benchmarks (loss-avoidance, earnings improvement, and analyst forecast) and leave one switch on (Fig. 1 ‘white space’) while turning off the other two. The benchmark that is turned on is not confounded by alternative goals or incentives to meet a second or third benchmark. When I examine the loss-avoidance benchmark and delete the earnings improvement benchmark and/or analyst forecast benchmark, I find that firms appear to be using discretionary accruals to manage earnings. Additional analysis shows that firms that have large changes in earnings drive the increase in the level of discretionary accruals for small loss firms. I find similar results when I examine the earnings improvement benchmark. When I ‘turn on’ the earnings improvement benchmark and delete firms from the small earnings decreases sample that may have responded to incentives to meet or just beat the other two benchmarks (loss-avoidance and analyst forecast), firms with small earnings increases have higher discretionary accruals than firms with small earnings decreases. Firms also appear to use discretionary accruals to avoid small negative earnings changes (beat the earnings improvement benchmark). Additional analysis shows that the increase in the level of discretionary accruals for small earnings decrease firms is partially driven by firms that are trying to just beat the loss-avoidance benchmark (obtain a small profit). One caveat18 is that I only consider alternative earnings benchmarks, when there are a large variety of incentives that may influence a firms discretionary accrual levels just below a benchmark. These other incentives might include debt covenants or compensation contracts and it is important to note that these additional incentives might also affect the results around earnings benchmarks. I leave these additional incentives or goals to future research. My results provide evidence as to why: (1) firms with small losses have discretionary accruals that are not significantly different from that of firms with small profits and (2) firms with small negative earnings changes have discretionary accruals that are not significantly different from that of firms with small positive earnings changes. The small loss sample just below the loss-avoidance benchmark contains firms that could be managing discretionary accruals to meet other earnings benchmarks (earnings improvement and/or analyst forecast). The small earnings decreases sample just below the earnings improvement benchmark contains firms that could be managing discretionary accruals to meet other benchmarks (loss-avoidance and/or analyst forecast). After controlling for firms that may have responded to incentives to meet or beat these alternative earnings benchmarks, results suggest that firms record discretionary accruals in a manner consistent with earnings management. I find some evidence that firms just above the loss-avoidance benchmark that also meet or beat the analyst forecast benchmark may be ‘discretion-saving’ to help out with future earnings. I also find evidence that firms not immediately around benchmarks in the earnings levels and earnings changes distributions appear to respond to opportunities to meet or beat alternative benchmarks: sometimes with abnormal accruals that suggest earnings management and sometimes with abnormal accruals that suggest discretion-saving behavior. These findings suggest a more complete earnings management story than the ‘loss avoidance story’.