We first review prior earnings management studies with an emphasis on discretionary accruals as a proxy for earnings management. Discretionary accruals are estimated using widely accepted models, such as the Jones model or its updated ones with additional control variables to improve their prediction power. Nonetheless, estimated discretionary accruals are still subject to model specification errors. Then, we review alternative methods to identify earnings management by evaluating earnings distribution properties and frequencies of digits in collected accounting numbers. These alternative methods can identify anomalies in earnings distributions or frequencies of digits but cannot explain how such anomalies take place. Accordingly, future studies of earnings management may employ these alternative methods in conjunction with discretionary accruals to offer a better insight into earnings management practices.
Meeting Analysts' Expectations Using Discretionary AccrualsMatsumoto (2002) reports that firms use discretionary accruals to meet or beat the mean of analysts' quarterly earnings forecasts. Firms that meet market expectations show a greater frequency of positive discretionary accruals than do their counterparts that miss them. Similarly, Payne and Robb (2000) report that discretionary accruals depend on the relative amounts of the mean of analysts' forecasts in the month preceding the annual earnings announcement and pre-managed earnings. If the former is higher than the latter, firms report positive discretionary accruals. If, however, the former is lower than the latter, firms report negative discretionary accruals.Dechow, Richardson and Tuna (2000) also report high discretionary accruals and working capital for firms that meet analysts' forecasts as compared to their counterparts that do not. Besides, these firms show high market capitalization and high market-to-book ratios with positive abnormal stock returns in the following year. As a result, managers of these firms anticipate optimistic financial outcomes in the near future and thus may attempt to exhibit their private information by meeting analysts' forecasts. Otherwise, their shareholders would be unfairly penalized for missing analysts' forecasts.Burgstahler and Eames (2006) analyze distributions of earnings forecast errors and report a very high frequency of zero and small positive forecast errors and a very low frequency of small negative forecast errors, consistent with the findings of Burgstahler and Dichev (1997). The unusually large number of zero and small positive forecast errors results from the combination of upward earnings management and downward forecast management. In other words, with the end of a year approaching, analysts revise their forecasts downward and managers may engage in managing both cash flows and discretionary accruals upward to meet revised earnings forecasts.In summary, earnings outperforms cash flow from operations in predicting future cash flows in general research settings. Depending on the underlying motivations o...