적자보고 회피 및 이익감소 회피 기업에 대한 재무분석가의 이익예측오차에 관한 연구

This study investigates the association between the error and bias in analysts' earnings forecasts for the year t+1 and the firms' tendency to meet or beat two earnings benchmarks, i.e., profit, and earnings increase, at year t. Prior studies on earnings management largely focus on discretionary accruals to proxy for the earnings management. Although the discretionary accruals are the most frequently used measure to proxy for the earnings management, it has certain limitation. First, the measurement of discretionary accruals requires various estimation processes which makes the measure vulnerable to measurement errors. There have been voluminous studies which pointed out this problem. Second, outside interest parties have difficulty in measuring the discretionary accruals because the processes to measure the accruals are confusing and can not be done easily without computers and by using statistical softwares. Thus, most lay people are not able to use the measure. In contrast, the two benchmarks and the firms' tendency to meet or beat the benchmarks are relatively easier to understand and less prone to the measurement problems. Thus, since the two seminal study of Burgstahler and Dichev(1997) and Degeorge et al.(1999) introduced the measures, many studies have used the measure in order to show the earnings management by firms. Such examples are Burgstahler and Dichev(1997), Burgsthaler and Eames(2003, 2006), Degeorge et al.(1999), and Matsumoto(1999, 2002). These studies consistently report that firms which are likely to fail to meet or beat the two benchmarks manage earnings slightly to move to or above the benchmarks. As a result, there exist a kink in the distributions of the earnings and earnings changes that the number of firms which report earnings or earnings changes exactly equal to zero or slightly above zero are significantly greater than the number of firms which report earnings or earnings changes just below zero. Initially, those prior studies use one more benchmark - the positive earnings surprise (actual earnings - analysts' consensus earnings forecasts) - but we do not use the two benchmark in our analyses because the definition of the third benchmark is directly related to what we are going to test in this study - the analysts' forecast error and bias. In this study, we examine the firms' tendency to manage earnings to meet or beat the two benchmarks in association with the analysts' earnings forecasts in the next period. When a firm meets or just beats the benchmarks, it is likely that the firm manage earnings. If analysts understand this, they may reflect the fact that the firm manage earnings in updating earnings forecasts for the future years. Thus, the tendency could be related to the characteristics of the earnings forecasts. We use the data collected over a period starting from year 2000 to year 2007. All the sample firms are non-financial industry firms and December-year-end firms. The earnings forecasts data available at the end of each quarters are used for the analyses. It is to compare the effects of different time horizon on the characteristics of the earnings forecasts. We collect analysts' earnings forecasts on March, June, September, and December in order to see differential effect of various time horizon. Thus, the sample size varies depending on the analyses. The empirical results are summarized as follows. First, the firms that report zero or slightly non-zero profit show significantly more accurate earnings forecasts and the forecasts are relatively pessimistic, compared with control samples that report slightly non-zero loss. Second, the firms that report zero or slightly non-zero earnings increase show the same trend that these firms show more accurate and less optimistic earnings forecasts. The trend is strongest among the loss avoidance firms, followed by earnings decrease avoidance firms. The findings of this study are interesting and provide various valuable implications. In view of academics, this study links the future earnings forecasts with current tendency for firms to manage earnings. The outside information users also can easily identify these tendency and use their findings in their decision-makings. Analysts and regulators also can use this information.