실제 이익조정이 장기 경영성과에 미치는 영향

Since Roychowdhury (2006) developed estimation models for real earnings management, such as abnormal cash flow from operation (ACFO), abnormal production cost (APC), and abnormal discretionary expenses (ADE), several researches have been conducted using his estimation models. Although his models have contributed to studies on real earnings management in terms of helping researchers overcome the limitations of previous studies that were only able to capture individual operating activities, these models are still vulnerable to measurement error problems. It is argued that managers tend to relax their credit policy to boost sales and they reduce research and development (R&D) expenses to improve accounting numbers. However, similar to discretionary accruals, managers’ activities considered to be real earnings management may deliver their private information (Subramanyam 1996). A manager who is aware that a customer whose operating performance will improve in a later period does not restrain the employee from selling products on credit. In addition, a manager who enhances the efficiency of their R&D process can achieve the same level with a smaller budget. Namely, Roychowdhury’s (2006) metrics may have positive as well as negative meaning. The former is related to private information, whereas the latter is associated with managers’ opportunistic earnings management. If Roychowdhury’s (2006) metrics as to real earnings management embrace managers’ private information instead of earnings management, it will be related to even an higher operating performance in succeeding period. This study investigates whether Roychowdhury’s (2006) measurement is negatively associated with a firm’s performance. Negative relationship can serve as the evidence that the measurement captures opportunistic earnings management on average. The followings are the empirical results of the study. First, real earnings management has a negative effect on long-term operating performance and stock performance. In other words, current real earnings management impairs not only the very next period’s but also the later period’s performances. We also find that stock price reflects this negative impact of real earnings management on long-term performance. Furthermore, this negative impact is more serious in long-term periods than in short-term periods, consistent with Graham et al. (2005). Unlike accruals-based earnings management, real earnings management distorts managers’ operating decisions damaging actual performance. Although discretionary accruals are reversed in next period without a longer effect, the negative impact of real earnings management accelerates with the lapse of time. Second, upward (downward) earnings management has a negative (positive) impact on long-term operating performance. We analyze the differential impact with the direction of real earnings management. We define the earnings management brackets that are suspected to have different earnings management incentives: upward, downward, and unclear. While long-term operating performance decreases with upward real earnings management bracket, it increases with downward real earnings management bracket. Stock price declines only with upward earnings management bracket, showing that investors are more sensitive to bad news. Third, we find that investors can obtain abnormal profit with hedge portfolio based on real earnings management. It implies that firms with larger real earnings management are overvalued because real earnings management brings about mispricing in stock market. This study has some contributions. First, we try to mitigate measurement error problem in Roychowdhury’s (2006) model. In addition to his estimation models, we add the return on asset (ROA) variable to control a firm’s performance like performance-matched discretionary accruals model (Kothari et al. 2005). Although Roychowdhury (2006) included net income as a control variable in the regressions, we cannot control operating performance with his model. It is because the proxy for the real earnings management is independent variable in our analysis. Thus, we eliminate the correlation between real earnings management and firm performance in the estimation stage to minimize the measurement error. Second, we examine the gradual long-term impact of real earnings management on firm’s operating performance with the modified real earnings management model. Our finding suggests that his model contains not managers’ private information but opportunistic earnings management on average. Third, we find that upward and downward real earnings management have differential impacts on the long term performance. While upward real earnings management aggravates future performance, it gets better with downward real earnings management. Our detailed analysis furthers the result of earlier studies.