상이한 이익조정 수단이 재무분석가 예측의 불확실성과 재무분석가 수에 미치는 영향

This study aims to examine the impact of two different earnings management methods, accrual-based and real earnings management, on the uncertainty of the information which analysts provide, and their incentives to forecast. When the information uncertainty is high, financial analysts have not only incentive but also disincentive in providing information. The higher the information uncertainty grows, the larger the investors’ demands are generated, which increases the value of the analysts’ information. On the other hand, information uncertainty also raises the risks for analysts to undertake, which discourages the incentive for analysts to provide information. Under these circumstances, we test how the two different types of earnings management distinguishably affect the incentive or disincentive of analysts to provide information to the market considering information uncertainty. For this, we examine the influences of the two different earnings management on the quality of analysts’ forecast and the number of analyst following. Differentiated from prior literature, this study compares the impact of real earnings management with accrual-based earnings management regarding information uncertainty that analysts face. Earnings management often includes managers’ discretionary decisions and opportunistic behaviors so it is broadly used to measure the level of information uncertainty between managers and investors(Schipper 1989; Kim et al. 2010; Cho et al. 2009). For this earnings management, two different methods are widely recognized, which include accrual-based and real earnings management. Between these two, accrual-based earnings management is promoted by managers’ discretionary choices of their accounting policies or by adjusting the timing to report their earnings(Schipper 1989). The results of accrual-based earnings management are usually reversed within short periods and it tends not to actually manipulate the firms’ real operational activities in the long perspective. However, real earnings management means managers’ abnormal decisions that manipulate firms’ actual operational activities and this harms firms’ long-term performances as it cannot be recovered once it happens(Roychowdhury 2006; Kim et al. 2008). In addition to these, accrual-based earnings management is subject to be audited by external auditors and supervisory institutions so it could be relatively easily found and is clearly restricted by laws and regulations. However, real earnings management is difficult to be distinguished from normal operational activities and also free from external regulations. Based on these ideas, we expect that the impacts of the two different types of earnings management on information uncertainty and the number of analyst following would be differential. Moreover, we analyze the changes in those results with highly earnings management suspicious firms. If firms are highly suspected for earnings management, the interests of the investors as well as information uncertainty on those firms would grow larger. In this case, although higher uncertainty gives the analysts higher information risks, their incentives to meet the investors’ demands by providing forecast information would increase sharply. For these tests, we collect analysts’ EPS forecast data between 2003 and 2007 from I/B/E/S database and the financial data from KIS-Value database provided by NICE Investors’ Services Co. Total 533 samples are used in the analyses. We estimate the firms’ discretionary accruals through Kothari et al.(2005)’s performance-matched discretionary accruals estimation model and the firms’ level of real earnings management using Roychowdhury(2006)’s real activities manipulation estimation model. Then we regress the influences of the two different earnings management methods over the analysts’ forecast accuracy, dispersion, and their coverage respectively. As results of the examination, we first find that accrual-based earnings management does not affect the analyst forecast accuracy and dispersion, but real earnings management significantly decreases the analyst forecast accuracy and increases the dispersion. Second, accrual-based earnings management does not have a significant influence on the number of analyst following, whereas real earnings management significantly decreases the number of analyst following. Third, in the analysis subject to the highly suspicious firms for earning management, real earnings management lowers the analyst forecast accuracy but promotes higher number of analyst following among them. In conclusion, this study implies that the impacts of the two different earnings management on the quality of analysts’ forecast and the number of analyst following are differential because the two different earnings management methods generate different level of information uncertainty. As information uncertainty grows larger, it prevents analysts from providing information in general. However, the grown demands for information as to the highly suspicious firms for earnings management urge analysts to forecast information. On the whole, the results provide evidences that real earnings management promotes higher information uncertainty than accrual-based earnings management because it lowers the analysts’ forecast accuracy and generates disincentive for them to provide information. However, if the value of information increases with grown attention, like those for suspicious firms, the analysts start to provide more information although taking the grown information risks. Together, our results support the idea that analyst coverage actually depends on their private benefits and costs to provide information(Barth et al. 2001).