내생성을 고려한 소유집중도가 이익조정에 미치는 영향: 패널데이터를 이용한 정적 및 동적 분석

This study examines whether a firm's corporate governance practices have an effect on the quality of its publicly released financial information. In particular, we examine the relationship between ownership concentration and the extent of corporate earnings management as measured by abnormal accruals and real earnings management activities. There is a long-standing debate on the effect of ownership concentration on earnings management in existing archival research. Our study has a distinctive feature since the Korean ownership structure is characterized by the dominance of one largest shareholder and his/her family members who generally participate in the management of the firm. The largest shareholders and specially related persons are highly likely to influence most of important corporate decisions whether directly or indirectly and thus ownership may be indifferent from management. Our study examines two conflicting hypotheses based on two conflicting theories: 1) the convergence of interest hypothesis (based on stewardship theory) and 2) the expropriation of minority shareholders hypothesis (based on agency theory). Agency theory suggests that the interests of opportunistic, self-interested agents conflict with those of principals whereas stewardship theory suggests instead that agents' interests will be aligned with firm interests and thus be more intrinsically motivated than predicted by agency theory. Since ownership is not generally separated from management in South Korea, we focus on agency relationships between controlling shareholders and minority shareholders rather than those between ownership and management. Under the convergence of interest hypothesis, agency costs between controlling shareholders and minority shareholders should be reduced which in turn reduce controlling shareholders' opportunistic behavior. On the other hand, the expropriation of minority shareholders hypothesis suggests that controlling shareholders may infringe on minority shareholders' right in their favor, by influencing managers to engage in earnings management. Although many of previous studies attempted to find true relationship between ownership concentration and earnings management considering two conflicting hypotheses, the results are still inconclusive. Empirical corporate accounting research, which attempts to explain the causes and effects of financial decision, often has serious issues with endogeneity. This is because it is generally difficult to find exogenous factors or natural experiments with which to identify the relations being examined. Despite previous literatures have reported mixed results, many of these papers suffer from model mis-specification, limited data and failure to control for endogeneity issues since their results are based on the normal OLS regression using pooled data which often ignores potential heterogeneity originating from unobservable factors. Noticeable gap in the literature is the inability of authors to control for firm heterogeneity, dynamic effects and also other factors such as firm characteristics. In this paper, we examine static and dynamic panel data analysis of the effect of ownership concentration on earnings management, econometrically considering possible endogeneity issues caused by unobservable heterogeneity. To address the relation between ownership concentration and earnings management, we borrow from an extensive previous literature to justify that our main variables are plausible proxies for earnings management. We specifically partition earnings management into two aspects: accrual based earnings management and real earnings management. We further use two measures of abnormal accruals; 1) the modified Jones model suggested by Dechow et al.(1995) and 2) the performance-adjusted modified Jones model suggested by Kothari et al.(2005) and real activities manipulation measure suggested by Roychowdhury(2006). Meanwhile, the percentage of ownership allocated to the largest shareholder and specially related shareholders is used as a proxy variable for ownership concentration. We examine the relation between ownership concentration and earnings management using random effect, fixed effect and the dynamic effect method for a panel of 2,984 firm-year observations over a period from 2000 to 2010. We also provide results from the pooled OLS and the between model regression for comparison purposes. The results suggest how serious econometrical bias may be when ignoring those potential endogeneity issues. Our empirical results are summarized as follows. First, ownership concentration overall exhibited positive sign(+) on earnings management activities from the fixed effect model in consideration of auto correlation, albeit strongly consistent evidence was not found since overall coefficients were either insignificant or marginally significant. The results are, in general, consistent with those obtained from dynamic effect model and other additional panel data analysis. Second, the effect of foreigners and institutional shareholders on earnings management was not significant. We illustrate why other commonly used estimators that ignore the endogeneity issues and dynamic relationship between ownership concentration and past firm history might be biased. The environment where ownership is highly concentrated (on average) is one of the distinctive features of Korean capital market. We expect to provide valuable insight into true association between ownership concentration and earnings management through a well-developed panel data analysis.