Empirical model Conclusion, Limitations and Future Research

215       t t t t t t t t t t t NI MTB SIZE OWNH EM EM OWNH EM EM OWNH EM EM Y                                     9 1 8 1 7 6 5 4 3 2 1 3 3 2 2 1 1 9 where Y REM, ACFO, APC, and ADE; REM ACFO+ADE+APC; ACFO 1×Residual from the estimation model of equation 1; APC Residual from the estimation model of equation 2; ADE 1×Residual from the estimation model of equation γ; EM1 1 in companies with NI slight ly greater than 0 0≤NI≤0.015, 0 otherwise; EM2 1 in companies with NI much greater than 0 0.075≤NI, 0 otherwise; EM3 1 in companies with NI much less than 0 NI≤-0.075, 0 otherwise; OWNH 1 if the proportion of majority shareholders‘ ownership is higher than median value in each year, 0 otherwise; SIZE Natural log of total assets; MTB Market value of equity book value of equity; and NI Net income beginning total assets. In hypothesis 2, to compare and analyze the earnings management behavior of companies under different earnings management incentives brackets, we used equation 9. The dependent variables in estimation model 9 are identical to those in equation 8 used to test hypothesis 1. The independent variables of interest are EM1, EM2, EM3, an interaction term between each variable and OWNH, a dummy variable that indicates whether the majority shareholders‘ ownership is high. Eε1 was used to test hypothesis βa with a sample of companies with earnings slightly greater than zero, EM2 to test hypothesis 2b with a sample of companies with earnings much greater than zero, and EM3 to test hypothesis 2c with a sample of companies with earnings much less than zero. The dummy variable, OWNH, takes a value of one if the majority shareholders‘ ownership is higher than the median, and a value of zero otherwise. Insert TABLE 2 about here 216 The expected signs of each variable of interest are summarized in TABLE 2. First, the companies with high majority share holders‘ ownership tend to decrease real earnings management. OWNH, the variable of interest in hypothesis 1, is therefore expected to show significantly negative values. Second, the EM1 bracket containing companies with earnings slightly greater than zero is suspected to contain companies with upward earnings management. However, real earnings management incurs the sacrifice of the future value of the company, i.e., business myopia. Thus, while companies with low majority shareholders‘ ownership will perform upward real earnings management, those with a high proportion of majority shareholders will significantly control upward earnings management. As a result, companies with a low proportion of majority shareholders will have high Y REM, ACFO, APC, and ADE , while those with high majority shareholders‘ ownership will have a smaller Y, according to the convergence-of-interests hypothesis. In a regression equation, the EM1 variable was positively correlated with Y, but the Eε1×OWNH variable was significantly negatively correlated. Third, the EM2 bracket containing companies with earnings much greater than zero is expected to practice downward earnings management. Their future corporate performance is expected to improve due to downward real earnings management, and it is therefore unlikely that majority shareholder monitor managers‘ myopic real earnings management. As a result, real earnings management will be done in such a way as to decrease earnings, regardless of whether majority shareholders‘ ownership is high or low. Thus, the EM2 variable is expected to show a negative correlation with the dependent variable and Eεβ×OWNH is predicted to be insignificant. That means real earnings management is not expected to differ depending on the level of ownership. 217 In the EM3 bracket containing companies with earnings much less than zero, however, it is difficult to predict the direction of real earnings management beforehand, regardless of the proportion of majority shareholders‘ ownership, because the impacts of upward earnings management for decreasing deficits and downward earnings management for the big bath both apply. Thus, it is expected that both the EM3 and Eεγ×OWNH variables will not have significant relationships with proxies for real earnings management i.e., REM, ACFO, APC, and ADE. As control variables, we first added the natural log of assets SIZE at the beginning of the period, considering that firm size has an impact on earnings management according to the political cost hypothesis, which states that a company tries to reduce earnings as it becomes bigger Jones, 1991. Second, we included the market-to-book ratio MTB, which accounts for growth opportunities in the market. We expect that as a company has more opportunities for growth, there is a downward earnings management incentive Roychowdhury, 2006. Third, Dechow et al. 1995 and Guay et al. 1996 pointed out that earnings management incentive is related to firm performance. Hence, net income NI was included to control for firm performance.

4.4 Data Collection

The sample was selected from companies listed on the Korea Stock Exchange as of December 31, 2007 that satisfied the following criteria: 1 companies except financial companies listed on the Korea Stock Exchange, with their accounts closing in December; 2 companies with financial statements available for extraction using the KIS-Value of Korea Investors Services; 3 companies with information available about majority shareholder ownership using TS2000 of the Korea Listed Companies Association. We used 17 years of data 1991 to 2007 from 7,358 companies that satisfied the above conditions. Samples were classified into 13 groups according to industry SIC code 218 to estimate real earnings management. To eliminate bias, 918 samples that were in an industry-year group with less than 30 samples were excluded from the sample pool. Furthermore, to eliminate the effect of outlier bias, the top and bottom 1 of independent and dependent variables were winsorized. A total of 6,440 firm-year samples were used for the analyses. TABLE 3 shows the majority shareholder ownership of the companies. The majority shareholders‘ ownership of all companies was about γ0. It was about 17 in the lower- level group, and 43 in the higher-level group. When we compared the majority shareholders‘ ownership of companies according to earnings management brackets Eε1, EM2, and EM3, we found that it was almost the same for EM1 and EM3 but relatively larger in EM2. Insert TABLE 3 about here

V. EMPIRICAL RESULTS 5.1 Descriptive statistics

The descriptive statistics of the variables used in the analyses are presented in TABLE 4. We compared the lower ownership group with the higher ownership group for all samples and for each earnings management bracket EM1, EM2, and EM3. First, we compared the difference between earnings management brackets. Abnormal cash flow from operations ACFO, abnormal production costs APC, and abnormal discretionary expenses ADE were relatively small in EM1 and EM3, but large in EM2 compared to the full sample. This difference implies upward earnings management in EM1 and EM3, and downward earnings management in EM2.