Empirical Results .1 Descriptive statistics

1905 Table 1 represents the descriptive statistics of RAM and its components for the each year covered by this study. Contrary to our expectations, CFO has slightly positive coefficients. Coefficients of production costs PROD and discretionary expenses DE are positive and negative, respectively, which is consistent with our expectations. Every component moves in direction consistent with RAM from 2003 to 2007, but the values of the coefficients do not keep up increasing or decreasing in the time series [as time goes by]; this indicates that the Accounting Reform Act, which was introduced in 2003, has hardly any influence on the size or direction of earnings management. Panel A: Annual Descriptive Statistics of RAM Year Components of RAM Total RAM CFO Prod DE Average Standard Deviation Average Standard Deviation Average Standard Deviation Average Standard Deviation 2003 0.035 0.197 0.842 0.641 -0.033 0.181 1.121 0.765 2004 0.791 4.046 8.961 31.46 -0.007 0.090 8.905 31.313 2005 0.081 0.378 1.234 1.313 -0.030 0.171 1.344 1.420 2006 0.030 0.213 0.866 0.657 -0.028 0.165 1.087 0.738 2007 0.010 0.199 0.824 0.647 -0.026 0.161 1.096 0.717 Panel B: RAM and Its Components Description Average Standard Deviation Min Value Max Value CFO 0.04572 0.14824 -1.03023 59.34589 PROD 1.79811 0.63668 -0.09344 8.13855 DE -0.02698 0.02767 -1.00000 0.00300 RAM 1.00843 0.66321 0.01153 8.51596 DAM -0.00063 0.21721 -4.75608 5.98581 1906 Table 1 RAM and Its Components, and Descriptive Statistics of DAM Table 2 presents Pearson correlation coefficients between RAM and its components and DAM. CFO PROD DE RAM DAM CFO 1 0.1261 0.000 -0.0016 0.906 0.1168 0.000 -0.1503 0.000 PROD 1 0.0039 0.779 0.9817 0.000 0.1244 0.000 DE 1 0.0071 0.607 0.0011 0.934 RAM 1 0.1137 0.000 DAM 1 Table 2 Pearson Correlation Coefficients between RAM and Its Components and DAM Zang 2006, Cohen and Zarowin 2008 argue that RAM and DAM are substitutes for each other from their finding of s strongly negative correlations between RAM and DAM. However, as can be seen from Table 2, RAM is positively correlated with DAM in the case of South Korea. This indicates that in Korea managers conduct earnings management making use of both DAM and RAM.

4.2.2 Analysis of RAM and a Proxy of Audit Quality

Table 3 displays regression coefficients for each item when comparing RAM with BIG4. Variable Intercept PROXY_AQ 1 if audited by Big4, and 0 otherwise LEV ln_SIZE GROW ∑ Year F Value Adj. R 2 Number of Samples Predicted - + - + 1907 Sign Actual Result 1.366 10.66 0.050 2.62 0.792 18.08 -0.06 -8.37 0.01 42.06 - 144.85 0.165 5087 Table 3 Relationship between RAM and the Reputation of Auditors Whether they are big4 or not We have results which have signs consistent with our expectations for the control variables of LEV, In_SIZE and GROW, and adjusted R-square is 0.161.In an analysis undertaken to examine the relationship between the reputation of auditors and audit quality, RAM and the status of auditors BIG4 or not are slightly positive correlated 0.005, which is to in contrast expectation and would lead to the rejection of the hypothesis that Big4 and RAM are negatively correlated. Possible reasons for this result are as follows: First, there may be limitations to the model we introduce. Or it may be possible that because of limitations of Roychowdhurys Model, we could not capture all of real-activity earnings management, which caused our results to contradict our expectations. Second, according to these results, because it is generally difficult to detect RAM as compared to DAM, Big4 cannot detect real-activity earnings management well. In fact, many cases of accounting fraud have been found in firms audited by Big4. This may be taken to imply that auditors, whether Big4 or not, pass over real-activity earnings management.

4.3 Additional Analyses

4.3.1 Additional Analysis of RAM ‘s Components 1 Comparison with Components of RAM 1908 We analyze the relations between audit quality and three components that make up RAM, i.e. abnormal CFO, production costs and discretionary expenses. The results in Table 4are as set out below: Variable Intercept PROXY_AQ 1 if audited by Big4, and 0 otherwise LEV ln_SIZE GROW F Value Adj. R 2 CFO 2.167 37.73 0.001 0.17 0.152 8.82 -0.090 -29.60 0.002 20.03 76.19 0.2924 PROD 0.969 5.84 0.095 4.51 0.614 12.33 -0.020 -2.28 0.004 14.03 38.34 0.1702 DE 2.159 44.96 0.025 4.12 0.002 0.18 -0.097 -38.27 0.002 26.25 142.07 0.4397 Table 4 Relationship between Components of RAM and Big4 The components of RAM and Big4 are positively correlated; for the Big4, we get higher values of RAM. 2 Analysis of Including Abnormal Cash Flows from Investing Activities within RAM As stated above, the value of RAM consists of abnormal CFO, production costs and selling and administrative expenses, which are components of the income statements. This means that although there may be some possibility of real-activity earnings management through items within the category of selling and administrative expenses in the income statements, these types of earnings management tend to be excluded. Therefore, to supplement to these facts in this study, we include real-activity earnings management, which is exercised through