Sample Selection and Research Methods

549 of research. In this research we use discretionary accruals as the proxy of earning management and we use 3 models to minimize errors from the model setting. First we use the modified Jones model Dechow et al. 1995. Prior research documents that the modified Jones model is generally effective. Our first model is described as follows: TABTA= b + b 1 ΔREV i - ΔREC i BTA i + b 2 PPE i BTA i +e i 1-1 Here, TA total accruals = NI net income – CFO cash from operations; REV = net sales revenue; REC = trade receivables; PPE = property, plant, and equipment; BTA = beginning total assets; Δ = change operator The discretionary accruals are obtained by subtracting fitted values of accruals that is, non-discretionary accruals, from the total accruals as follows: DAi = TAi BTAi –[b0 + b1 ΔREVi –Δ RECiBTAi + b2 PPEiBTAi]+e i 1-2 Yoon and Miller 2002 document that the modified Jones model does not fit well, particularly for Korean firms. Kothari et. al 2005 suggest that it is better to give an additional independent variable to control firms‘ performance in the modified Jones model when estimating discretionary accruals. Yoon and Miller 2002 find that cash from operations is the major determinant of accruals. So we include cash from operations as an additional independent variable as a control variable of performance. Our second model is described as follows: TABTA= b + b 1 ΔREV i - ΔREC i BTA i + b 2 PPE i BTA i + b 3 CFO i BTA+e i 2 The way to have the discretionary accruals and variables are same with model 1-2. Lastly Kothari et al. 2005 suggest that discretionary accrual models may be mis- specified when applied to firms with extreme past performance, proposing that a matching procedure based on performanceROA is more appropriate for these firms. Return on assetsROA is net income deflated by total assets. Consistent with 550 Kothari et al. β005, we implement the ‗performance-matched discretionary accrual model. Our third model is described as follows: TABTA= b + b 1 ΔREV i - ΔREC i BTA i + b 2 PPE i BTA i + b 3 ROA i +e i 3 The way to have the discretionary accruals and variables are same with model 1-2. 3. 2. 2. Regression models First we examine whether corporate name changes have an effect on earnings management. NC is the main variable in our model and we add control variables, we regress DA1 2, 3 on NC and controlling for CFO, LEV, SIZE, GRW. DA1 DA2, DA3 it = b o +b 1 NC+ b 2 CFO it +b 3 LEV it +b 4 SIZE it +B 5 GROW it +e it 4 Next we regress of DA 12, 3 on LC, IC, OC and controlling for CFO, LEV, SIZE, GRW to investigate whether there are differences among the reasons for corporate name changes. LC is a dummy variable which has a value 1 when a firm report current loss and IC is a dummy which has a value 1 when a firm change industry from consolidation or change the primary products. OC is a dummy which has a value 1 when largest stockholders are changed for corporate name change firms. DA1 DA2, DA3 it = b o +b 1 LC+ b 2 IC+b 3 OC+b 4 CFO it +b 5 LEV it +b 6 SIZE it +B 7 GROW it +e it 5 Model 6 is for investigating the effect of interaction of multiple factors on discretionary accruals. We add 4 interaction variables in model 5. LCOC is an interaction dummy for name change from loss-reporting and largest stockholders change. LCIC is an interaction dummy for name change from loss-reporting and industry change. ICOC is an interaction 551 dummy for name change from largest stockholders change and industry change. LCICOC is interaction dummy from loss-reporting and largest stockholders change and industry change. Other control variables are same with model 4. DA1 DA2, DA3 it = b o +b 1 LC+ b 2 IC+b 3 OC+b 4 LCIC+b 5 LCOC+b 6 ICOC+b 7 LCICOC +b 8 CFO it +b 9 LEV it +b 10 SIZE it +B 11 GROW it +e it 6

3. The Trend of Corporate Name Change

In this section, we briefly describe the corporate name changes in Korea in terms of disclosure frequency, industry dispersion, the reasons for change, the pattern of changed name. And we compare financial features with non-changing firms and changing firms. 3. 1. The frequency of corporate name change According to KIND of Korea Exchange, 131 KOSDAQ firms and 53 KSE firms have changed their name in 2008 and they were 110 firms of KOSDAQ and 40 firms of KSE in 2007. Figure 1 report the number of corporate name change firms from 2003 to 2008. We find that they are steadily increased, especially in the KOSDAQ market. [Figure 1] Frequency of corporate name change 50 100 150 200 250 2005 2006 2007 2008 KSE KOSDAQ TOTAL 2004 552 From 2004 to 2008, there are 24 firms changing their corporate names twice in a year and 91 firms have changed their name twice for 5 years. Twenty four KOSDAQ firms have changed their names twice in a year and three firms have changed their names up to five times over the period of 2004 to 2008. So we examine the features of corporate name change based on KOSDAQ firms. Figure 2 reports industry dispersion of corporate name change firms from 2004 to 2008. 98 Electronic·computer Manufacturing companies changed their names and 90 Broadcast and media companies and 70 service companies changed their names. It means corporate name changes happen in all over the industry, not in some industry. [Figure 2] Industry dispersion of corporate name changes 3. 2 The Reason for Corporate name changes According to the previous study, the reasons for corporate name change could be due to show expanded product offerings and strategic direction, to reflect company diversification and expansion, to provide a more universally representative name, and to reflect new identity following a change in ownership. Based on previous study we partitioned the reasons for corporate name changes into 6 cases from 2004 to 2008. The largest stockholders change, the foreign ownership and export proportion data were retrieved electronically from KIS-VAUE database and the data about primary product change, consolidation and diversification during the study period were manually collected from KIND of Korea Exchange. Table 1 report the partition of reasons for corporate name change firms who were able to figure out. 20 40 60 80 100 120 Computer Hardware Electronic manufacturing Wholesale Trade BroadcastingMedia Services Non-elect. Machinery Others.