Japanese managers' earnings management using several different types of reserve accounts.
The Japanese economy was booming in the late 1980s. Now it is slowly trying to regain its economic glory days which preceded the market crash in 1990. Many of its leading industries lost their worldwide leadership positions to South Korea and other countries. Samsung has the highest market share in the world TV market surpassing Sony and Panasonic, even though Toyota is still the best auto manufacturing company in the world (http//news.chosun.com, February, 27, 2007). Since Japan does not own natural resources such as oil and lumber, most of their success depends on exports and the world economy. Currently China and other developing countries are demanding increasing quantities of oil. As a consequence, Japan still will continue to face hard times as long as oil prices remain high.
Japan is a unique country from the capital structure perspective. The Japanese capital structure has moved toward that of the U.S. and other western countries' systems after the market crash,. However, their government still has a large impact and their system is more tax oriented than those of western countries. In addition, Japanese corporations have more reserve accounts than those of any other country in the world. It is interesting to see how their corporate managers use these reserve accounts to manage corporate earnings to meet their goals.
Previous studies have used U.S. capital market data to understand managers' motivations regarding earnings management. Managers may have motivation to manage earnings to maximize their bonuses or to maintain target income levels. However, the U.S. and Japan have different capital and business environments. Under different capital and business environments, managers may or may not behave the same way to satisfy their self-interests. This is an interesting issue that we need to test using actual Japanese data.
The purpose of this study is to test the earnings management behavior of Japanese managers using different types of reserve accounts. Typical Japanese firms maintain ten or more reserve accounts and they are legally allowed to set up reserves for their guaranteed obligations starting from April 20, 1982 (Takahashi, 1984). (1) The total amounts of these reserve accounts are very large for some firms, creating the proper environment for possible earnings management using these accounts. This is the first empirical study to study Japanese managers' earnings management motivations using different types of reserve accounts.
The paper is organized as follows: Section 2 presents the background of earnings management and the different capital and business environments in Japan. The third section describes our research methodology and data collection procedures. The fourth section reports results and discusses sensitivity analyses. The last section concludes with a summary of our findings and future research avenues.
Evans and Sridhar (1996) defined "earnings management" as an activity which incorporates any accrual-based manipulation of economic earnings by the manager. Their study predicted that within less flexible financial reporting systems, managers' compensation would depend on reported earnings. However, if there are ten or more reserve accounts like may be present in the Japanese financial system, mangers are tempted to use some of these reserve accounts to manage earnings to meet their target income. Moses (1987) studied firm specific factors that may provide motivation for U.S. managers to manage earnings by choice of accounting methods. He showed that earning management was associated with factors such as firm size, the existence of bonus compensation plans, and the variation of actual earnings from expected earnings.
Several studies used discretionary accruals to study managers' earnings management behavior. While these studies are not in the same context as our study, they found that managers used their discretion to adjust their firm's reserve accounts and some of the same motivational factors may be relevant for our study. See, for example, Jones (1991), Dechow and Sloan (1991), DeAngelo et al. (1994), Maydew (1997), and Kasznik (1999).
Early studies tested earnings management motivation using only U.S. data. However, even if we consider the unique financial and business environments in Japan, there is some empirical evidence to sustain the U.S. findings in the Japanese context. Herrmann and Inoue (1996) showed that under certain operating conditions, firm size, income taxes, capital intensity, deviation in operating activities, and earnings variability are all related to motivations of managers to manage earnings using depreciation changes in the Japanese context. Herrmann et al. (2003) later also find that income from sales of assets and management forecast error have a negative relationship after controlling for the debt-to-equity ratio, firm size, growth, and last year's income from asset sales.
Ali and Hwang (2000) find that value relevance is lower for Japanese firms since Japan has bank-oriented financial systems, private-sector bodies are not involved in the standard-setting process, uses the Continental accounting model, uses tax-based accounting, and more is spent on external auditing services.
Darrough et al. (1998) find that the debt-to-equity and asset hypotheses hold in Japanese firms only after the 1990 market crash. Even though Japanese managers may have different financial and business environments than their U. S. counterparts, such as Japanese accounting is more tax oriented, their firms share cross-holdings (keirutsu) with other firms, and their managers are more long-term goal oriented, Japanese managers still choose income increasing accounting accruals to increase their bonuses and increase the amount of outside funding.
Other factors may change managers' earnings management behavior across countries. For example, Black and White (2003) compared earnings and balance sheet information in Japan, Germany and the U.S. They assumed that accounting in Japan and Germany is conservative, more debt is financed with main banks, and accounting is more tax-based when compared to U.S. Their findings show that in the U.S. positive earnings are more value relevant than the book value of equity, but not negative earnings. However, Germany has the opposite results and Japan shows mixed results.
Kumar and Hyodo (2001) studied Japanese and U.S. price-earnings (P/E) ratios between 1975 and 1995. Their study used parent-only reported earnings which is more reliable than consolidated earnings in Japan. They find higher price-earnings ratios in Japan compared with the U.S. Possible reasons for higher P/E ratios in Japan are: Japanese firms' reported earnings are smaller because they allow special reserves for future uncertainties such as retirement benefits, product returns and repairs and also allow larger depreciation expenses. Some of these factors are relevant for our current study.
Income Smoothing Measure
The absolute value of the differences between the changes in earnings before reserve (EBR) accounts with the change in earnings (E) deflated by sales is used as a measure of earnings management (EM) as was used by Herrmann and Inoue (1996): (2)
EM = |[EBR.sub.t] - [EBR.sub.t-1]| - |[E.sub.t] - [E.sub.t-1]| / SALES
where EM = earnings management
EBR = earnings before reserve accounts in period t and t-1
E = earnings in period t and t-1.
The higher book values of each reserve account or any combination of several reserve accounts maintained by Japanese firms may increase the opportunities of managers to smooth earnings (Darrough et al., 1998). Sales are used as a deflator because sales are less prone to manipulation. The expected sign of EM is positive as positive values of EM are consistent with Japanese managers' earnings management motivations.
Various Factors for Earnings Management
Prior studies used the following factors as independent variables for studying accounting choices and economic consequences.
Larger firms usually have higher political costs since their activities can be easily exposed via media and they need to protect their reputation (Watts and Zimmerman, 1986). If larger firms miss their target income or have bad news, investors or creditors will pay more attention. Therefore, managers may have higher motivation to manage earnings. However, this may not be the case for Japanese firms since most of their larger firms are multinational firms and they can diversify their risk easily. In addition, most larger firms are backed by main banks and belong to one of six keiretsu groups. Some larger firms may belong to the same keiretsu group as media firms. Therefore, it may be easy for Japanese firms to mitigate their political costs. From the above discussions, no sign is predicted for the relation between firm size and the earnings management variable. Firm size is measured by the natural log of total sales consistent with most other studies.
Generally, individuals or corporations do not want to pay higher taxes. For Japanese managers their requirement of tax-financial reporting conformity provides higher motivation to manage earnings by reporting systematically low profits and consequently paying less tax (Choi and Meek, 2005). It is expected that firms with greater current tax liabilities are more likely to manage earnings. Therefore, the EM variable and the tax variable are expected to have a positive relationship. Current year income taxes are also deflated by sales.
Moses (1987) provides evidence using U.S. data that managers have motivation to manage earnings to maximize their bonus compensation. However, the bonus amount paid to Japanese managers is relatively smaller when compared with that of U.S. managers (Herrmann and Inoue, 1996). Additionally, Japanese managers sometimes work as a team with bonuses distributed based on seniority. Therefore, the impact of a bonus compensation variable in Japanese firms is expected to be small. However, the sign of bonus variable is the same as U.S. studies, which is positive, since the managers of Japanese firms who have higher stakes will have more motivation to manage earnings.
Japanese firms, on average, have a higher ratio of depreciation expense compared to total fixed assets. The firms with higher depreciable asset book values have more opportunities to use changes in depreciation methods to manage earnings (Herrmann and Inoue, 1996). Thus, a depreciable assets variable is included in our model to control confounding effects. The sign of this variable is expected to be positive with the EM variable.
Operating Activity Deviations
Japanese firms focus on sales or growth rather than production as their main goal (Radebaugh and Gary, 1997). In our study, sales are used to measure deviation in operating activities since sales are less prone to manipulation. Managers may have motivations to manage earnings if actual operating results deviate from expected operating results. Therefore, the sign of the operating activity deviation variable is expected to be positive with the earnings management variable.
Moses (1987) finds evidence using U.S. data that there is a negative relationship between earnings variability and earnings management. However, the variability of earnings may measure market risk (Daley et al., 1988). Therefore, fluctuations of earnings may increase risk and increase borrowing costs. If this is the case, managers' earnings management motivations and earnings variability will have a positive relationship.
The debt-to-equity ratio variable has generally shown significant income increasing choices in U.S. studies. However, Darrough et al. (1998) show evidence that the debt-to-equity ratio is not significant using Japanese data. Interestingly, however, this result changes after the market crash of 1990 in Japan. Their results suggest that the debt and equity environment of Japan since 1990 is moving toward to that of U.S.. The general public does not own many shares in Japan compared to banks or other institutional investors (Black and White, 2003). Usually Japanese firms depend on debt financing rather than equity financing. Therefore, the debt-to-equity ratio is included in our model to control confounding effects. The expected sign of this variable is positive with the EM variable.
Data Collection and Descriptive Statistics
Our study used only March 31 fiscal year-end and non-financial firms from the PACAP database for Japan compiled by the Pacific-Basin Capital Market (PACAP) Research Center at the University of Rhode Island. We include only March 31 fiscal year end firms to ensure similar information environments. Approximately 2/3 of Japanese firms fall in this category. Banking firms are excluded due to a non-homogeneous business environment and firm characteristics that may lead to confounding effects. Table 1A presents basic summary statistics. A total of 11,866 sample firm data are available from the PACAP database for our study period. Overall, the EM variable is 1.4% of sales. The average size of our sample firms is 56.247 million Yen in sales. Tax payments are 2.1% of sales. Bonuses are 37.34% of operating income and depreciable assets are 23.24% of total assets. The absolute value of changes in sales divided by sales is 9.31% and the absolute value of the three year average earnings divided by sales is 1.6%. For Japanese firms, debts are three time higher than equity in our sample. Table 1B presents the correlation matrix. None of the variables have serious correlation problems based on Table 1B. EM and TAX has 26.78% and SIZE and VAR has -23.87% correlations.
Table 2 presents number of firms and total reserve account changes by year and by classifications of firms. Overall, the numbers of earnings management firms are slightly less than those of non-earnings management firms. In addition, profit and loss firms are also similar in numbers, but it changes year by year.
RESULTS AND ADDITIONAL TESTS
Table 3 summarizes Chi-Square test results of earnings management behavior based on differing operating conditions. The Chi-Square test is performed to test the equal frequency of operating conditions. The Chi-Square test result is significant ([chi square] = 200.12, P < .01). Therefore, this result shows that earnings management differs based on operating condition. Of the 11,866 sample firms, 2,931 (24.7%) report changes in reserve accounts and are classified with a profit and positive EBR. In this situation, managers are expected to choose income decreasing reserve account change methods to manage earnings (Herrmann and Inoue, 1996). Only 1,259 (43%) of 2,931 firms managed earnings by choosing income decreasing reserve account changes. A total of 3,732 firms (31.5%) report a profit and negative EBR and 5,203 firms (43.8%) report a loss during the test period. For firms with a profit and negative EBR, managers may choose income increasing reserve account changes to manage earnings. Of the 3,732 firms with profit and negative EBR, 2,178 (58.4%) managed earnings by income increasing reserve account changes. For firms reporting losses, managers may choose income increasing reserve account change methods to manage earnings. Of the 5,203 firms with losses, 2,364 (45.4%) managed earnings by income increasing reserve account changes. However, firms with a profit and negative EBR may have contradictory motivations for managers. Only firms with a minimum tax burden may increase income by changing reserve accounts.
Table 4 reports full sample results of regression analysis with a fixed group effect. (3) For the overall model, adjusted R2 = 24.56% and the model is significant (F = 2.40, P < .01). Firm size, income taxes, depreciable assets, deviation in operating activities, and debt to equity variables are significant. The bonus variable is not significant as expected since Japanese managers have less incentive for earnings manipulation for a bonus (Kaplan, 1994). However, firm size and depreciable assets variables have negative signs, rather than the positive signs that we anticipated from the results of previous U.S. studies. These are interesting outcomes. Usually larger firms have higher motivation to smooth income to reduce political costs in the U.S. business environment, but for the Japanese firms, smaller firms more actively manage earnings using changes in reserve accounts.
Smaller firms may have higher stakes to smooth income which could reverse the political cost motivation. For large Japanese firms that are usually backed by main banks or a keiretsu relationship, they feel less pressure for generating capital. However, smaller firms need to show a smooth stream of income in the capital market or to their major stakeholders, if they want to maintain their reputation or banks may exert pressure on them withdrawing their investments. In addition, the smaller firms may be under increased pressure to avoid delisting.
The depreciable assets variable, computed as the ratio of depreciable assets to total assets, is expected to have a positive sign. However, for this study, the amount of depreciable assets is far less than the amount of reserve accounts even though they are not highly correlated (r = -0.0268). This could be a possible reason for the negative sign of the depreciable assets variable. Earnings variability is not significant for the Japanese case. The debt-to-equity ratio has a negative sign and significant. Again this result differs from those of previous U.S. studies. Most Japanese firms have higher debt ratios than U.S. firms and if a firm can maintain those higher ratios, this may signal to the market that the firm is backed by main banks or other keiretsu firms.
Table 5 reports the results of regressing each reserve account against our independent variables. Discussion of each of the individual regressions is relative to the overall regression model. For the reserve for loan losses accounts, size and tax have the same sign as the overall model, and are statistically significant. However, the depreciable assets, operating activity deviation, and debt to equity variables are not significant compared with all the reserve account model in Table 4. Additionally, the sign is positive and significant for the reserve for loan loss model, while it was negative and insignificant for the overall model.
For the reserve for corporate taxes, size and debt to equity are not significant unlike the overall model, while the tax variable continues to be positive and significant. Unlike the overall model, bonus is negatively significant. Three variables continued to be significant, but changed their signs: depreciable assets, operating activity deviation, and earnings variability.
For the reserve for business taxes, size and debt to equity are not significant, while the depreciable assets variable continues to be positive. Unlike the overall model, bonus is negatively significant. Two variables continued to be significant, but changed their signs: taxes and operating activity deviation.
For the reserve for bonuses, there is no overlap on significant variables. Three variables continued to be significant, but changed their signs. Size changed from negatively significant to positively significant, while taxes and operating activity deviation changed from positively to negatively significant. Depreciable assets and debt to equity cease to be significant, while the bonus variable becomes positively significant.
For special reserves, the tax variable continues to be positive and significant. Both size and debt to equity change from significant with negative signs to significant with positive signs. Debt to equity now is also positively significant. However, depreciable assets and operating activity deviation cease to be significant.
For both the legal reserve and the capital reserve, the depreciable asset variable is not significant. The other independent variables have the same signs and significance as those in the overall model. This reserve account is the most consistent with the results of combined reserve accounts.
For the earned reserve, there is no overlap in the sign and significance of the independent variables. Five variables that were significant in the overall model changed signs. Size and depreciable assets are now positively significant. However, tax, operating activity deviation, and earnings variability are negatively significant. Earnings variance becomes negatively significant.
For the reversal of legal reserve, size, depreciable assets, and operating activity deviation cease to be significant. Two significant variables change signs: tax and debt to equity. However, tax and earnings variability, which was not significant in the overall model, are negatively significant. Debt to equity is marginally significant and positive.
For the provision for special reserve, earnings variability is negatively significant and debt to equity is marginally significant and positive. However, for this and reversal of legal reserve accounts, a lot of data are missing and their results are not stable.
Table 6 presents the results based on the size of earnings before reserve (EBR) accounts. (4) For small EBR firms, the tax and operating activity deviation variables are positive and significant, while the size, depreciable assets, and debt to equity variables are negative and significant. Each of these is consistent with the overall model results. For large EBR firms, the results are the same with one exception. Operating activity deviation ceases to be significant. This implies that operating activity deviation is not an important factor for larger EBR firms.
Table 7 reports regression results based on operating conditions. For profit and positive EBR firms, size is negative and significant and tax is positive and significant as in the overall model. Operating activity deviation and debt to equity are no longer significant. This means that for profit and positive EBR firms, operating conditions and debt to equity are not important factors for earnings management. However, earnings variability is now significant and positive impact on earnings management. Depreciable assets changed from negative and significant to positive and significant.
For profit and negative EBR firms, the signs of statistically significant variables are exactly the same as in the overall model. However, the results differ considerably from those of the profit and positive EBR firms. The depreciable assets variable is negatively significant, but operating activity deviation is positive and significant. This is an interesting result. This implies that for profit and positive EBR firms, depreciable assets have a positive impact, but for profit and negative EBR firms, depreciable assets have a negative impact on earnings management. However, the debt-to-equity ratio becomes significant for profit and negative EBR firms. This implies that for profit and negative EBR firms, higher debt-to-equity ratios may require more caution for earnings management. This might be a result of increased oversight by lenders as debt to equity ratios and risk rise.
For loss firms, the results are similar to those of the overall model and the profit and negative EBR firm model. However, the depreciable assets and operating activity deviation variables cease to be significant. For loss firms, these variables may not be important factors to motivate mangers to manage earnings.
Sensitivity Analysis by Time Period
Japan experienced its market crash in 1990 and its equity values dropped more than one-half during the bubble. Therefore, it is possible that there may be an economic structure shift after 1990. To compare the results before 1990 and after 1990, we have performed sensitivity analyses. We tested each variable by a different time period after eliminating the extreme top and bottom 5%. For Group 1, the time period is from 1983 to 1990 and for Group 2 the time period is from 1992 to 1999. 1983 is the first year companies were legally required to have reserve accounts in Japan. Ending analysis in 1999 results in the same number of years in the before and after market crash time periods. The transition year, 1991, is deleted. Table 8 reports the results of this comparison test.
From Table 8, there are some differences before and after Japan's recession. The size and tax variables are significant with the large top 25% of EBR firms. Both of these variables are significant before 1990, but not significant in the later period. The depreciable assets variable is significant for both periods, but the difference is not significant. Debt to equity is significant in the earlier period, but not significant in the later period for the top 25% of EBR firms. From the results, the size and tax variables are important factors for earnings management before the 1990 period, but not the later period.
For the bottom 25% of EBR firms, tax and deviation in operating activities are significantly different between the before and after market crash periods. For small EBR firms, however, the tax variable is a more important factor in the later period and the deviation in operating activities variable is only important in the later period. For the bottom 25% of EBR firms, the size variable is only significant before the 1990 period, but the deviation in operating activities variable is significant only after the after 1992 period. These results suggest that there may have been structural changes in the Japanese economy after 1990. For small EBR firms, the managers' motivations to manage earnings are more consistent across time periods, but for large EBR firms, managers' motivations to manage earnings reduced dramatically after the 1992 period. This implies that Japanese business environment is moving toward a more western style.
SUMMARY AND CONCLUSIONS
This study shows that the Japanese managers manage earnings by using changes in reserve accounts, but this differs significantly by the size of EBR as well as by reserve account. For profit and positive EBR firms, the size, tax, depreciable assets, and earnings variance variables are significant. For profit and negative EBR firms, however, the sign of depreciable assets is the opposite and deviation in operating activities and debt to equity variables are significant. This implies that profit and negative EBR firms are more cautious in managing earnings as the depreciable assets and debt to equity ratios increase, but more aggressive in managing earnings, if there is a deviation in operating activities. For loss firms, higher depreciable assets or deviations in operating activities do not result in earnings management. From the results of the full sample and individual reserve account regression analyses, the presence of bonus is not significant. Therefore, a bonus is not a motivator for Japanese managers. This result is consistent with previous studies. Earnings variability is not significant for the full sample model, but it may have an impact on individual reserve accounts. The sign of the size variable is negative for Japanese firms. This is an interesting result for the Japanese business environment. For Japanese firms, smaller firms manage earnings more aggressively to meet their target income, but larger firms may have other resources such as main bank support or keiretsu support which are unique in their environment. For the top 25% EBR firms, managers use only depreciable assets is only significant after the Japanese market crash. However, for the bottom 25% EBR firms, impact of the tax, depreciable assets, and debt to equity ratio variables are consistent before or after the market crash.
There are limitations in this paper. EM is measured using a single period. In addition, this paper only focuses on one EM alternative. However, this paper documented earnings management behavior of Japanese managers using total changes in reserve accounts and the change in each reserve account based on the size of EBR and compared the results before and after the market crash of 1990. The change in the reserve accounts may be caused by changes in the macroeconomic factors and management may be responding to these changes rather than managing earnings. Therefore, other macroeconomic factors may need to be controlled in future research. Further analyses of each reserve account and its unique financial environment to understand Japanese mangers' earnings management behavior may be fruitful avenues for the future research.
Ali, A. & L.S. Hwang (2000). Country-specific factors related to financial reporting and the value relevance of accounting data. Journal of Accounting Research, 38(1), 1-21.
Black, E.L. & J.J. White (2003). An international comparison of income statement and balance sheet information: Germany, Japan and the US. European Accounting Review, 12(1), 29-46.
Choi, F.D.S. & G.K. Meek (2005). International accounting (Fifth Edition). Upper Saddle River, NJ: Prentice-Hall.
Daley, L.A., D.W. Senkow & R.L. Vigeland (1988). Analysts' forecasts, earnings variability, and option pricing: empirical evidence. The Accounting Review, 63(4), 563-585.
Darrough, M.N., H. Pourjalali & S. Saudagran (1988). Earnings management in Japanese companies. The International Journal of Accounting, 33(3), 313-334.
DeAngelo, H., L. DeAngelo & D.J. Skinner (1994). Accounting choice in troubled companies. Journal of Accounting and Economics, 17, 113-144.
Dechow, P.M. & R.G. Sloan (1991). Executive incentives and the horizon problem: An empirical investigation. Journal of Accounting and Economics, 14, 51-89.
Evans III, J.H. & S.S. Sridhar (1996). Multiple control systems, accrual accounting, and earnings management. Journal of Accounting Research, 34(1), 45-65.
Hausman, J.A. (1978). Specification test in econometrics. Econometrica, 46(6), 1251-1271.
Herrmann D. & T. Inoue (1996). Income smoothing and incentives by operating condition: An empirical test using depreciation changes in Japan. Journal of International Accounting, Auditing & Taxation, 5(2), 161-177.
Herrmann D., T. Inoue & W. Thomas (2003). The sales of assets to manage earnings in Japan. Journal of Accounting Research, 41(1), 89-108.
http://news.chosun.com, February 27, 2007.
Jones, J. (1991). Earnings management during import relief investigations. Journal of Accounting Research, 29, 193-228.
Kaplan, S. N. (1994). Top executive rewards and firm performance: A comparison of Japan and the United States. Journal of Political Economy, 102(3), 510-546.
Kasznik R. (1999). On the association between voluntary disclosure and earnings management. Journal of Accounting Research, 37(1), 57-81.
Kumar, S. & K. Hyodo (2001). Price-earnings ratios in Japan: Recent evidence and further results. Journal of International Financial Management and Accounting, 12(1), 24-49.
Maydew E. L. (1997). Tax-Induced earnings management by firms with net operating losses. Journal of Accounting Research, 35(1), 83-96.
Moses, O.D. (1987). Income smoothing and incentives: Empirical tests using accounting changes. The Accounting Review, LXII(2), 358-377.
Radebaugh, L.H. & S. Gary (1997). International accounting and multinational enterprises (Fourth Edition). New York: John Wiley.
Takahashi, K., Y. Kurokawa & K. Watase (1984). Corporate bankruptcy prediction in Japan. Journal of Banking and Finance, 8, 229-247.
Watts, R. & J. Zimmerman (1986). Positive accounting theory. New Jersey: Prentice Hall.
Wikil Kwak, University of Nebraska at Omaha
Richard File, University of Nebraska at Omaha
(1.) Reserve accounts used for this study are reserve for loan losses, reserve for corporate taxes, reserve for business taxes, reserve for bonuses, special reserves, legal reserve, capital reserve, earned reserve, reversal of legal reserve, and provision for special reserve which are identified in the PACAP database. These accounts have existed since 1982. This study uses the period from this date to the market crash in 1990 as one period. We then use a comparable length period after the crash for sensitivity analysis by time periods.
(2.) For the purpose of this paper, earnings management is equivalent to income smoothing even though income smoothing is a form of earnings management.
(3.) Data estimation, intrinsically, allows for heterogeneity across panel units and across time. This means that earnings management behaviors might be different across time and firms in the model. The fixed effect model that we used in this study confines that heterogeneity to the intercept terms of the relationship. However, that heterogeneity is absorbed in the error structure in the random effects model. The Hausman (1978) specification test is used to compare the fixed model with the random effect model under the null hypothesis that individual effects are uncorrelated with other regressors in the model. A fixed effect model is preferred if effects are correlated.
(4.) Each reserve account was used to compute the dependent variable instead of aggregate reserve accounts and the test was rerun using the same sets of independent variables and the results are reported in Table 5.
Table 1A : Summary of Basic Statistics Variable N Mean Standard Minimum Maximum Deviation EM 11866 0.0144 0.0396 -0.0434 0.3069 SIZE 11866 10.7501 1.0746 8.1167 13.632 TAX 11866 0.0206 0.0205 -0.0093 0.1165 BONUS 11866 0.3734 0.6668 -2.5816 6.3709 DASSET 11866 0.2324 0.1188 0.0131 0.6014 DEV 11866 0.0931 0.0807 0.0013 0.4779 VAR 11866 0.0161 0.0267 0.0003 0.2504 DBT/EQT 11866 3.3328 3.886 0.1164 34.3934 Table 1B: Correlation Matrix EM SIZE TAX BONUS EM 1 SIZE -0.0475 1 TAX 0.2678 -0.0918 1 BONUS -0.0564 0.0041 -0.1443 1 DASSET -0.0268 -0.1548 -0.0508 0.0393 DEV 0.1268 -0.0994 0.0832 -0.0976 VAR -0.0006 -0.2387 -0.0792 -0.0597 DBT/EQT -0.126 0.0894 -0.3629 -0.0319 DASSET DEV VAR DBT2EQT EM SIZE TAX BONUS DASSET 1 DEV -0.1072 1 VAR 0.0798 0.1617 1 DBT/EQT -0.177 0.0512 0.0446 1 Variable definitions: Dependent variable: EM = |[EBR.sub.t] - [EBR.sub.t-1]| - |[E.sub.t] - [E.sub.t-1]|/SALES EM = earnings management EBR = earnings before reserve accounts in period t and t-1 E = earnings in period t and t-1. Independent variables: SIZE = log of sales TAX = tax payment/sales BONUS = bonus/operating income DASSET = depreciable assets/total assets DEV = |[Sales.sub.t] - [Sales.sub.t-1]|/[Sales.sub.t-1] VAR = 1/3 [3.summation over (t=1)]| [E.sub.t] - [E.sub.t-1]/[Sales.sub.t]| DBT/EQT = debt-to-equity ratio Table 2:Number of Observations in Reserve Accounts by Year Year Total Total Income Non Profit Loss Number Reserve Smoothing Income of Accounts smoothing Firms Changes 1976 368 367 216 152 238 130 1977 407 406 198 209 235 172 1978 426 425 185 241 335 91 1979 395 394 193 202 294 101 1980 382 382 216 166 206 176 1981 380 380 172 208 184 196 1982 372 371 180 192 169 203 1983 345 343 168 177 217 128 1984 349 349 185 164 261 88 1985 364 363 164 200 185 179 1986 377 377 175 202 199 178 1987 385 385 218 167 319 66 1988 426 426 221 205 341 85 1989 491 491 289 202 363 128 1990 578 578 178 400 369 209 1991 611 610 256 355 221 390 1992 646 646 298 348 186 460 1993 667 665 343 324 294 373 1994 680 679 366 314 401 279 1995 682 680 317 365 435 247 1996 700 698 369 331 413 287 1997 683 678 280 403 254 429 1998 612 611 303 309 193 419 1999 540 537 311 229 351 189 Total 11866 11841 5801 6065 6663 5203 Table 3: Chi-Square Test Results of Smoothing Behavior by Operating Conditions Profit and Profit and Loss Total Positive Negative EBR EBR Earnings Management 1259 2178 2364 5801 Non-Earnings Management 1672 1554 2839 6065 Total 2931 3732 5203 11866 Computed Chi-Square statistic = 200.12 Critical Chi-Square Value with degree of freedom 2 at 1% = 9.21 Table 4: Full Sample Regression Results with Fixed Group Effect and Time Effect Variable Coefficient t-statistic Intercept 0.12816 5.64 *** SIZE -0.01013 -5.90 *** TAX 0.60669 20.03 *** BONUS -0.00029 -0.52 DASSET -0.01675 -2.24 ** DEV 0.02797 5.60 *** VAR -0.0059 -0.36 DBT/EQT -0.00113 -6.99 *** Adj [R.sup.2] = 0.2456 F test = 2.40 *** Hausman test: 80.82 *** Number of Observation: 11866 Note: F test rejects the null hypothesis that there are not any fixed effects. Hausman test for random effects shows a fixed effect model is preferred to a random effect model. *** Significant at the 1% level. ** Significant at the 5% level. * Significant at the 10% level. Variable definitions: Dependent variable: EM = |[EBR.sub.t] - [EBR.sub.t-1]| - |[E.sub.t] - [E.sub.t-1]|/SALES EM = earnings management EBR = earnings before reserve accounts in period t and t-1 E = earnings in period t and t-1. Independent variables: SIZE = log of sales TAX = tax payment/sales BONUS = bonus/operating income DASSET = depreciable assets/total assets DEV = |[Sales.sub.t] - [Sales.sub.t-1]|/[Sales.sub.t-1] VAR = 1/3 [3.summation over (t=1)]| [E.sub.t] - [E.sub.t-1]/[Sales.sub.t]| DBT/EQT = debt-to-equity ratio Table 5: Full Sample Regression Results by Each Reserve Account Dependent EMTO EM29 EM43 Intercept 0.128 0.008 -0.004 5.64 *** 1.22 -0.72 SIZE -0.01 -0.001 0.000 -5.90 *** -2.64 ** 0.46 TAX 0.607 0.024 0.025 20.03 *** 3.46 *** 3.53 *** BONUS 0.000 0.000 0.000 -0.52 0.50 -2.61 ** DASSET -0.017 0.000 0.003 -2.24 ** 0.15 1.82 * DEV 0.028 -0.001 -0.018 5.60 *** -0.97 -15.32 *** VAR -0.006 0.021 -0.011 -0.36 5.25 *** -2.62 ** DBT/EQT -0.001 0.000 0.000 -6.99 *** -1.33 0.000 Adj R2 0.2456 0.1331 0.1691 No of Obs 11866 9690 10702 Dependent EM55 EM58 EM59 Intercept -0.002 0.045 0.041 -1.49 4.32 *** 4.02 *** SIZE 0 -0.004 -0.003 1.69 * -4.43 *** -3.93 *** TAX 0.004 0.137 0.145 2.31 ** 9.82 *** 10.75 *** BONUS 0.000 0.000 0.000 -0.94 -0.93 -1.24 DASSET 0.000 0.001 0.000 0.24 0.35 -0.12 DEV 0.001 0.007 0.006 1.85 3.00 *** 2.85 *** VAR 0.001 -0.011 -0.011 0.93 -1.45 -1.44 DBT/EQT 0.000 -0.001 -0.001 2.14 ** -7.16 *** -7.57 *** Adj R2 0.1335 0.1838 0.1898 No of Obs 10075 11866 11711 Dependent EM44 EM45 Intercept 0.003 -0.004 1.11 -3.00 *** SIZE 0.000 0.000 -0.33 3.95 *** TAX -0.014 -0.007 -5.09 *** -4.14 *** BONUS 0.000 0.000 -2.57 ** 4.86 *** DASSET -0.002 0.001 -3.03 *** 1.67 DEV -0.005 -0.005 -11.67 *** -18.08 *** VAR 0.001 0.001 0.62 1.14 DBT/EQT 0.000 0.000 0.35 -0.39 Adj R2 0.1592 0.1801 No of Obs 8983 11778 Dependent EM60 EM70 EM71 Intercept -0.002 0.003 0.000 -2.37 ** 0.32 -0.08 SIZE 0.000 0.000 0.000 2.74 *** -0.17 -0.07 TAX -0.021 -0.032 0.013 -19.63 *** -2.89 *** 1.82 BONUS 0.000 0.000 0.000 1.21 -1.43 -0.21 DASSET 0.001 -0.001 0.002 3.77*** -0.27 1.44 DEV -0.001 0.001 -0.001 -8.00 *** 1.15 -1.07 VAR -0.001 -0.029 -0.009 -2.43 ** -6.11 *** -2.13 ** DBT/EQT 0 0 0 -2.98 *** 1.88 * 1.69 * Adj R2 0.1805 0.3383 0.0161 No of Obs 11665 1654 733 * Variable definitions are the same as in Table 4. EM29: reserve for loan losses, EM43: reserve for corporate taxes, EM44: reserve for business taxes, EM45: reserve for bonuses, EM55: special reserves, EM58: legal reserve, EM59: capital reserve, EM60: earned reserve, EM70: reversal of legal reserve, and EM71: provision for special reserve. Table 6: Full Sample Regression Results by EBR Variable Coefficient t-statistic EBR Large--Above Median (N=5881): fixed group and time effect considered Intercept 0.06294 2.19 ** SIZE -0.00393 -1.69 * TAX 0.54078 13.36 *** BONUS -0.00028 -0.45 DASSET -0.01958 -2.19 ** DEV 0.00644 1.13 VAR -0.02829 -1.62 DBT/EQT -0.00061 -3.94 *** Adj [R.sup.2]= 0.2583 F test= 1.83 *** Hausman test= 38.07 *** EBR Small--Below Median (N=5901)): fixed group and time effect considered Intercept 0.25904 6.29 *** SIZE -0.02128 -6.48 *** TAX 0.60325 13.00 *** BONUS 0.00011 0.12 DASSET -0.0423 -3.00 *** DEV 0.04262 5.11 *** VAR 0.00893 0.29 DBT/EQT -0.00342 -6.33 *** Adj [R.sup.2] = 0.3178 F test= 2.93 *** Hausman test= 70.02 *** *** Significant at the 1% level. ** Significant at the 5% level. * Significant at the 10% level. * Variable definitions are the same as in Table 4. Table 7: Regression results by operating conditions Variable Coefficient t-statistic Profit and Positive EBR (n=2772)): fixed group and time effect considered Intercept 0.05190 2.22 ** SIZE -0.00566 -2.94 *** TAX 0.08412 2.52 ** BONUS -0.00029 -0.45 DASSET 0.01663 2.12 ** DEV -0.00085 -0.14 VAR 0.04892 2.91 *** DBT/EQT 0.00001 0.07 Adj [R.sup.2] = 0.3386 F test= 1.43 *** Hausman test= 27.13 *** Profit and Negative EBR (n=3589)): fixed group and time effect considered Intercept 0.15896 2.83 *** SIZE -0.00897 -1.91 * TAX 0.29990 3.36 *** BONUS -0.00536 -1.49 DASSET -0.04302 -2.03 ** DEV 0.03203 2.57 ** VAR 0.00131 0.02 DBT/EQT -0.00777 -11.48 *** Adj [R.sup.2] = 0.3959 F test= 1.82 *** Hausman test= 83.47 *** Loss Companies (n=5123)): fixed group and time effect considered Intercept 0.09044 3.36*** SIZE -0.00782 -3.70*** TAX 0.65921 16.12*** BONUS 0.00024 0.45 DASSET 0.00133 0.15 DEV 0.00835 1.24 VAR -0.00647 -0.34 DBT/EQT -0.00056 -2.98*** Adj [R.sub.2] = 0.2858 F test= 1.70 *** Hausman test= 76.93 *** *** Significant at the 1% level. ** Significant at the 5% level. * Significant at the 10% level. * Variable definitions are the same as in Table 4. Table 8: Regression results of EBR by time periods after deleting the extreme 5% EBR Group 1: 1983 - 1990 H0: [[beta]. sup.Group1] = [[beta].sup. Group2] Variable Coefficient t-statistic t-statistic EBR Large--Top 25% Intercept 0.1153 2.49 ** 2.88* SIZE -0.01067 -2.06 ** 2.63** TAX 0.36332 4.72 *** 7.69*** BONUS 0.00024 0.23 1.17 DASSET -0.04924 -2.42 ** 1.64 DEV -0.00721 -0.7 0.73 VAR -0.0194 -0.48 0.59 DBT/EQT -0.00063 -2.50 *** 1.03 Adj [R.sup.2] 0.3709 0.3201 N 829 675 EBR Small--Bottom 25% Intercept 0.67679 3.14 *** 2.15** SIZE -0.04906 -2.63 *** 1.21 TAX 0.51766 2.57 ** 5.37*** BONUS 0.00529 1.44 0.96 DASSET -0.20460 -2.62 *** 0.63 DEV 0.02949 1.00 2.01** VAR -0.06259 -0.32 0.16 DBT/EQT -0.00407 -1.79 * 0.34 Group 2: 1992 - 1999 H0: [[beta]. sup.Group1] = [[beta].sup. Group2] Variable Coefficient t-statistic t-statistic EBR Large--Top 25% Intercept 0.05304 0.8 2.88* SIZE -0.00492 -0.74 2.63** TAX 0.04753 0.47 7.69*** BONUS 0.0007 0.65 1.17 DASSET 0.04502 1.87 ** 1.64 DEV 0.00461 0.33 0.73 VAR -0.00938 -0.24 0.59 DBT/EQT -0.00046 -1.3 1.03 Adj [R.sup.2] 0.3709 0.3201 N 829 675 EBR Small--Bottom 25% Intercept 0.14134 1.66 * 2.15** SIZE -0.00933 -1.49 1.21 TAX 0.41482 5.49 *** 5.37*** BONUS -0.00078 -0.83 0.96 DASSET -0.04851 -2.04 ** 0.63 DEV 0.02749 2.31 ** 2.01** VAR -0.00913 -0.24 0.16 DBT/EQT -0.00135 -1.66 * 0.34 Adj. [R.sup.2] 0.5204 0.2952 N 704 1785 *** Significant at the 1% level. ** Significant at the 5% level. * Significant at the 10% level. * Variable definitions are the same as in Table 4.
|Printer friendly Cite/link Email Feedback|
|Author:||Kwak, Wikil; File, Richard|
|Publication:||Journal of International Business Research|
|Date:||Jan 1, 2009|
|Previous Article:||Cultural differences in television celebrity use in the United States and Lebanon.|
|Next Article:||Individualism and collectivism: reconsidering old assumptions.|