SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 The Effect of Market Share and Leverage Interaction Toward Earnings Management Practices By Dian Anita Nuswantara State University of Surabaya ABSTRACT The purpose of the study is to investigate whether market share, institutional ownership and leverage affect the existence of earnings management and whether the interaction between market share and leverage will affect the existence of earnings management. To conduct this research we use regression and regression with interaction method as statistical tools. Total observation are 81 they are 27 companies that listed on Bursa Efek Jakarta (BEJ) by 1998 to 2000. This study supports the first and fourth hypothesis but could not accept the second and third hypothesis. The results show the evidence that market share effect the existence of earnings management and the interaction between market share and leverage also effect the earnings management. Keywords: earnings management, market share, leverage, and institutional ownership 1. INTRODUCTION Earnings is a measurement of the company performance summary that is accomplished under accrual-based accounting. It becomes important because it is also considered by the users as the measurement of the company performance summary, for instance it determines the executives compensation and investor or creditor decision making. Accrual background is underlain the problem solving of the company performance measure when the company is in the continued operation. Income recognition and matching principles are expected to be able to overcome the time and verification problems attached in the treasury flow so that the earnings can reflect the company performance (Dechow, 1994). However, the use of accrual is still problematic, due to the discretion management that is related. A management authority can be used to inform the private matters or to manipulate earnings for personal benefit. Many earnings management researches are carried on, and it is find out that many companies employs the earnings management for several purposes; like, for example: it is used to fill in the expectation of analysts (Burghstahler and Eames, 1998), to fill in the investor expectation (Bushee, 1998), to fill in the company budget (Kasznik, 1999), to fill in the debt binding (Healy and Palepu, 1990) and others. Numbers resulted from accounting process are used by regulator in making a decision, but there is no research showing that regulators consider earnings management practice in their political decision (Healy and Wahlen, 1999). This condition triggers companies that are sensitive toward the political decision to employ earnings management practices (Jones, 1991; Cahan, 1992; Key, 1997). Earnings management practices is also occurred in Indonesia, for example Saiful and Hartono (2002) find out that the management withdraw the future earnings to raise the current earnings around their first offering to public. This is aimed to maximize their own utility. Investors and other market participants cannot detect this condition, it might be due to the developing market characteristic that is its 170 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 participants are less capable. Some researches also show that corporate governance mechanism can be counted on as one of the factors which can control the opportunistic manager’s action (Ching et al., 2002). Shleifer and Vishny (1997) cited that corporate governance is related to the way how an investor feels sure that they will obtain their investment back, how an investor push a management to give the half of their earnings to them, how they can feel sure that the management will not embassle the money they supplied for or how the management does not invest in bad project, how they control the company. External control is one of the instruments to reach the corporate governance. Therefore, this research is expected to be able to test the effect of leverage, institution owner and market authority toward earnings management practice and interaction influences between leverage and market authority toward earnings management practice. Based on the above explanation, this article attempts to test if the market authority of a company influences positively toward a company earnings management, if degree of leverage influences negatively toward a company earnings management, and if the relationship between market authority and earnings management practice are influenced by the degree of leverage. The organization of the article is as the following: part one is the introduction containing the background, problems, and objective of the study, part two contains the review to related literature that presents some theories on earnings management, political costs, agency relationship, and corporate governance, then continued to hypothesis formulation; part three describes the research method; part four presents the data analysis and findings; and the last is conclusion and findings of hypothesis test, and the last is conclusion and the limitation of the research. 2. THEORETICAL FRAMEWORKS AND HYPOTHESIS FORMULATION Many studies on earnings management had been carried on different background. McNichols and Wilson (1988) and Perry and Wiliams (1994) tested several measurement for earnings management in companies practice which will only produce the description of the company pseudo performance, therefore the investors who make use of the information to make a decision, they will buy the company shares which have counterfit shares. Jones (1991) investigated if companies manage earnings during the investigation of import liberation, and this study showed that a company would decrease its earnings when he need an import protection through tariffs, quotas, and marketing agreements. Cahan (1992) and Key (1997)also support Jones study. Keating and Zimmerman (2000) support hypothesis that company manage its earnings to avoid violation of debt covenant, to increase executives compensation, and/or to reduce the possibility of executive change. The study show that manager who choose to change its accounting policy that increase earnings have bad performance and high degree of leverage. Ching et al (2002) try to relating earnings management with corporate governance issues. Regulator is in charge to identify possible situations to encourage earnings management. Therefore, the regulator and investor need to consider corporate governance in making a decision (Ching et al., 2002) because the possibility of earnings management is connected to a specific environment where the company runs its operations. A study done by Ching in Hongkong agrees with the idea that earnings management which is around seasoned equity offerings depends on the structure of corporate governance. A company which employs seasoned equity 171 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 offerings has a greater motivation to manage its earnings when they have greater number of board of directors in compare with those who have smaller. This indicates that the smaller number of board of directors will give a better control function than the greater number. 2.1. Political Cost Hypothesis Positive theory is a theory that used to predict what happen in this real life and what the consequences should be taken into account. Positive accounting theory states that company management is carried out in the most efficient way so that the company existence can be maintained maximally. The freedom to choose the accounting policies enable the manager to be opportunist one. Positive accounting theory assumes that both the manager and investors are rational and able to do whatever they want to do on their own interest. Although, there is a mechanism used to anticipate the opportunist manager, like for example: through determining the remuneration plan. Political costs can be proxied by several measurements like company size, risk, capital intensity, and industrial concentration ratio (Watts & Zimmerman, 1986). Political costs hypothesis predicts that manager will refuse the possibility of political impact of welfare transfer by choosing the accounting strategy that decrease earnings (Watts and Zimmerman, 1978 in Watts and Zimmerman, 1986). Zmijewski and Hagerman (1981) suspect that political costs are in variation with firms risks and that high risks company tend to choose accounting procedure portfolio that decrease earnings. The motivation is high risks company tend to have high variance of earnings changes so that the company tends to report high earnings and because information costs, boards, politicians, and bureaucrats may not adjust risk when considering profit reported. Manager of company whom subjects to political costs tends to take investment with lower risks to avoid high profit (Peltzman, 1976 in Watts and Zimmerman, 1986) Zmijewski and Hagerman (1981) argue that costs information, boards, politicians, and bureaucrats do not adjust earnings reported for capital opportunity costs therefore capital intensive companies tend to be subject of political costs so that they will choose accounting procedure that decrease earnings. Anti–trust division uses industrial concentration ratio to justify the competency level in an industry. One of the factors that is found out significantly influences the accounting procedure selection is the industrial concentration ratio (Utama 2000). This ratio is the percentage of the total sales in an industry which is attained by several big companies in industry. Industry with concentration ratio which is high will be curricul on laws, like anti–trust policy, to avoid the law implementation, the company with high industrial concentration tend to choose the accounting policy which decreases the comings. Industrial concentration ratio is reversed with the competition level. The closer the ratio to 1 means the less the competitors. In live with this point of view Shleifer & Vishny (1997) state that product market competition will decrease company profitability. If a company is not efficient so it will decrease the earnings. Therefore, a manager of a company who owns low profitability will manipulate the earnings so that an investor will be willing to trust their capital in the company. If the company 172 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 market segment is small so that the company does not have strong position in the competition, this triggers the company to manipulate the earning so it look good. Contrary to the above fact, there are many advantage for the companies which have large market segment, one of them is that the company become a price– determiner – it is caused by the power of big companies that enable to compete with other companies without hesitation (Brinkley, 1999). This makes the company profitability higher than companies which have small market segment. The excess of the advantage, the companies become sensitive toward every government policy. This is caused by the government hesitation that the company will commit illegal monopoly, transfer of wealth, subsidy abandoning, and government investigation. Therefore, the companies tend to admit share management. Based on that illustration, the hypothesis are formulate as follows. Hypothesis 1: The power of market reflected by company market share positively influences earnings management practice. 2.3. Agency Theory Agency relationships defined as a contract taken by principal party and agent, in which in agent work on the behalf of principal name and given to an authority to make a decision. If both parties are the ultimate utility so there are reasons for the agent to rely on not only to do thing for the sake of the principal advantages. Principal can limit the distortion by determining the right incentive for the agent and provide supervision cost that is designed for watching the agent’s work that might distort the rule. One of the action of manager’s distortion is earnings management. The existence of this management is probably the function from specific environment where the company work. For instance, the corporate governance mechanism which can influence the earnings management level and how important the factors influence the earnings management. These are influenced by domestic jurisdiction factor of a country (Ching et al, 2002). Corporate governance can be defined as system that adopted by company in the mean of conducting and controlling the company (Cadbury, 1992 in Ching et al, 2002). Corporate governance focus on process for which designed to assure the accountability of the company directors and manager jobs. They must perform their job in high integrity and must be a subject to check and balances to avoid using power inappropriately. External supervision is one of the important components from corporate governance mechanism, and ownership of an institution and loan – use are another from of external supervision. 2.4. The Institution Ownerships The institution ownership is important for there is an interest conflict between a manager & shareholders (Jensen & Meckling, 1976). It is because of the institutional investor who has information more than the individual investor, consequently the announcement for company which has great institutional ownerships contains less information (Szewczyk, 1992 in Filbeck and Hatfield, 1999). This is caused by: 1. Institutional investor has more resources than individual investor to allocate information. 173 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 2. Beside that, institutional investor has professionalism in information analysis so that information quality can be detected. 3. Institutional Investor, generally, has business relation so that an information access is large them individual investor. 4. Institutional investor has motivation for monitoring more tightly the company activities them individual investor. 5. Institutional investor’s trade is more active than individual investor so it can raise the probability for new information instantly that is reflected from the price. Gillian and Starks (2000) state that institutional ownership is useful for it can push the manager to think for the company long-term performance. However, not all institutional investor give an advantage for management, for instance, pension insurance, because this cannot give an in put for the management (Murphy and Vans Nuys, 1994 in Wilopo and Mayangsari, 2002). Lins (2002) finds out that a separation between owner & management influences positively forward the company values forward countries which have law protection for share holder. In other side, concentration influence of ownership in the company external party has positive influence for company values. Filbeck and Hatfield (1999) find out that there is a direct relationship between level of institutional ownership and member of respond of share price. They hypothesized that this relationship is not significant for the company public utility as for as the regulator role takes place of the institutional investor role in decreasing the assymetrics information on the contrary, if the ownership is concentrated on small number of investor with high percentage of ownership for every investor, the collective action of the investor will be easier, for instance through voting (Shleifer & Vishny, 1997). Although, the distortion forward voting right mostly happened in developing countries which have unstable law system. Therefore, it is hypothesized: Hypothesis 2: The institutional ownership influences negatively toward earnings management 2.5. The Leverage Debt could reflects the ability of creditor to have a control beyond his debtor. Specifically, we can say that debt is a contract where the debtor raise funds from the creditor and promise to give a flow of payment predetermined to the creditor. Besides, the debtor usually make a statement that he won’t break the covenant (Smith and Warner, 1979). If the debtor broke the covenant, especially when he cannot fulfill the payment, the creditor may take over the collateral or the possibility of bankruptcy. Those above show that company will try to obey the covenants in the contracts to avoid the creditors’ use of his right. Those rights, at the end, are one of substance of external monitoring. Thus, we hypothesized: Hypothesis 3: the leverage influences negatively toward earnings management As discussed above, the company which have market power is more possible to conduct earnings management. But the magnitude of those opportunistic behavior will be declined when there is sufficient corporate governance mechanism. Therefore, the degree of leverage that reflects external monitoring, that is creditor, is hypothesized 174 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 could be influence the relationship of company market power with its possibility to practice earnings management. Thus, we hypothesized: Hypothesis 4: The bigger the market power, the greater the possibility of earnings management practice, especially if there is less external monitoring (low leverage). 2.6. The Size of Company The size of company could be reflects by the magnitude of company’s assets. Big company has a higher exposure than the smaller company so that it will be an object of high quality auditor investigation. This condition push the reporting become more conservative and tend to manipulate earnings downward. 3. RESEARCH METHODOLOGY 3.1. The Sample The population is public company that listed in Jakarta Stock Exchange, except those are in financial sector because they are high regulated, high risks, and public entrust sector (Samlawi and Sudibyo, 2000). The sample is conducted with purposive sampling to obtain sample which represents the criteria. The sample are those have: 1. three years financial reporting from 1998-2000, and the years must be ended at 31 December 2. not in banking and financial industry as classification by Indonesian Capital Market Directory 3. have data needed; those are discretionary accruals, market share, and leverage 4. has in accordance with classification of Biro Pusat Statistik 1998 5. sales can be group based on industry of parents DATA COLLECTION METHODS Data used in this research are: 1. Secondary data, annual financial statements with the notes to financial statements from Jakarta Stock Exchange 1998 to 2000 to get discretionary accruals, institutional ownership, and leverage 2. Market share taken from Biro Pusat Statistik 1998-2000 to get market share. 3.3.VARIABLE MEASUREMENT 3.3.1. DEPENDENT VARIABLE Dependent variable used in this research is discretionary accruals counted by modified Jones model, because this measurement less bias (Dechow, 1995). The procedure to count are as follows: First, determine the total accruals by the formulas: TAt=(CAt-CLt-Casht+STDt-Dept)/(At-1) Where, TAt is total accruals firm t at time t CAt is the change in liquid assets at time t CLt is the change in short-term debt at time t Casht is the change in cash and equivalent cash at time t STDt is the change in long-term debt mature in current year at time t Dept is depreciation and amortization costs at time t At-1 is total assets at time t-1 175 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 Second, regress this formula to get coefficient for nondiscretionary accruals TAt= a1(1/At-1)+ a2(REVt)+ a3 (PPEt)+ t Where, REVt is the change in revenue in year t divided by assets in year t-1 PPEt is gross assets in year t divided by assets in year t-1. Third, determine nondiscretionary accruals by substitute a1, a2, a3 into formulas below: NDAt=1(1/At-1)+ 2(REVt)+3(PPEt) Where, NDAt is nondiscretionary accruals at time t 3.3.2. Independent Variable Independent variable used in this research are: 1. Market share is the firm sales in year t divided by industry sales in the same year 2. leverage is debt divided by assets 3. institutional ownership is proportion of shares owned by institutional investor, whose institutional investor defined as Jones, 1991. 3.3.3. Moderating Variable Moderating variable is leverage with the same explanation as above 3.3.4. Control Variable Control variable is assets which measure as natural log of company assets. This measurement used to minimize variance between sample Analysis Techniques To test hypothesis 1, 2, and 3 we use multiple regression as follows: ABSDACC = (MARKET SHARE, LEVERAGE, INSTITUTIONAL OWNERSHIP) ABSDACC is the absolute of company’s discretionary accruals. The company’s discretionary accruals must be absolute first because we do not consider whether the earnings management are earnings increasing or earnings decreasing but we regard to the degree of company’s discretionary accruals. To test hypothesis 4 we use multiple regression with leverage as moderating variable as follows: ABSDACC = (MARKET SHARE, LEVERAGE, INSTITUTIONAL OWNERSHIP, MARKET SHARE*LEVERAGE) As discussed above, leverage is argued as proxy of external monitoring thus market power will be multiplied by leverage. Before testing the hypothesis we have to fulfill classic assumptions. To test the heteroscedasticity assumption we conduct Glejser test. To test the multicolinearity we see the variance inflation factor and tolerance. To see whether there is an autocorrelation we use Durbin-Watson d statistic. 176 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 4. EMPIRICAL RESULTS 4.1. Descriptive Statistics Table 2 is the summary of descriptive statistics on the research sample Table 2 Descriptive Statistics of The Research Sample N Minimum Maximum Mean Std. Deviation DISCRETIONARY 81 .01 23.04 .4868 2.5419 ACCRUALS ASSETS 81 34359.00 129684726.00 5207570.6543 19499718.8955 MARKET SHARE 81 .06 41.28 3.3924 7.7126 INSTITUTIONAL 81 .11 .95 .6458 .2143 OWNERSHIP LEVERAGE 81 .10 1.84 .6825 .3517 Valid N (listwise) 81 The summary of descriptive statistics shows that the dependent variable, discretionary accruals, has minimum result 0.01 with maximum result 23.04, mean value 0.4868, and 2.5419 standard deviation. Assets have minimum result 10.44 with maximum result 18.68, mean value 13.2466, and 1.7856 standard deviation. Market share variable has the highest variability compared to other variable, it can be seen from its 7,7126 standard deviation with the minimum result 0.06 and the maximum result 41,28. Institution ownership has minimum result 0.11 with maximum result 0.95, mean value 0.6458, and 0.2143 standard deviation. Degree of Leverage has minimum result 0.10 with maximum result 1.84, mean value 0.6825, and 0.3517 standard deviation. 4.2. Classical Assumption Test Results Classical assumption test on the relationship between dependent variable and independent variable is important because of the two characteristics in multivariate analysis (Hair, et al, 1998). First, relationship complexity, the more variables used into the equation, the larger the distortion and bias that might have been happened if the assumption has been ignored. Second, analysis and result complexity might covered the sign of assumption ignorance that has shown in a more simple analysis, univariate analysis. Homoscedasticity This assumption suggest that based on the independent variable value, the variation in the error term is equal to all observation thus, it can be said that the variation in the error term for each independent variable is the same constant positive value with its standard deviation (Hair et al, 1998). In short, it be said that Y population connected with several X value will have the same variation. Variation around the regression line (upper and lower) wil be the same for every X value. This assumption is connected especially on the relationship between variables. This condition suggest that dependent variable shows the same variation degree along predictor variable continuum. Glejser test results is shown in the following table: 177 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 Table 3 The Result of Residual Variable Regression on Weighted Independent Variables (Assets, Market Share, Institution Ownership, and leverage) Unstandardized Coefficients B (Constant) -.276 LNASSETS -2.544E-04 MARKET SHARE .230 INSTITUTIONAL .365 OWNERSHIP LEVERAGE -7.020E-02 a Dependent Variable: RESIDUAL Std. Error .874 .063 .015 .495 .302 Standardized Coefficients Beta t p-value .000 .890 .039 -.315 -.004 15.703 .738 .753 .997 .000 .463 -.012 -.232 .817 Table 3 shows that there is independent variable which is significant with its residual, that is market share variable. This problem is overcome by weighted least squares method (Gujarati, 1995). The test result after weighing with the residual from original regression is presented on table 4. Table 4 shows that the significant of independent variable on its residual is no longer exist. Thus, it can be concluded that heteroscedasticity is not happening. Table 4 The Result of Residual Variable Regression on Weighted Independent Variables (Assets, Market Share, Institution Ownership, and leverage) Unstandardized Coefficients B (Constant) .308 LNASSETS -2.098E-03 MARKET SHARE -3.943E-04 INSTITUTIONAL -.149 OWNERSHIP LEVERAGE .171 a Dependent Variable: RESIDUAL Std. Error .402 .030 .008 .242 .149 Standardized Coefficients Beta t p-value -.012 -.008 -.104 .767 -.071 -.048 -.615 .448 .944 .962 .542 .188 1.145 .259 178 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 Table 5 The Result of Residual Variable Regression on Independent Variables (Assets, Market Share, Institution Ownership, and leverage) and The Interaction between Leverage and Market Share Unstandardize d Coefficients B (Constant) -1.041 LNASSETS 6.404E-02 MARKET SHARE 7.773E-02 INSTITUTIONAL .449 OWNERSHIP LEVERAGE -5.218E-02 LEVERAGE*MARKET .307 SHARE a Dependent Variable: RESIDUAL Standardize T p-value d Coefficients Std. Beta Error .908 -1.147 .255 .068 .062 .948 .346 .043 .324 1.819 .073 .503 .052 .893 .375 .313 .106 -.010 .534 -.166 2.890 .868 .005 Glejser test for the second regression equation shows that heteroscedasticity is happened as shown in table 5. As the first regression, the second one is also done by residual weighting from the regression equation to overcome the problem. The result is shown in the following table: Table 6 The Result of Residual Variable Regression on Independent Variables (Assets, Market Share, Institution Ownership, and leverage) and The Interaction between Weighted Leverage and Market Share Unstandardized Coefficients B (Constant) .651 LNASSETS -1.986E-02 MARKET SHARE -7.060E-03 INSTITUTIONAL -.110 OWNERSHIP LEVERAGE -.136 LEVERAGE*MARKET 1.209E-02 SHARE a Dependent Variable: RESIDUAL Standardized Coefficients Std. Beta Error .303 .024 -.142 .070 -.168 .178 -.104 .113 .158 T Pvalue 2.149 -.842 -.101 -.619 .037 .404 .920 .539 -.220 -1.209 .127 .076 .233 .940 Heteroscedasticity problem can be overcame by weighting, as showed in table 6 that variables which have significant value less than 5% is no longer exist. Therefore, it can be said that the second regression equation is also free from heteroscedasticity problem 179 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 4.2.2. Multicolinearity This assumption suggest that multicolinearity is not existed among the independent variables. Multicolinearity means that there isn’t perfect linear correlation or near perferct linear correlation between independent variables. This research is using variance inflation factor (VIF) and tolerance to detect multicolinearity. The result shows that the entire independent variable have VIF value far below 10 (the lowest one is 1,101 and the highest one is 1,255) and TOL value closed to 1 (the lowest one is 0,797 and the highest one is 0,908). It can be concluded that this assumption is fulfilled, that is no multicolinearity. Table 7 The Diagnosis Result of Multicolinearity of The First Regression Collinearity Statistics (Constant) LNASSETS LEVERAGE INSTITUTIONAL OWNERSHIP MARKET SHARE Tolerance VIF .651 .820 .635 1.536 1.220 1.575 .848 1.180 a Dependent Variable: ABDACC b Weighted Least Squares Regression – Weighted by Unstandardized Residual There is multicolinearity problem in the second regression because VIF and TOL value of market share variable and the interaction between leverage market share variable are more than 10 as shown in table 8. Table 8 The Diagnosis Result of Multicolinearity of The Second Regression Collinearity Statistics (Constant) LNASSETS LEVERAGE INSTITUTIONAL OWNERSHIP MARKET SHARE LEVERAGE*MARKET SHARE Tolerance VIF .687 .092 .860 1.456 10.877 1.163 .822 .086 1.216 11.649 a Dependent Variable: ABDACC b Weighted Least Squares Regression – Weighted by Unstandardized Residual 180 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 4.2.3. Auto correlation Auto correlation is defined as the correlation between time series observation or cross section observation (Gujarati, 1995). This assumption suggest that there must be no auto correlation, it means that every predicted value is independent Based on Durbin-Watson table, it can be found that for 81 observation and four independent variables the is dL equal to 1,534 and dU is equal to 1,743 on 5% degree. The result of calculation shows that the d is equal to 2,188 on the null hypothesis acceptance zone, it means that there is no correlation. The second regression is also shows d equal to 2,123, it means that there is no autocorrelation. 4.3. Hypothesis Test After the assumption test has been done, the next step is conducting a test on hypothesis. A test on all of the hypothesis as have been explained in chapter 3 using multiple regression and interacted multiple regression. The result of the first regression is presented in the following table: Table 9 The results of Discretionary Accruals Regression on Independent Variables (Assets, Market Share, Intitution Ownership, and Leverage) Expected Unstandardized Sign Coefficients B (Constant) LNASET PANGSA PASAR KEPEMILIKAN INSTITUSI LEVERAGE + - -.331 -2.078E-03 .558 .447 - 9.518E-02 T Std. Error .321 -1.030 .022 -.093 .002 312.589 .328 1.361 .163 .585 p-value .310 .927 .000 .181 .562 aDependent Variable: ABUDACC b Weighted Least Squares Regression – Weighted by Unstandardized Residual Table 9 shows that constant coefficient is 0.331 with the negative sign –0,002; 0,558; 0,447; and 0,09 in order for assets, market share, institution ownership, and leverage variables. Regression result shows the acceptance of the first hypothesis that market share has a positive correlation with profit management on 1 % significance degree. It means that the bigger the marker share is, the bigger the possibility of company conducting profit management. Positive accounting hypothesis stated that a company which bear a bigger politics burden will tends to conduct earnings management which decrease earnings. This research do not make the differences whether company conducting earnings management which decreasing or increasing earnings because the number of sample is insufficient if considering the problem. Besides this research do not focus on positive accounting hypothesis but emphasize more on market share which represent market power of the company. Based on table 9, it is known that institution ownership and leverage variable is insignificance on earnings management so that the second hypothesis and the third one do not proved. These insignificance of the two variables may caused by the 181 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 financial institution are not professional so that they cannot detect whether there is earnings management or not. Murphy and Van Nuys (in Wilopo and Sekar, 2002) supporting this findings, they found that not all institutional ownership affect positively on earnings management, but its depend on the kind of institution, such as pension funds. This study also not considering the size of institution, while the size of institution might be the caused of the insignificance of this variable because small institution are less active in pressure the management activities than the bigger one. Besides, the investor probably are short-term oriented investors. Thus, the investors do not consider the prospect of company in the long-term. They do not also consider the process of creating the numbers in the financial report because the only thing they consider is the magnitude of current profits. The Free-rider also could lead to the insignificance of this variable, because if the majority of ownership is held by one institution so that the other investor are just free-rider. The third hypothesis also do not supporting. May be because the creditor monitor its financial performance of the debtor loosely. This loose monitoring will motivate earnings management, in other words we can say that this monitoring mechanism could not prevent company from earnings management. The pooled debt, short-term debt and long-term debt, could lead to inappropriate proxy because this mislead the function of external monitoring that do not held by shot-term debt. Assets variable also found insignificance to earnings management. The results of the second regression are as follows:: Table 10 The Results of Regression of Discretionary Accruals to Independent Variables: Assets, Market Share, Institutional Ownership, Leverage, and Interaction between Leverage with Market Share Variables Assets Market share Institutional ownership Leverage Market share*leverage Expected Koefisien Standard Sign Regresi Error T p-value -0,002 0,018 -0,129 0,898 + 0,907 0,083 10,885 0,000 0,586 0,216 2,711 0,010 0,429 0,115 3,723 0,001 -0,795 0,190 -4,190 0,000 The second regression also weighted by its residual to overcome heteroscedasticity problems. All coefficient of independent variables are now become -0,002; 0,907; 0,586; 0,429 and –0,795 in order for assets, market share, institutional ownership, leverage and interaction between leverage with market share. All independent variables, except assets, are significant at 1%. Market share has positive sign as predicted and significant at 1%, it means that market share influence earnings management of company. Institutional ownership has positive sign not like the prediction and ignificat at 1%. It means that this study not supporting the statement that the institutional ownership influence earnings management but in the different sign. These may be because the size of institution that ignored, the investor orientation, and also the existence of free-rider. Leverage also in different sign but significant at 1%. This may be because the monitor of creditor are weak, thus the use of debt as external funding do not affect company earnings management. The fourth hypothesis that leverage interact with 182 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 market share in affect the earnings management supported and in the same direction (as predicted). We can say that bigger market share company if accompanied by sufficient external monitoring (proxied by high leverage) hence the possibility of earnings management will decline. 5. CONCLUSION AND LIMITATION 5.1. Conclusion This research is aim at testing the effect of market power (represented by market share) and external control (represented by institution ownership and leverage) on profit management. The first regression results only support the first hypothesis that market share variable which reflecting market share power on influencing profit management are in the positive direction on the degree of significancy 1%. The second and the third hypothesis did not supported. The results of the second regression support upon the fourth hypothesis that market power in influencing profit management interacts with leverage in the positive direction. It means that the magnitude of earnings management of company with sufficiently large market share will decline when there are an appropriate control from external parties, that is creditor. This condition indicate that the control from creditor on company can decrease profit management even when company has a big market share. 5.2. Research Constraint There are several research limitation that should be considered when we will make an inferences about this study. They are: 1. The number of observation are only 81 which is came from 27 manufacturer in 1998-2000 period. This is done because this research needs to identify company market share based on the group of parent company. In addition, the use of industry sales from BPS has made the author to constrain the research period because there are any differences in the industrial classification before 1998 and after ward. Therefore the next researcher is expected to expand the test by adding the number of observation, such as including non-manufacture sector. 2. The interaction between market share and leverage might have made the emergence of multicolinearity problem that the readers should be careful in interpreting the result of this research. The problem is that with the increase of correlation degree between variables, OLS predictor can still be acquired but the standard deviation tends to grow. Therefore the probability of the type II error will increase, the predictions of OLS parameter and its standard deviation will be very sensitive to the change in the sample data even the smallest one, if the multicolinearity is high, maybe R2 can be high but the prediction of regression coefficient which is significant in statistic term is only a few. 3. Institution ownership variable is not supporting the hypothesis because institution ownership include small institution or incompetent in the field of capital market. The next researcher is expected to be able to sort the size of institution so that the effect can be seen in the big or competent institution only. 183 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 4. Leverage is also not supporting the hypothesis because it include short term account payable which is not reflecting external control. The next research is expected to be able to overcome this problem, for example by expanding the sample . References Cahan, Steven F. 1992. The effect of Antitrust Investigations on Discretionary Accruals: A Refined Test of The Political-Cost Hypothesis. The Accounting Review Vol.67, No. 1. DeAngelo, L., 1988, Managerial competition, information costs and corporate governance: the use of accounting performance measures in proxy contests, Journal of Accounting and Economics 10.) Erickson, Merle dan Shiing-wu Wang, 1999, Earnings Management by Acquiring Firms in Stock for Stock Mergers, Journal of Accounting and Economics 27. Jain, Bharat A. dan Omesh Kini, 1994, The post-issue operating performance initial public offerings, Journal of Finance 49 (5). Jones, Jennifer J., 1991, Earnings management during import relief investigation, Journal of Accounting Research 29. Kasznik, Ron, 1999, On the association between voluntary disclosure and earnings management, journal of accounting research 37. Key, Kimberly Galligan, 1997, Political incentives for earnings management in the cable television industry, Journal of Accounting and Economics 23. Keating, A Scott dan Jerold L. Zimmerman, 2000, Depreciation-policy changes: tax, earnings management, and investment opportunity incentives, Journal of Accounting and Economics 28. Koch, Timothy W. dan Larry D. Wall, 2000, The use of accruals to manage reported earnings: theory and evidence. Loughran, Tim dan Jay R. Ritter, 1997, The Operating Performance of Firms Conducting Seasoned Equity Offerings, The Journal of Finance, vol. 52, no.5. McNichols, M dan G.P. Wilson, 1988, Evidence of earnings management from the provision for bad debts, Journal Accounting Research (supl.) Michelson, S.E., J.J. Wagner dan C.W. Wotton, 1995, A market based analysis of income smoothing, Journal of Business, Finance, and Accounting (des). Penman, Stephen H. dan Xiao-Jun Zhang. 2002. Accounting Conservatism, the Quality of Earnings, and Stock Returns. The Accounting Review Vol. 77 No. 2. Perry, Susan E. dan Thomas H. Williams, 1994, Earnings management preceeding management buyout offers, Journal of Accounting and Economics 18. 184 SNA VII DENPASAR – BALI, 2-3 DESEMBER 2004 Saiful dan Jogiyanto HM, 2002, Hubungan manajemen laba dengan kinerja operasi dan retur saham di sekitar IPO, Simposium Nasional Akuntnasi 5. Salno, Hanna Meilani dan Zaki Baridwan, 2000, Analisis perataan penghasilan, factor-faktor yang mempengaruhi dan kaitannya dengan kinerja saham perusahaan curric di Indonesia, Jurnal Riset Akuntansi Indonesia 3(1). Samlawi dan Bambang Sudibyo, 2000, Analisis perataan laba didasarkan pada kinerja perusahaan di pasar, Simposium Nasional Akuntansi 3. Scott, William R., 1997, Fianacial Accounting Theory, Prentice Hall, Inc. Siregar, Sylvia, 2002, Pengaruh pertumbuhan utang dan asimetri informasi terhadap penilaian pasar atas discretionary accruals, Simposium Nasional Akuntnasi 5. Shivakumar, Laksmanan, 2000, Do firms mislead investor by overstating earnings before seasoned equity offerings, Journal of Accounting and Economics (29). Teoh, Siew Hong,Ivo Welch dan T.J.Wong, 1998, Earnings management and the post-issue-performance of seasoned equity offerings, Journal of Financial Economics (okt). Teoh, Siew Hong, T.J. Wong dan Rao, 1998, Earnings management and the long term market performance of initial public offerings, Journal of Finance (des). Utama, Siddharta. 2000. Teori dan Riset Akuntansi Positif: Suatu Tinjauan Literatur. Jurnal Ekonomi dan Bisnis Indonesia. Vol. 15 No. 1. Watts, Ross L. dan Jerold L. Zimmerman. 1986. Positive Accounting Theory. Prentice/Hall International, Inc. Wolks, Harry I., Michael G. Tearney, dan James L. Dodd, 2001, Accounting Theory: A Conceptual and Istitutional Approach, South-Western College Publishing. Zimmerman, J.L., 1983, Taxes and firm size, Journal of Accounting and Economics 185