“CEO behaviour; the influence of CEO income and supervisory board salary”

(1)

“CEO behaviour; the influence of CEO income and

supervisory board salary”

Author Ewoud Eldert Venema

Student reference 1482726

University Univerity of Groningen Faculty Faculty of Economics and Business

Study Corporate Financial Management

University supervisor Prof. dr. D.M. Swagerman

(2)

Acknowledgements

Leadership is one of the most important elements in determining whether a company is successful or not. Especially the 2008 credit crunch boosted my interest for the way CEOs manage their companies and the effects of motivational tools for CEOs. This interest resulted in starting a research in the topic of CEO behaviour.

Writing this thesis was an interesting process. Not always was it an easy job. For the reason that thesis writing is a long term job and a goal so far away is not always easy to pursue. Therefore, I would like to use this opportunity to thank the people who helped me finishing this master thesis. First of all, I would like to thank my supervisor Dirk Swagerman, who let me determine the pace for working on this thesis. Furthermore, with his feedback I was able to find the right direction for this study. Secondly, I would like to thank my parents. Not only for the thesis writing process, but also for the support they gave me throughout the course of my studies.

(3)

Table

of

After the credit crunch of 2008, the corporate governance practices of large (multi) national firms received renewed attention. The paradox of firms making huge losses and firing lots of personnel on one hand, and gigantic chief executive officer (CEO) salaries on the other hand intensified the attention on the way firms are governed. Especially, the power the board of directors had and still has, comes forward by this paradoxical situation. Already in the 1990s firms recognized the importance of incentive alignment of CEOs with those of the company and shareholders. Firms started to come up with new ways of stimulating CEOs to maximize long term company value. However, the 2008 credit crunch made governments realize that their liberal view on how business should be run might not have been the best practice.

(6)

For these reasons, this study investigates whether the level of certain income components of CEOs changes their behaviour. Subsequently, this thesis examines the relation between the executive board and the supervisory board. In the Netherlands supervisory board members spend on average only 1.5 day a month on their board function. This amount of time might be too little to execute the tasks and responsibilities of the supervisory board function. For the limited time they spend on controlling the firm, they cannot bear the amount of responsibility they currently have. A research by the Erasmus University of Rotterdam indicates that 75% of the alumni of a supervisory board study program find that the functioning of the supervisory board is not optimal. Due to this limited functioning of the supervisory board, CEOs might be able to influence the board and exert power over them easily.

(7)

This study uses firm specific panel data for the 2001 to 2007 time span. Firms in this study are Dutch and listed at the Euronext Amsterdam. All firms have revenues larger than €100.000.000,- for 2008. Furthermore, the study uses data on all CEO income components and supervisory board income for the year 2006 and 2007.

Firstly, this paper will outline the theory on this subject by telling something about the long term incentive plans, company earnings management, the effectiveness of supervisory boards and the Dutch remuneration practices of the last decade. Secondly, the theoretical background leads to the formulization of the research hypotheses. Subsequently, this thesis presents the methodology of the research. Next, the description of the data follows presenting the summary statistics of this thesis’ data. Then, this research continues with the presentation of the results. Finally, this paper draws the conclusion by naming the most important results and implications of this study.

I. Theoretical background

A. Long term incentive plans

(8)

(9)

structure of the LTIPs creates openings for CEOs to increase their salaries. LTIPs differ a lot between companies and are therefore hard to compare. CEOs make use of this heterogeneity and the complexity of these structures to increase their own salary. Buck et al.(2003) find that the sensitivity of income to shareholder return decreases with the presence of LTIPs in executive compensation packages. This result supports the ‘power perspective’ which argues that CEOs try to increase their secure income, i.e. income that is certain and irrespective of firm performance. This suggests that executive compensation increased due to the complexity of the compensation packages. And the existence of LTIPs is an opportunity for CEOs to increase their income. This result implies that LTIPs do not improve incentive alignment. However, the result of this research is based on a one year regression analysis and hence, does only give an indication of possible misalignment due to the existence of LTIPs. Buck et al. argue that these new reward systems might induce managers to enrich themselves at the cost of the shareholders and other stakeholders of the firm.

B. Earnings management

(10)

on the balance sheet as accounts receivable, therefore the results seem higher than the actual results will turn out to be. In fact, he should have posted an item on the balance sheet that corrects for the sales that will not be converted into cash. If the CEO does not create this item, he manages the company earnings to enhance company results. Hence, via this channel and via the other income accruals, companies can manage their earnings. Beneish and Vargus (2002) investigate the way executives manage company earnings in order to make their company results look better. They compared earnings management with insider trade activity among firms. So they investigate the behaviour selling and buying of company stocks by the managers of that firm. Their study finds that for firms with abnormal stock buying the one year ahead persistence of the income accruals was significantly higher. This result implies that managers improve the company results to enhance stock prices, and therefore the value of their stock holdings. For firms with abnormal insider selling the one year ahead persistence of the income accruals was significantly lower. After selling their own stock holdings, company results are less important to managers and consequently stock prices are lower.

These results suggest that managers manipulate company results to maximize

(11)

(12)

conflicts with the goals of shareholders who want to maximize their value and would therefore accept all positive NPV projects.

C. Role of the supervisory board

(13)

whether a certain policy that is favourable to executives is executed will depend on the amount of outrage the policy will generate among company stakeholders.

D. Dutch remuneration practices

(14)

(15)

II. Hypotheses

CEOs manage company earnings to maximize their income (Bergstresser and Philippon (2006), Vargus and Beneish (2002) and Kuang (2008)). First, LTIPs of companies increase the incentives for CEOs to manage earnings and maximize their salaries. The outline of LTIPs attaches CEO salary to company performance and therefore these incentives exist. Jensen (2003) argues that the existence of performance target bonuses causes CEOs to game the system and maximize their income. He argues that executives destroy much value because managers change their behaviour when budgets and targets are to be made. Jensen points out that these targets create incentives for managers to lie about the company results in order to make the target and receive the bonus, even though this means that realisation of the successive target becomes harder. Cools (2008) proposes to get rid of target bonuses. He believes that bonuses do not improve incentive alignment. In stead, managers are ‘incentivized’ to realize the performance targets rather than caring about the long term company performance. The foundation of target bonuses shifts the focus of managers from company performance to their own income. For these reasons this research hypothesizes:

1. There exists a positive relation between the variable fraction of income and earnings management.

(16)

have more possibilities to manipulate company results too. Therefore, I expect LTIPs to increase earnings management:

2. There exists a positive relation between the equity fraction of income and earnings management.

Not only the high variable income may indicate the existence of earnings management, also the general level on CEO income might be an indication for earnings management. The power perspective suggests that managers try to maximize their secure income. If managers are able to enhance their salaries, they might be able to exert more power over the level of earnings of the firm too. Therefore the study hypothesizes:

3. There exists a positive relation between the total CEO salary and earnings management

(17)

(18)

decisions in the best interest of the company. The role they have as an advisor and controller of the firm should not correlate with company performances. However, the Tabaksblat code does not indicate a general compensation level for supervisory board members. Cardinaels (2009) investigates the relation between supervisory board pay and their monitoring function on CEOs. After controlling for, among other variables, hospital size, he finds that CEOs of Dutch hospitals receive a higher compensation if the compensation for supervisory board members is higher too. He suggests that the monitoring abilities of the supervisory board members are hampered if these board members receive higher remuneration. This supervisory board culture causes a lack of monitoring within companies, and enables CEOs to manipulate company earnings. Therefore, this study hypothesizes the following:

4. Higher supervisory chairmen remuneration positively relates to company earnings management.

III. Methodology

In order to be able to test the behaviour of CEO this study uses a model used by Kuang (2008). This model is a revision of the model of Dechow and Dichev (2002). Accordingly, this study predicts the earnings management via the abnormal accruals. The previous scholars present the following ordinary least squares regression model: t j t j t j t j t j t j t

j x xCFO x CFO x CFO x v x PPE v

(19)

The variables in this formula represent the following: ) ( ) ( , , , , ,t jt jt jt jt j CA Cash CL STDEBT

TCA       , firm j’s total current accruals in year t.

t j

CA_,

 stands for firm j’s change in current assets between year t-1 and t.

t j

Cash_,

 stands for firm j’s change in cash holdings between year t-1 and t.

t j

CL_,

 stands for firm j’s change in current liabilities of firm between year t-1 and t.

t j

STDEBT_,

 stands for firm j’s change in debt between year t-1 and t.

t j t j t j NIBE TA

CFO_,  _,  _, , firm j’s cash flows from operations in year t.

t j

NIBE, stands for firm j’s net income before extraordinary items in year t. )

( )

( _, _, _, _, _,

,t jt jt jt jt jt

j CA Cash CL STDEBT DEBN

TA        , firm j’s total

accruals in year t.

t j

DEBN _, stands for firm j’s change in depreciation and amortization expense in year t.

t j

v_, Re

 stands for firm j’s change in revenues between year t-1 and t.

t j

PPE_, stands for firm j’s gross value of property, plant and equipment in year t.

t j

v _, measures the value of firm j’s abnormal accruals.

(20)

Dutch listed firms. The firms are listed at the Euronext Amsterdam. This study runs a cross sectional time series analysis because the parameter coefficients are likely to differ across firms (Dechow and Dichev 2002). However, the study only investigates 8 years, which is a too short period to fully rely on a time series analysis. Hence, it uses a panel data analysis to increase accuracy of the earnings management estimation. The Hausman test of Appendix A points out that this sample needs a fixed effects model for analyses. The variable v_j_,_t is the residual estimate of the regression. This variable measures the value of abnormal accruals. The sign of the variable v_j_,_t tells in what direction the firm manages its earnings. During periods of economic growth companies have incentives to present lower earnings than actual earnings in order to build up reserves. In contrary, during years of economic deterioration companies have incentives to camouflage bad results, and thus might manage earnings upwards. During these periods the sign of the accruals can be positive. During periods of company wellbeing earnings management will be negative. This study uses the absolute value of abnormal accruals because it does not make a distinction between the directions of earnings management. So whether the variable

t j

(21)

squares method (Amemiya (1973, 1985)). Thus, for the hypotheses testing this research uses the following Tobit regression model:

j j j

j

j y yVAR Control Industry

AA  ₀  ₁   

ln (2)

j

AA

ln stands for the natural logarithm of the absolute value of total abnormal accruals. This study uses the natural logarithm in order to improve the distribution of the sample. Var stands for the variable part of income and is the ratio of variable income to total CEO income. The variable part of income consists of bonuses, stocks and options. Control_j represents the control variables for the regression analysis.

j

Industry controls for the industry effects on the dependent variable.

(22)

For the testing of the influence of the equity part of CEO income I use the following regression:        _j _j _j

j y y Equity Control Industry

AA ₀ ₁

ln (3)

Regression (3) estimates the relation between the equity part of income and earnings management. Equity is the ratio of stocks and options to total income. I expect that this parameter positively relates to earnings management.

Subsequently, this study tests how the parameter ln Total, which is the natural logarithm of the total value of CEO income, relates to earnings management. The following regression examines the effects of the total salary of a CEO on earnings management:        _j _j _j

j y y Total Control Industry

AA ln

ln ₀ ₁ (4)

Lastly, this study examines the influence of supervisory board salary on earnings management:        _j _j _j j y y SB Control Industry AA ln ln ₀ ₁ (5)

(23)

(24)

IV. Data analysis

(25)

and the dividend receipts are 4%. Subsequently, for the data on supervisory board salary this research subtracts data from the annual reports of the specific firms. The study omits the firms for which the data set is incomplete and the firms for which the VEB does not present data on CEO salaries. The final data set contains 66 firms and includes 118 firm years for the regression analysis.

A. Dependent variable

(26)

Table I. Statistics for abnormal accrual analysis

Variable TCA CFO t+1 CFO t CFO t-1 dRev PPE

Median -203518 236300 196411 176062 14577 145022 Mean -1129137 1189761 1160374 1078982 -13820 867143 Standard deviation 3013884 3203874 3226248 3198793 1408535 1970409 Min -26900000 -3277000 -4415000 -6821405 -16700000 0 Max 1122871 29400000 29400000 29400000 13400000 11900000 Skewness 5,27 5,17 5,27 5,34 -2,40 3,20 Kurtosis 36,17 35,07 36,17 37,55 69,96 13,27 OBS 451 443 449 448 458 457

Table II. Correlation matrix abnormal accrual analysis

Variable CFO t+1 CFO t CFO t-1 dRev PPE

CFO t+1 1

CFO t 0,93 1

CFO t-1 0,87 0,93 1

dRev -0,09 -0,18 -0,29 1

PPE 0,74 0,69 0,62 -0,07 1

(27)

and 15,70 and has a standard deviation of 1,96. More importantly, the skewness of this variable is approximately 0 and the kurtosis value approximates 3, these statistics indicate that the variable is normally distributed.

Table III. Summary statistics dependent variable

Abnormal accruals (A) Absolute value of A (AA) ln AA

Median 1663 43146 10,67 Mean 0,002 220208 10,67 Standard deviation 573921 529889 1,96 Min -6579254 249,43 5,52 Max 3121341 6579254 15,70 Skewness -2,97 5,95 -0,12 Kurtosis 48,52 55,80 2,71 OBS 439 439 439 B. Independent variables

(28)

€ 638.241 respectively. These values indicate that there exist large differences in the value of stock and options plans among firms. The variable part of income consists of bonuses, stocks and option plans paid out to CEOs. The variable part of income is on average 43%. This part of CEO income fluctuates between 0% and 84% of the total earnings of CEOs. The subsequent column shows the summary statistics for Equity. The statistics for this parameters show that the average part of equity payments to CEOs is 0.17. Again, because of the large standard deviation, there exist large differences in the equity payments between firms. Next, the average of total CEO salary is € 1.527.406 over 2006 and 2007. The highest salary of a CEO is € 9.663.913 which is the combined salary of both CEOs that were head of Ahold in 2007. The highest salary paid to a single CEO was € 8.545.000 paid to the CEO of Wolters Kluwer in 2007. The height of this salary can mainly be subscribed to the option plan she received worth € 5.620.000. The lowest CEO salary of this sample is € 142.916 paid to the CEO of Kardan in 2007. The last column of Table VI. shows that the chairmen of supervisory boards earn on average over 2006 and 2007 €43.974. The salaries of supervisory board members range from € 0, paid to the supervisory board members of AFC Ajax, to €112.500, paid to the supervisory board chairman of Akzo Nobel in 2007.

Table IV. summary statistics independent variables

Base Bonus Pension Remain Stocks Options Var Equity Total SB

(29)

C. Control variables

Table V. presents the control variables. To control for the effects of leverage on earnings management this study includes the debt value of the company. Furthermore, column 2 represents the percentage sales growth. Sales growth is a control measure for the increase of income accruals. The volatility of cash flows controls for abnormal accruals that occur due to large deviations in cash flows. The volatility of cash flows is the standard deviation of cash flows over the 2001 to 2008 time frame. At last, the market to book value controls for the size of the firm.

Table V. summary statistics control variables

ln Debt Size BM Sgr ln Volcfo

Mean 12,46 13,26 3,26 0,15 11,82 Median 12,45 13,36 1,33 0,04 11,71 Std. Dev. 1,56 1,99 8,75 0,96 1,37 Min 9,65 6,91 0,18 -0,86 9,58 Max 17,05 17,40 71,97 14,70 16,33 Skewness 0,49 -0,29 6,84 12,75 0,62 Kurtosis 2,98 3,44 50,96 188,01 3,20 OBS 302 118 118 282 302

(30)

to fall in the dummy variable trap, this study compares the effect of the specific industry towards the Oil and Gas industry.

(31)

Table VI. Correlation matrix independent variables

Var Equity ln Total ln Sb ln Debt Size BM Sgr ln Volcfo

Var 1 Equity 0,78 1 ln Total 0,76 0,64 1 ln Sb 0,62 0,50 0,80 1 ln Debt 0,58 0,39 0,81 0,77 1 Size 0,71 0,48 0,82 0,74 0,86 1 BM -0,27 -0,20 -0,25 -0,17 -0,29 -0,54 1 Sgr 0,13 0,05 -0,13 -0,07 -0,21 -0,08 0,07 1 ln Volcfo 0,54 0,31 0,73 0,77 0,87 0,76 -0,13 -0,10 1

Furthermore, Table B2. shows the correlation matrix for the income variables. There exists a high correlation (0.76) between the base salary of CEOs and the bonuses they receive. Many independent variables correlate with each other. High correlations (0.85) exist between the base salary and bonuses of CEOs. Furthermore, supervisory board salaries correlate to many salary components of CEO salary and show especially a high correlation between the base salary of CEOs (0.63) and the variable part of CEO income (0.62).

V. Results

(32)

The first analysis tests the H1, expecting that companies that pay their CEOs larger shares of variable income have higher earnings management. Table VII. presents the results of this Tobit model. The first column of the Table shows the coefficients of the parameter Var. The coefficient of this parameter is positive (0.32) but statistically insignificant, indicating that companies that pay their CEOs a relative high share of variable income do not have significantly higher earnings management. Remarkably the variables Var and Size do not show significant coefficients. The insignificance of these parameters might be due to the high correlation of the two variables with each other. Table VI. presents the correlations between the two variables, which is 0.72. In addition, the correlation between Var and ln Debt is 0.61. Moreover, the standard deviation of the variables Var and Size is relatively high. The standard error for Var is 0.69 which is high compared to its coefficient (0.32). For Size this number is 0.13, compared to a coefficient of 0.14, this is high too. The high correlation and the relative size of the standard errors indicate that there might exist multicollinearity between the variables. This multicollinearity influences the significance of the results and could influence the value of the coefficients. Thus, this study cannot ascribe the movement of the dependent variable to one of these independent variables properly because the parameters behave in a similar manner. Hence, the result of this regression does not support H1, which states that higher variable parts of CEO income positively relate to earnings management.

(33)

insignificant. Again, this insignificance might be due to the high correlation of the parameter with Size and ln Debt. Table VI. shows that correlation of Equity with Size and ln Debt is 0.51 and 0.43 respectively. Furthermore, again the standard error of Equity is 0.53 which is high compared to the coefficient of 0.23. The high correlation and the relative high standard deviation imply that multicollinearity influences the significance of the parameter Equity. So, this model does not support H2. The third column of Table VII. shows the Tobit model including the independent variable ln Total. H3 states that higher total income of a CEO enhances the level of earnings management of a company. The model does not find results supporting this hypothesis however. The parameter ln Total has a positive coefficient which is insignificant. Lastly, this study investigates the effects of the salary of supervisory board chairmen. Table VII. reports the results for this regression. Surprisingly, the table shows that ln SB has a negative sign, however not significant. H4 argues that the sign and the relation between supervisory board salary and earnings management should be positive. The second column of Table VII. shows that the standard error of ln SB is particularly high (compared to its coefficient). This high standard deviation, together with the correlation between ln SB and ln Debt and Size of 0.77 and 0.74 respectively, could mean that there exists multicollinearity between the variables influencing the coefficients and significance of the parameters.

(34)

coefficient of ln Debt varies between 0.80 and 0.92. Logically, this parameter has a positive sign as firms with more debt tend to have higher earnings management. The coefficient of Size varies between 0.14 and 0.19. Although generally, the income accruals of larger firms are easier to predict, the size of the accruals creates higher earnings management fluctuations. As expected, the coefficient of Sgr is positive and significant. The results of growing firms are harder to predict and therefore the abnormal accruals of these firms are larger. Hence, the coefficient of this parameter is positive and significant in this model. Book to market ratio has a positive significant coefficient in all models. This result indicates that firms with a larger book to market value have higher abnormal accruals. This result is contraire to the expectations, which predicted a negative relation for this parameter because the predictability of company income accruals increases as book value increases. However, a positive sign of BM can be explained to the size effects of firms creating higher abnormal accruals. Next, the effects of different industries are not significant. The number of industries is too large to identify significant effects on company earnings management. This same reasoning holds for the change of CEO position. There were too few changes of CEO positions to identify significant effects of this parameter. During the 2006/ 2007 time frame there were only eight changes of CEO position. This number is low compared to the total number of firm years (128).

(35)

(36)

Table VII. regression on earnings management

Dependent variable: ln (aa) Method: tobit regression model

Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error

C -3.38 1,31 -3,28 1,08 -4,47 2,09 -1,79 3,45 Var 0,32 0,69 Equity 0,23 0,53 ln Total 0,09 0,21 ln SB -0,25 0,41 ln debt 0,91 *** 0,12 0,80 *** 0,15 0,89 *** 0,16 0,92 *** 0,14 Size 0,14 0,13 0,19 0,13 0,14 0,14 0,18 0,13 BM 0,06 *** 0,01 0,06 *** 0,02 0,06 *** 0,02 0,06 *** 0,02 Sgr 1,08 * 0,59 0,13 ** 0,07 1,13 * 0,57 0,91 0,57 ln volcfo 0,01 0,00 0,08 0,15 0,02 0,15 0,06 0,15 dCEO -0.03 0,39 -0,14 0,39 -0,07 0,40 -0,07 0,39 Ind1 0,21 0,65 0,03 0,64 0,26 0,65 0,13 0,64 Ind2 0,10 0,52 0,01 0,53 0,11 0,53 -0,07 0,53 Ind3 -0,27 0,57 -0,49 0,57 -0,25 0,58 -0,40 0,57 Ind4 -0,04 0,56 -0,24 0,55 -0,01 0,56 -0,21 0,58 Ind5 -0,53 0,87 -0,57 0,89 -0,48 0,88 -0,68 0,87 Ind6 -0,28 0,55 -0,32 0,56 -0,25 0,55 -0,25 0,56 R² 0,34 0,34 0,34 0,34 N 116 118 116 110

(37)

Table VIII. shows the Tobit regression including only the firms that pay their CEOs a base salary higher than €500.000,- per year. The sample for this model includes only 44 companies. The coefficient of Var has a value of 2.72 and is significant at the 5% level. This positive sign indicates that companies that pay their CEOs a larger share of variable income have higher abnormal earning accruals. The pseudo r squared value of this model is 0.37 which is even higher than the ones examining the complete sample indicating that the explanatory power of this model is relatively high. Moreover, the Var remains significant with control variables ln Debt and Size included in the model. Therefore, the result of this model indicates that CEOs manage earnings when their base salary is higher than €500.000.

(38)

independent variables Var and ln Total. However, points around €500.000,- show comparable regression outcomes for both variables.

(39)

Table VIII. regression on earnings management

Dependent variable: ln (aa) Method: tobit regression model Base salary

>€500.000,-Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error

C -5,96 *** 2,14 -3,06 -1,61 -17,76 *** 5,65 -16,65 * 8,45 Var 2,72 ** 1,35 Equity 0,88 1,03 ln Total 1,08 ** 0,46 ln SB 1,03 0,76 ln debt 1,57 *** 0,49 1,16 ** 2,44 1,16 ** 0,47 1,20 ** 0,48 Size -0,49 0,56 -0,09 -0,16 -0,26 0,51 0,04 0,52 BM -0,20 0,25 -0,14 -0,55 -0,01 0,25 -0,03 0,26 Sgr 1,54 * 0,90 0,14 ** 1,93 1,78 ** 0,86 2,35 *** 0,91 ln volcfo 0,10 0,25 -0,01 -0,04 0,09 0,25 -0,07 0,25 dCEO 0,55 0,44 0,26 0,56 -0,07 0,48 0,53 0,45 Ind1 -0,43 0,70 -0,31 -0,46 -0,06 0,65 -0,05 0,68 Ind2 -0,25 0,60 0,21 0,36 -0,18 0,58 0,20 0,59 Ind3 -0,52 0,62 -0,41 -0,72 -0,41 0,59 0,00 0,61 Ind4 -0,41 0,73 -0,31 -0,48 -0,92 0,81 0,43 0,74 Ind5 -1,42 0,91 -0,72 -0,83 -1,01 0,85 -0,71 0,88 Ind6 -0,90 0,54 -1,28 ** -2,22 -0,91 * 0,53 -0,78 0,57 R² 0,37 0,35 0,38 0,36 N 44 46 44 44

(40)

VI. Conclusion

(41)

towards clients or the firm) as taken from a corporate perspective (generating sustainable- instead of short term profits).

The examination of the entire sample, containing more than 110 firm years, does not show any significant results on the CEO salary components and the salary level of the supervisory board. The insignificance of these statistics might be due to the multicollinearity between the independent variables of this research, the income variables, and the firm control variables Debt and Size. Hence, the results from this study do not follow the line of the empirics of Kuang (2009), Bergstresser and Philippon (2006) and Vargus and Beneish (2002). These scholars find support for the premise that incentive compensation increases the level of earnings management. Furthermore, this research does not find support for the paradigm that companies that pay their supervisory boards higher salaries have higher earnings management. Therefore, there is no indication that supervisory boards that earn higher salaries, have a more coolant attitude towards CEOs. This finding is contradictory to what Bebchuk and Fried (2004) argue. They expect that supervisory boards have an accommodating attitude towards CEOs.

(42)

option packages. It is a complex and time consuming procedure to calculate the value of each option package individually. Using a fixed rate for these parameters, significant differences between the perceived value of option packages and the real value of these packages could come into being. Therefore, the real variable part of income could differ substantially. Hence, this valuation difference could jeopardize the validity of the regression analyses.

(43)

(44)

Appendix A

Table AI. Hausman test

Coefficients (b) fixed Coefficients (B) random (b-B) difference sqrt (diag (V b- V B) S.E.

CFO t+1 0,20 -0,14 0,34 0,02

CFO t -0,78 -0,77 -0,01 .

CFO t-1 0,35 0,21 0,14 .

dRev 0,29 0,20 0,09 .

PPE -1,21 -0,38 -0,83 0,09

b = consistent under H0 and Ha; obtained from xtregression

B = inconsistent under H0 and Ha, efficient H0: obtained from xtregression

Chi2 174,99

(45)

Appendix B

Table BI. panel regression output

Dependent variable: TCA

Method: panel regression fixed effects

Coefficient Std. Error C 199.051,30 * 109766,20 Cfo t+1 0,20 *** 0,04 Cfo t -0,78 *** 0,04 Cfo t-1 0,35 *** 0,03 dRev 0,29 *** 0,03 Ppe -1,21 *** 0,10 R squared within 0,62 N 439

(46)

Appendix C

This study categorizes according to the Industry Classification Benchmark (ICB). The sample contains 7 different industries:

Table CI. Industry distribution

Based on the Industry Classification Benchmark (ICB).

Number of firms

Oil and gas 7

(47)

Appendix D

Table DI. regression on earnings management without control variables ln Debt and Size

Dependent variable: ln abs(aa) Method: tobit regression model

Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error

C 9,06 0,84 -1,50 1,22 -8,53 *** 2,22 -7,65 * 4,18 Var 4,54 *** 0,75 Equity 1,15 * 0,60 Total 0,70 *** 0,20 ln SB 0.71 0,48 BM 0,03 * 0,01 0,02 0,01 0,02 * 0,01 0.01 0,01 Sgr 0,02 0,85 0,04 0,08 0,49 0,64 0.17 0,69 ln volcfo 0,00 *** 0,00 0,99 *** 0,09 0,77 *** 0,11 0,89 *** 0,14 dCEO 0,61 0,57 -0,11 0,46 -0,28 0,45 -0.07 0,48 Ind1 -0,04 0,95 0,38 0,76 0,66 0,76 0,58 0,8 Ind2 -0,40 0,76 -0,36 0,62 -0,09 0,61 -0,28 0,66 Ind3 -0,24 0,84 -0,72 0,67 -0,42 0,67 -0,50 0,72 Ind4 -0,97 0,81 -0,44 0,65 -0,20 0,65 -0,04 0,73 Ind5 0,92 1,26 1,16 1,02 0,86 1,00 1,04 1,06 Ind6 -1,21 0,80 -0,65 0,65 -0,37 0,64 -0,24 0,7 R² 0,16 0,25 0,25 0,23 N 116 118 118 111

(48)

Appendix E

Table EI. Correlation matrix Base >500.000

Var Equity Ln Total Ln SB Size Ln Debt BM Sgr Ln Volcfo

(49)

References

Periodicals:

Amemiya, Takeshi, 1973, Regression Analysis when the Dependent Variable is Truncated Normal, Econometrica. 41, 997–1016.

Bartov, Eli., Givoly, Dan, and Hayn, Carla, 2002, The rewards to meeting or beating earnings expectations, Journal of Accounting and Economics 33, 173-204.

Bebchuck, Lucian A., and Fried, Jesse M., 2003, Executive compensation as an agency problem, Journal of Economic Perspectives, 71-92

Becker, Connie L., DeFond, Mark L., Jiambalvo, James, and Subramanyam, K. R, 1998, The effect of audit quality on earnings management, Contemporary Accounting Research 15, 1–24.

Beneish, Messod, and Vargus, Mark E., 2002, Insider Trading, Earnings Quality and Accrual Mispricing, The Accounting Review. 77, 755–91.

Bergstresser, Daniel, and Philippon, Thomas, 2006, CEO Incentives and Earnings Management, Journal of Financial Economics, 80, 511–29.

(50)

Cheng, Qiang, and Warfield, Terry D., 2005, Equity Incentives and Earnings Management, The Accounting Review. 80, 441–76.

Dechow, Patricia, and Dichev, Ilia D. ,2002, The Quality of Accruals and Earnings: The Role of Accrual Estimation Errors, The Accounting Review 77, 35–59.

Duffhues Pieter, and Kabir, Rezaul, 2008, Is the Pay Performance Relationship Always Positive? Evidence from the Netherlands, Journal of Multinational Financial Management, 18, 45-60.

Graham, John.R., Harvey, Campbell R., and Rajgopal, Shiva, 2005, The Economic Implications of Corporate Financial Reporting, Journal of Accounting and Economics, forthcoming.

Hribar, Paul, and Nichols, Craig D., 2007, The Use of Unsigned Earnings Quality Measures in Tests of Earnings Management, Journal of Accounting Research. 45, 1017–53.

Kahneman, Daniel, and Tversky, Amos, 1979, Prospect Theory: An Analysis of Decision under Risk, Econometrica. 47, 263-291.

(51)

Kuang, Yu Flora, 2008, Performance vested stocks and earnings management, Journal of Business Finance & Accounting 35, 1049-78

Skinner, Douglas J., and Sloan, Richard G., 2002, Earnings Surprises, Growth Expectations, and Stock Returns or Don't Let an Earnings Torpedo Sink Your Portfolio, Review of Accounting Studies 7, 289-312

Swagerman, Dirk M., and Terpstra, Erik, 2007, The Effectiveness of Dutch Executive Pay Packages, Compensation & Benefits Review 39, 47-57

Monograph (Books):

Amemiya, Takeshi, 1985, Advanced Econometrics

Magazines, Newspapers and Websites:

Cools, Kees, 2008, Bonus veroorzaakt manipulatie, Weblog De Volkskrant, October 23.

University papers

Hau, Harald, and Thum, Marcel, 2009, Subprime crisis and board (in)competence: private vs. public banks in Germany, Cesifo working paper 2640

“CEO behaviour; the influence of CEO income and supervisory board salary”