1
Master Thesis
Corporate Governance in the Netherlands, does good
governance have an effect on performance?
J.A. de Vries
Studentnr: 1468278
August 2009
University of Groningen
Faculty of Economics and Business
MSc Business Administration
2 Corporate Governance in the Netherlands, does good governance have
an effect on performance?
J.A. de Vries
Rijksuniversiteit Groningen
August 2009
Abstract
This paper studies empirically the relationship between corporate governance and firm performance in the Netherlands. Some interesting results are found. Firstly, separation of the CEO and chairman function has a significant negative effect on return on equity. Secondly, in contrast to earlier research in the Netherlands, “good” board composition has a significant positive relationship with accounting performance. Thirdly, shareholder rights seems to have no effect on firm performance. Lastly, “good” compensation has a significant negative relationship with market valuation, while it has a positive (non-significant) relationship with accounting performance.
Keywords: corporate governance, performance, the Netherlands
Introduction
3 reduce agency costs, like too high management wages. But agency costs can also exist as managers make investments that are only in their own interest and not in the interest of the shareholders. If corporations have a good corporate governance structure, this agency costs can be reduced. This means that good corporate governance can increase firm performance. Empirically, evidence for this relationship is mixed. Gombers, Ishii & Metrick (2003) find positive abnormal stock returns for firms with high shareholder rights. On the other hand, Bauer, Guenster & Otten (2004) find a negative relationship between corporate governance indicators and accounting performance. Furthermore, a lot of authors find weak results for the relationship between corporate governance and firm performance. Therefore, investigating this relationship is an interesting research objective. Except for earlier research of this topic done by Van Ees, Postma & Sterken (2003), there’s no recent research of this relationship for Dutch firms a far as I know.
So I want to investigate the relationship between corporate governance and firm performance for Dutch listed companies (AEX, AMX and AScX index). This research is relevant because there is still no hard proof of this relationship to exist. New in this research is that is for one typical geographical region (the Netherlands). Furthermore, corporate governance of firms will be measured based on the Dutch code Tabaksblat. So this research can also bring in some new insights in the discussion if governance codes add value for firms or society. The result of this research is that some corporate governance variables have a significant relationship with performance. It appears that separation of the CEO and chairman function has a significant negative effect on return on equity. Secondly, in contrast to earlier research in the Netherlands, “good” board composition has a significant positive relationship with accounting performance. Thirdly, shareholder rights seems to have no effect on firm performance. Lastly, “good” compensation has a significant negative relationship with market valuation, while it has a positive (non-significant) relationship with accounting performance.
4 After the data and methodology part, the results are discussed. The last section concludes.
Literature review & hypotheses
As mentioned in the introduction, there has been a lot of research done already in investigating the relationship between corporate governance and firm performance. One of the first papers on the corporate governance topic was by Berle & Means (1932). According to Kyereboah-Coleman, Adjasi & Abor (2006), they (Berle & Means, 1932) described the agency problem in firms where there is a separation of ownership and control. These firms are run by professional managers (agents) who are unaccountable to dispersed shareholders (principals). Jensen & Meckling (1976) define this agency relationship as “a contract under which one or more persons (principal) engage another person (agent) to perform some service on their behalf, which involves delegating some decision-making authority to the agent.”
5 behave themselves, because they already do that intrinsically. This means that corporate performance is not influenced by the corporate governance structure.
Departing from this theoretical discussion, a link between corporate governance and firm performance will be tried to be established. A lot of authors tried to find this link.
6 In most studies, several corporate governance and performance variables are distinguished. The most used corporate governance variables are board size, CEO-duality, the composition of the board of directors, shareholder rights and management compensation. Performance variables can be divided in accounting performance (like return on assets or return on equity) and market performance/valuation (like stock returns and Tobin’s Q). Below I will discuss the corporate governance variables and the performance variables shortly.
Board size
Theoretically, it can be argued, according to Kiel & Nicholson (2003), that a larger board is more likely to be watchful for agency problems because a greater number of people will be reviewing management actions. Furthermore, Goodstein, Gautam & Boeker (1994) state that resource dependence theorists have argued that by increasing in size and diversity, boards help to link the organization to link the organization to its external environment and secure critical resources. But according to Jensen (1993), small boards can help improve performance. He states: “When boards get beyond seven or eight people they are less likely to function effectively and are easier for the CEO to control”. Empirically, Yermack (1996) presented evidence that small of directors are more effective. Using Tobin’s Q as an approximation of market valuation, he found an inverse association between board size and firm value in a sample of 452 large U.S. industrial corporations in the period 1984-1991. On the other hand, based on meta-analysis of 131 samples, Dalton, Daily, Ellstrand and Johnson (1999) provide systematic evidence of nonzero, positive, true population estimates of board size – performance (accounting and market) relationships.
So overall, evidence regarding the relationship between board size and corporate performance seems to be mixed.
CEO-duality
7 disapproved. Best practice III.8.1 says: “The chairman of the (management) board shall not also be and shall not have been an executive director.”
According to the Agency theory, CEO-duality will reduce the effectiveness of board monitoring and therefore this role should be divided (Finkelstein and D’Aveni, 1994). On the other hand, according to Stewardship theory, one person in both roles may improve firm performance. This is according to Yermack (1996) due to a greater understanding of the company’s operations. Based on a meta-analysis of previous research, Boyd (1995) states that CEO duality might be contingent on the company’s size and challenges. He states: “Evidence presented here indicates that duality can help firm performance – under the right circumstances.” In their research between CEO-duality and firm performance, Baliga, Moyer & Rao (1996) concluded that the market is indifferent (there are no significant announcement effects) to changes in a firm’s duality status, that there is little evidence op operating performance changes around changes in duality status and that there is only weak evidence that duality status affects long-term performance.
Composition of the board of directors
According to Schleifer & Vishny (1997), the structure of boards varies greatly even across developed economies, ranging from two-tier supervisory and management boards in Germany, to insider-dominated boards in Japan, to mixed boards in the United States. In the Dutch Tabaksblat code it is assumed that most companies have a two-tier board structure. In this structure, the role of the supervisory board is “to supervise the policies of the management board and the general affairs of the company and its affiliated enterprise, as well as to assist the management board by providing advice.”
8 relationship between board composition and firm performance. According to Rhoades, Rechner and Sundaramurthy (2000), also based on meta-analysis, the proportion of outside directors had a very small positive relationship with firm performance.
Shareholder rights
A good functioning corporate governance structure also requires good shareholders rights. The Dutch corporate governance code writes the following on this topic: “Good corporate governance requires the fully-fledged participation of shareholders in the decision-making in the general meeting of shareholders. It is in the interest of the company that as many shareholders as possible take part in the decision-making in the general meeting of shareholders. The company shall, in so far as possible, give shareholders the opportunity to vote by proxy and to communicate with all other shareholders.”
Theoretically, not much is known about the impact of shareholder rights on corporate performance. Gompers, Ishii and Metrick (2003) say the following on this issue: “Presumably, shareholders accept restrictions of their rights in hopes of maximizing their wealth, but little is known about the ideal balance of power.” Therefore, Gompers, Ishii and Metrick (2003) tried to find a relationship between shareholder rights and corporate performance empirically. They found some significant evidence that firms with better shareholder rights have better operating performance than firms with worse shareholder rights. Furthermore, an investment strategy that purchased shares with the strongest shareholder rights and sold shares with the weakest shareholder rights earned an abnormal return of 8.5 percent per year.
Compensation
9 there is a tenuous link between the performance of managers and their compensation: “First, the compensation package may be poorly structured. (...) Second, managers often seem to manage to maintain their compensation stable or even have it decreased despite poor performance. (…) Third, managers may succeed in “getting out on time”. (…) Finally, managers receive large golden parachutes for leaving the firm.” Therefore, it is no surprise that the Dutch Tabaksblat code devotes a significant part of the code to compensation of managers and directors. It says the following about the amount and composition of the compensation: “The amount and structure of the remuneration which the management board members receive from the company for their work shall be such that qualified and expert managers can be recruited and retained. If the remuneration consists of a fixed and a variable part, the variable part shall be linked to previously-determined, measurable and influenceable targets, which must be achieved partly in the short term and partly in the long term. The variable part of the remuneration is designed to strengthen the board members’ commitment to the company and its objectives.”
In the literature, the focus on compensation is on CEO-compensation. Bertrand & Mullainathan (2000) empirically examined two competing views of CEO pay. In the contracting view, pay is according to them used to solve an agency problem: the compensation committee optimally chooses pay contracts which give the CEO incentives to maximize shareholder wealth. In the skimming view, pay is the result of an agency problem. According to them, CEOs have managed to capture the pay process so that they set their own pay, constrained somewhat by the availability of cash or by a fear of drawing shareholders’ attention. Core, Holthausen & Larcker (1999) also linked CEO-compensation to agency theory. They suggest that firms with weaker governance structures have greater agency problems, that CEO’s receive greater compensation and that these firms perform (in terms of ROA and stock return) worse.
Performance
10 performance, the most used measures are return on assets (ROA) and return on equity (ROE), but also other measures are used. Rhoades et al (2000), for example used besides this two measures also profit margin as accounting measure for estimating the relationship between board composition and financial performance.
For performance in the financial market, stock returns and market valuation are used mostly. For example, in the case of stock returns, Bauer, Guenster & Otten (2004) found some evidence that corporate governance affects stock returns positively. The most used market valuation measure is Tobin’s Q. This variable describes the market value of the assets relative to the replacement cost of the assets. Yermack (1996), for example, used this measure in linking board size to market valuation.
Hypotheses
After describing the several corporate governance variables and the performance variable, I want to draw up a hypothesis for this research. With regard to the corporate governance variables, I will use the variables CEO-duality, board composition, shareholder rights and compensation. Board size will not be used because I think that this variable lies not in the spirit of this research and this variable is too dependent on too may contingencies in my opinion. Therefore, the hypotheses are the following:
H0a: There is a significant relationship between CEO-duality and firm performance.
H0b: There is a significant relationship between board composition and firm performance.
H0c: There is a significant relationship between shareholder rights and firm performance.
11 Data & Methodology
The methodology that will be used is quite similar to earlier work like Larcker, Richardson & Tuna (2007). It consist of a multiple regression based on ordinary least squares (OLS) of the following form:
Dependent variablet = ΣγControls + ΣβGovernance factorst + εt
Where α, β are parameters, εt is an error term, controls are control variables. For
the dependent variable is as well as accounting performance as market performance used. For accounting performance, return on assets (ROA) and return on equity (ROE) will be used. Market capitalization / Shareholders’ equity (Mcap/SE) will be used as a proxy for market performance. It shows the equity market value relative to the book value of the equity. There will also be some control variables included. For controlling for size, sales (natural log) will be used. For controlling for capital structure, gearing (percentage of total debt to total assets) will be used. And lastly, there will be dummy variables for industry effects. Based on the data, the firms are placed in 6 sectors. These are Manufacturing, Service, Real Estate, Retail and Construction. Firms that can’t be placed in one of these 5 industries are labeled Other.
As mentioned before, I will use 4 governance variables. There are named DUALITY (CEO-duality), COMPOSITION (board composition), SHAREHOLDER (shareholder rights) and COMPENSATION (compensation of managers and directors). I will measure values of this variables based on compliance with the Dutch code Tabaksblat. The Code Tabaksblat gives a lot of best practices for good corporate governance. After connecting the relevant best practices with the variables, I give each firm a score for this variable based on the compliance rate. This compliance rate is measured as the percentage compliance on the relevant best practices1. The following example will illustrate this. For the duality variable, there are two relevant best practices. If a firm complies to one and does not comply to the other, it scores 0.5 (50%) on the duality variable.
Data needed for this regressions is obtained from several sources. As mentioned before, the firm data consists of 75 firms. These are the firms in the Dutch AEX,
1
12 AMX and AScX indices. Financial data, needed for the performance measures and the control variables, is obtained from the Amadeus top 250,000 database. Data is obtained for the years 2006 and 2007, this gives the opportunity to do the regressions for more than one year. Corporate governance data is obtained from the annual reports and the websites of the firms in question. Assumed is that this data is the same for both 2006 and 2007.
The total dataset will not include all 75 firms. There are several reasons for this. Firstly, not all firms in the three indices have a statutory place in the Netherlands. Officially, they don’t have to comply to the Dutch corporate governance code and therefore these firms (9 firms) will not be included in the dataset. Secondly, for some firms there was lack of data availability (9 firms). After these two adjustments, a total of 57 firms remains2. To provide an short overview of this dataset, table I shows descriptive statistics.
It appears that the compliance rates of the governance variables are high. On average compliance rates are above 90% for CEO-duality, board composition and shareholder rights. The compliance rates are the highest for the duality variable. Only two firms score 50% for this variable, the other firms score 100%. Concerning non-compliance for the compensation variable, it appears that a lot of firms do not comply to the best practice concerning severance pay (best practice II.2.7 of the code Tabaksblat). The high compliance rates are in accordance to research by Akkermans et al (2007). They provided an overview of the application of the Tabaksblat code in 2004 and found also a high level of compliance with the code.
The regressions will be estimated first for the total dataset (both years together), these are called the pooled regressions. Secondly, regressions will be estimated for both individual years. After the regressions are estimated, obtained coefficients for the various governance factors are analyzed and discussed.
2
13 Table I
Descriptive statistics dataset
This table shows the descriptive statistics of the dataset used. The following variables are the independent variables: Mcap/SE (Marketcap/Shareholders’ Equity), ROE (Return on equity) and ROA (Return on assets). Duality (CEO-duality), Composition (composition of supervisory board), Shareholder (shareholder rights) and Compensation (compensation of board [management + supervisory] members) are the governance variables. The control variables are Gear (Gearing), Sale (LogSales) and the sector (dummy) variables.
14 Results
Results based on market valuation
Table II shows the results from the OLS regression equation linking corporate governance and performance based on the marketcap/SE-ratio. It appears that the pooled model has a relative low squared value (0.276). The Adjusted R-squared value is slightly lower (0.195).
Looking at the corporate governance variables, it appears that all corporate governance variables are insignificant at the 5% significance level. The duality variable appears to have a positive effect (+3.637) on firm valuation. However, the p-values of both the pooled and the individual years show that this variable is statistically insignificant. The composition variable has very varying values. For 2006, to composition variable has a coefficient of +0.585 and for 2007 this coefficient is -0.639. The pooled coefficient has a value of -0.003. Furthermore, the coefficients are very insignificant. The shareholder variable has also very varying values (pooled: +0.827; 2006: -0.735; 2007: +2.292) and all coefficients are also very insignificant. The compensation variable has more stable coefficients. For the 2006 data, the coefficient is -3.336, for the 2007 data this coefficient is -3.870 and for the pooled data this coefficient is -3.472. Furthermore, the coefficients for 2007 and for the pooled data are significant at the 10% significance level.
Results based on accounting performance (ROE)
15 Table II
OLS Regression results of Marketcap/SE-ratio
This table shows the results of the OLS regressions of Marketcap/SE-ratio (dependent variable) on Corporate Governance variables and Control variables. The pooled results are the results based on the OLS regression of the pooled data. The results for 2006 and 2007 are the results based on the cross-sectional OLS regression of the individual yearly data. P-values of coefficients are in parentheses.
Pooled 2006 2007
R-squared 0.276 0.236 0.418
Adj. R-squared 0.195 0.041 0.269
Independent variable coefficient coefficient coefficient
(p-value) (p-value) (p-value)
DUALITY 3.637 3.477 4.186 (0.234) (0.512) (0.246) COMPOSITION -0.003 0.585 -0.639 (0.998) (0.801) (0.685) SHAREHOLDER 0.827 -0.735 2.292 (0.789) (0.887) (0.541) COMPENSATION -3.472 -3.336 -3.870 (0.076) (0.331) (0.093) GEARING 0.003 0.002 0.003 (0.218) (0.518) (0.217) SALES -0.011 -0.081 0.103 (0.919) (0.637) (0.444) RETAIL 2.844 5.134 -0.153 (0.381) (0.338) (0.970) MANUFACTURE 0.876 3.124 -2.080 (0.783) (0.549) (0.601) CONSTRUCTION 1.316 3.542 -1.635 (0.684) (0.505) (0.687) REAL ESTATE 0.510 3.086 -2.762 (0.862) (0.524) (0.450) SERVICE 2.093 4.592 -1.087 (0.511) (0.381) (0.784) OTHER 2.376 4.611 -0.581 (0.465) (0.389) (0.886)
16 Table III
OLS Regression results of Return on equity
This table shows the results of the OLS regressions of Return on equity (dependent variable) on Corporate Governance variables and Control variables. The pooled results are the results based on the OLS regression of the pooled data. The results for 2006 and 2007 are the results based on the cross-sectional OLS regression of the individual yearly data. P-values of coefficients are in parentheses.
Pooled 2006 2007
R-squared 0.539 0.627 0.612
Adj. R-squared 0.489 0.534 0.515
Independent variable coefficient coefficient coefficient
(p-value) (p-value) (p-value)
DUALITY -72.881 -113.224 -34.877 (0.012)* (0.010)** (0.345) COMPOSITION 29.384 67.728 -7.201 (0.021)* (0.001)** (0.657) SHAREHOLDER -11.691 17.021 -31.033 (0.688) (0.682) (0.421) COMPENSATION 10.501 5.363 21.060 (0.561) (0.843) (0.360) GEARING -0.114 -0.114 -0.076 (0.000)** (0.000)** (0.012)* SALES 5.831 4.616 7.392 (0.000)** (0.002)** (0.000)** RETAIL 7.157 1.263 -9.141 (0.816) (0.977) (0.827) MANUFACTURE -9.967 -10.488 -30.902 (0.741) (0.803) (0.454) CONSTRUCTION 2.345 3.023 -21.826 (0.939) (0.944) (0.604) REAL ESTATE 4.891 0.317 -11.073 (0.860) (0.994) (0.769) SERVICE 3.248 -1.717 -13.069 (0.914) (0.968) (0.751) OTHER 14.645 12.529 -5.279 (0.635) (0.772) (0.900)
17 Table IV
OLS Regression results of Return on assets
This table shows the results of the OLS regressions of Return on assets (dependent variable) on Corporate Governance variables and Control variables. The pooled results are the results based on the OLS regression of the pooled data. The results for 2006 and 2007 are the results based on the cross-sectional OLS regression of the individual yearly data. P-values of coefficients are in parentheses.
Pooled 2006 2007
R-squared 0.511 0.442 0.616
Adj. R-squared 0.457 0.302 0.520
Independent variable coefficient coefficient coefficient
(p-value) (p-value) (p-value)
DUALITY -7.369 -11.730 -4.319 (0.556) (0.597) (0.778) COMPOSITION -0.205 5.694 -5.033 (0.970) (0.558) (0.459) SHAREHOLDER -21.656 -14.479 -28.627 (0.090) (0.505) (0.080) COMPENSATION 3.568 1.640 5.305 (0.651) (0.908) (0.580) GEARING -0.061 -0.056 -0.067 (0.000)** (0.001)** (0.000)** SALES 1.804 1.821 1.742 (0.000)** (0.014)* (0.004)** RETAIL 21.895 15.160 29.215 (0.104) (0.500) (0.099) MANUFACTURE 11.766 4.450 20.045 (0.371) (0.839) (0.247) CONSTRUCTION 13.342 6.515 21.051 (0.320) (0.770) (0.234) REAL ESTATE 17.790 11.383 24.900 (0.144) (0.576) (0.119) SERVICE 15.921 8.186 24.465 (0.228) (0.710) (0.159) OTHER 23.021 15.292 31.631 (0.089) (0.498) (0.076)
18 The composition variable has also very varying values. In 2006, this variable has a large positive value (+67.728) but in 2007 this value is negative (-7.201). For the pooled regression, this coefficient has a positive value (+29.384). For the 2006 data, this coefficient is very significant (p-value: 0.001). In the pooled data case, the composition variable is significant at the 5% significance level. Also the shareholder variable has very varying values (pooled: -11.691; 2006: +17.021; 2007: -31.033). This coefficient is in all cases insignificant. Contrary to the previous regression model (the Mcap/SE-model), the compensation variable has positive values (pooled: +10.501; 2006: +5.363; 2007: +21.060), but these coefficients are insignificant in all cases.
Results based on accounting performance (ROA)
19
Sensitivity analysis
20 Table V
OLS Regression results of Marketcap/SE-ratio (method 2)
This table shows the results of the OLS regressions (method 2) of Marketcap/SE-ratio (dependent variable) on Corporate Governance variables and Control variables. For method 2, the input governance data is different compared to the first method. First, the average compliance rate for all four governance variables is calculated. Then, firms score a value of “1” when their compliance rate is above the average compliance rate and a “0” when this is not the case. The pooled results are the results based on the OLS regression of the pooled data. The results for 2006 and 2007 are the results based on the cross-sectional OLS regression of the individual yearly data. P-values of coefficients are in parentheses.
Pooled 2006 2007
R-squared 0.314 0.289 0.437
Adj. R-squared 0.237 0.107 0.293
Independent variable coefficient coefficient coefficient
(p-value) (p-value) (p-value)
DUALITY_2 0.504 -0.273 1.413 (0.738) (0.914) (0.439) COMPOSITION_2 0.380 0.698 0.028 (0.467) (0.428) (0.965) SHAREHOLDER_2 2.189 3.414 0.855 (0.243) (0.268) (0.715) COMPENSATION_2 -1.881 -1.964 -1.904 (0.013)* (0.145) (0.034)* GEARING 0.001 0.001 0.001 (0.741) (0.890) (0.776) SALES -0.033 -0.094 0.062 (0.732) (0.538) (0.614) RETAIL 3.645 4.194 2.738 (0.084) (0.208) (0.322) MANUFACTURE 1.539 1.987 0.730 (0.446) (0.532) (0.786) CONSTRUCTION 2.031 2.506 1.175 (0.338) (0.451) (0.677) REAL ESTATE 1.202 2.137 -0.174 (0.491) (0.440) (0.940) SERVICE 2.563 3.272 1.484 (0.202) (0.302) (0.575) OTHER 3.022 3.458 2.200 (0.139) (0.284) (0.414)
21 Table VI
OLS Regression results of Return on equity (method 2)
This table shows the results of the OLS regressions (method 2) of Return on equity (dependent variable) on Corporate Governance variables and Control variables. For method 2, the input governance data is different compared to the first method. First, the average compliance rate for all four governance variables is calculated. Then, firms score a value of “1” when their compliance rate is above the average compliance rate and a “0” when this is not the case. The pooled results are the results based on the OLS regression of the pooled data. The results for 2006 and 2007 are the results based on the cross-sectional OLS regression of the individual yearly data. P-values of coefficients are in parentheses.
Pooled 2006 2007
R-squared 0.535 0.596 0.619
Adj. R-squared 0.484 0.495 0.523
Independent variable coefficient coefficient coefficient
(p-value) (p-value) (p-value)
DUALITY_2 -36.950 -53.194 -25.446 (0.013)* (0.019)* (0.181) COMPOSITION_2 10.381 22.502 -1.967 (0.044)* (0.005)** (0.765) SHAREHOLDER_2 2.964 12.147 3.745 (0.871) (0.650) (0.878) COMPENSATION_2 7.893 8.334 12.723 (0.248) (0.446) (0.138) GEARING -0.107 -0.121 -0.050 (0.000)** (0.001)** (0.145) SALES 5.919 4.681 7.648 (0.000)** (0.001)** (0.000)** RETAIL -25.931 -15.454 -57.173 (0.205) (0.590) (0.051) MANUFACTURE -41.829 -25.428 -79.450 (0.036)* (0.357) (0.006)** CONSTRUCTION -29.268 -10.853 -70.669 (0.159) (0.707) (0.020)* REAL ESTATE -24.692 -11.653 -57.015 (0.149) (0.627) (0.022)* SERVICE -27.688 -15.260 -60.150 (0.157) (0.577) (0.033)* OTHER -15.636 0.599 -52.388 (0.432) (0.983) (0.067)
22 Table VII
OLS Regression results of Return on assets (method 2)
This table shows the results of the OLS regressions (method 2) of Return on assets (dependent variable) on Corporate Governance variables and Control variables. For method 2, the input governance data is different compared to the first method. First, the average compliance rate for all four governance variables is calculated. Then, firms score a value of “1” when their compliance rate is above the average compliance rate and a “0” when this is not the case. The pooled results are the results based on the OLS regression of the pooled data. The results for 2006 and 2007 are the results based on the cross-sectional OLS regression of the individual yearly data. P-values of coefficients are in parentheses.
Pooled 2006 2007
R-squared 0.510 0.462 0.599
Adj. R-squared 0.456 0.327 0.499
Independent variable coefficient coefficient coefficient
(p-value) (p-value) (p-value)
DUALITY_2 -11.882 -16.546 -7.351 (0.064) (0.131) (0.367) COMPOSITION_2 1.303 4.658 -1.847 (0.556) (0.219) (0.514) SHAREHOLDER_2 2.608 6.054 -2.029 (0.743) (0.645) (0.847) COMPENSATION_2 4.785 5.622 4.167 (0.108) (0.298) (0.256) GEARING -0.049 -0.043 -0.056 (0.000)** (0.011)* (0.000)** SALES 1.766 1.710 1.771 (0.000)** (0.010)** (0.002)** RETAIL -0.529 -2.855 3.094 (0.952) (0.839) (0.802) MANUFACTURE -10.731 -13.198 -6.583 (0.211) (0.332) (0.584) CONSTRUCTION -9.469 -11.396 -5.935 (0.293) (0.423) (0.639) REAL ESTATE -3.632 -5.063 -0.905 (0.624) (0.668) (0.930) SERVICE -6.292 -9.205 -1.801 (0.458) (0.494) (0.879) OTHER 1.478 -0.744 5.407 (0.864) (0.957) (0.654)
23 Table VI shows the results from the OLS regression equation linking corporate governance (based on method 2) and accounting performance based on return on equity. It appears that the fit of the data (measured by R-squared and adjusted R-squared) is more or less the same for both methods. Duality had very negative coefficients (pooled: -72.881; 2006: -113.221, 2007: -34.877) at the first method and these coefficients are also negative at the second method (pooled: 36.950; 2006: -53.194; 2007: -25.446). Furthermore, similar to the first method, the coefficient for the pooled data is statistically significant at the 5% level. The composition variable also shows similar results for both methods. The pooled coefficients are positive (method 1: +29.384; method 2: +10.381). Furthermore, the pooled coefficients are statistically significant at the 5% level for both methods. The shareholder variable shows some different results. For the first method, the pooled coefficient is negative (-11.691) and for the second method, this coefficient is positive (+2.964). Furthermore, the coefficients are not statistically significant. The compensation variable shows some more stable results. For both methods, the coefficients are positive. However, the coefficients are statistically insignificant for both methods.
24 method 2: +4.785) for both methods. However, the coefficients are not statistically significant for both methods.
Summarizing, two methods are used for estimating the regression models. The first method used the compliance rates as input data (for example: a score of 0.85 for firm X on governance variable Y). For the second method, a firm could a score a “0” (if it scores lower than the average of the sample for a particular governance variable) or a “1” (if it scores higher than the average of the sample for a particular governance variable). The significant outcomes were the following. Compensation has a negative effect on market valuation. Furthermore, duality has a negative effect on return on equity. On the other hand, composition has a positive effect on return on equity. For the return on assets models, shareholder rights have a negative effect on return on assets. However, this outcome is not statistically significant for all methods. The same holds for the negative effect of duality on return on assets. This is statistically significant for some methods but not for all.
Conclusions
The hypotheses at the beginning of this study were the following:
H0a: There is a significant relationship between CEO-duality and firm performance.
H0b: There is a significant relationship between board composition and firm performance.
H0c: There is a significant relationship between shareholder rights and firm performance.
H0d: There is a significant relationship between compensation and firm performance.
25 appears that the CEO-duality variable has a significant negative effect on return on equity. This means that by having good corporate governance provisions in place, return on equity will be lower on average. For the return on assets, this effect is also negative. This result is a little bit in conflict with previous empirical findings. Most authors found no clear link between role duality and accounting performance. Baliga, Moyer & Rao (1996), for example, investigated duality status changes and suggests that the market is indifferent to changes in a firm’s duality status and that there is little evidence of operating performance changes around changes in duality status. However, other authors like Rhoades et al (2001) found that firms with a separation of the two roles consistently have higher accounting returns. Therefore, this result seems to be conflicting with these previous findings. For the market valuation (measured by marketcap/SE), having good provisions concerning CEO-duality pays off. Those firms are valued higher in the market on average. However, this effect is not statistically significant. This result seems to be more in accordance with previous empirical research. For example, Hanifa & Hudaib (2006) state that companies with role duality seem not to perform as well as their counterparts with separate board leadership (for market valuation and ROA). Linking these results to the agency/stewardship-theory discussion, it appears that the market believes that there are significant agency costs when the function of CEO and chairman is not separated. Because it awards firms for having a good governance structure on CEO-duality. While in fact actual (accounting) performance is higher for firms where there is a less “good” structure on CEO-duality. This accounting result is in conflict with the research by Fama & Jensen (1983) and Jensen (1993), who state that the agency problem may be eliminated by separating management and supervising functions.
non-26 executive directors do not seem to affect performance regardless of the measures used. Kiel & Nicholson (2003) found for Australian firms a positive relationship between the proportion of inside directors and the market-based measure of firm performance. In the Netherlands, Van Ees, Postma & Sterken (2003) found a negative relationship between the number of outsiders on the board and performance in the Netherlands. This last result seems to be in contrary with the result of this study, because the Tabaksblat code prescribes that there only should be one insider on the board maximally. Unfortunately, because of these different definitions about the insider/outsider dilemma, a direct comparison between both studies can’t be made. Furthermore, Van Ees, Postma & Sterken (2003) used a different time frame (1996), so future research should clarify how board composition influences performance. However, I provide some strong statistical evidence that good (according to the code Tabaksblat) board composition has a positive relationship with firm performance.
Shareholder rights seems to have no effect on firm performance. The results of this variable are mixed and are in none of the cases significant. In the empirical literature, shareholders’ rights are associated positively with firm performance. Gombers, Ishii & Metrick (2003) used an investment strategy model that bought firms with the strongest shareholders’ rights and sold firms with the weakest shareholders’ rights and found that this strategy earned an abnormal return of 8.5 percentage per year. They also found that firms with stronger shareholder rights had higher firm value, higher profits, higher sales growth, lower capital expenditures, and made fewer corporate acquisitions. Furthermore, Core, Guay & Rusticus (2006) found that firms with weak shareholders rights exhibit significant operating underperformance. However, I found no significant results that shareholders’ rights have a relationship with firm performance in the Netherlands. An explanation for this can be that shareholders’ rights are already very well incorporated in Dutch law, so the best practice provisions in the Dutch code Tabaksblat are somewhat redundant.
27 can be showed in the accounting returns that firms who pay their managers/board members in accordance to the best practice provisions, score higher accounting returns. In the empirical literature, Core, Holthausen & Larcker (1999) suggest that firms with weaker governance structures have greater agency problems and that CEO’s at firms with greater agency problems receive greater compensation. Furthermore, firms with greater agency problems perform worse. Their performance is based both on stock returns and on ROA. So maybe there is a agency problem for firms that score low on the compensation best practices in my study. However, this is not statistically significant. On the other hand, the market values firms higher that score low on the compensation best practices and this effect is statistically significant.
Summarizing, this study brings some interesting results. Firstly, separation of the CEO and chairman function has a significant negative effect on return on equity. Secondly, “good” board composition (according to the code Tabaksblat) has a significant positive relationship with accounting performance (return on equity). Thirdly, shareholder rights seems to have no effect on firm performance. Lastly, “good” compensation (according to the code Tabaksblat) has a significant negative relationship with market valuation, while it has a positive (non-significant) relationship with accounting performance.
This study could suffer from some limitations. Firstly, there was relative small number of firms in the data, this could affect the generalizability of the results. Secondly, also data related, there is only data used for the years 2006 en 2007, so it is unknown if the results hold for other time periods. Furthermore, there could be a time lag between performance and corporate governance, this study did not took account for such time lagging effects. Future research should clarify this. Lastly, corporate governance is measured based on the code Tabaksblat. In this study it is assumed that the code prescribes “good governance”. However, it remains unclear that this is the case. Furthermore, because the code is based on self-description, firms could window dress.
28 compensation structure really has a negative effect on market valuation and a positive effect on accounting performance.
References
Akkermans, D.H.M., Ees, H. van, Hooghiemstra, R., Laan, G. van der, Hermes, N.L.M., Postma, T.J.B.M. and Witteloostuijn, A. van (2007), “Corporate
Governance in the Netherlands: an overview of the application of the Tabaksblat Code in 2004”, Corporate Governance: An International Review, vol. 15, nr. 6, pp. 1109-1118.
Baliga, B.R., Moyer, R.C. & Rao, R.S. (1996), “CEO Duality and Firm
Performance: What’s the Fuss?”, Strategic Management Journal, vol. 17, nr. 1, pp. 41-53.
Bauer, R., Guenster, N. & Otten, R. (2004), “Empirical evidence on corporate governance in Europe: The effect on stock return, firm value and performance”,
Journal of Asset Management, vol. 5, nr. 2, pp. 91-104.
Berle, A.A. & Means, G. (1932), “The Modern Corporation and Private Property”, Macmillan, New York.
Bertrand, M. & Mullainathan, S. (2000), “Agent with and without Principals”,
American Economic Review, vol. 90, nr. 2, pp. 306-308.
Boyd, B.K. (1995), “CEO Duality and Firm Performance: A Contingency Model”,
Strategic Management Journal, vol. 16, nr. 4, pp. 301-312.
Core, J.E., Guay, W.R. & Rusticus, T.O. (2006), “Does Weak Governance Cause Weak Stock Returns? An Examination of Firm Operating Performance and
29 Core, J.E., Holthausen, R.W. & Larcker, D.F. (1999), “Corporate Governance, Chief Executive Officer Compensation, and Firm Performance”, Journal of
Financial Economics, vol. 51, nr. 3, pp. 371-406.
Dalton, D.R., Daily, C.M., Johnson, J.L., Ellstrand, A.E. (1999), “Number of
Directors and Financial Performance: A Meta-Analysis”, Academy of Management
Journal, vol. 42, nr. 6, pp. 674-686.
Davis, J.H., Schoorman, F.D. & Donaldson, L. (1997), “Toward a Stewardship Theory of Management”, Academy of Management Review, vol. 22, nr. 1, pp. 20-47.
Donaldson, L. & Davis, J.H. (1991), “Stewardship Theory or Agency Theory: CEO Governance and Shareholder Returns”, Australian Journal of Management, vol. 16, nr. 1, pp. 49-65.
Van Ees, H., Postma, T.J.B.M. & Sterken, E. (2003), “Board Characteristics and Corporate Performance in the Netherlands”, Eastern Economic Journal, vol. 29, nr. 1, pp. 41-58.
Fama, E.F., & Jensen, M.C. (1983), “Separation of Ownership and Control”,
Journal of Law & Economics, vol. 26, nr. 2, pp. 301-326.
Finkelstein, S. & D’Aveni, R.A. (1994), “CEO Duality as a Double-Edged Sword: How Boards of Directors Balance Entrenchment Avoidance and Unity of
Command”, Academy of Management Journal, vol. 37, nr. 5, pp. 1079-1108.
Gombers, A., Ishii, J.L. and Metrick, A. (2003), “Corporate Governance and Equity Prices”, Quartely Journal of Economics, 118, pp. 107-155
30 Haniffa, R. & Hudaib, M. (2006), “Corporate Governance Structure and
Performance of Malaysian Listed Companies”, Journal of Business Finance and
Accounting, vol. 33, nr. 7/8, pp. 1034-1062.
Jensen, M.C. & Meckling, W.H. (1976), “Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure”, Journal of Financial Economics, vol. 3, nr. 4, pp. 305-360.
Jensen, M.C. (1993), “The Modern Industrial Revolution, Exit, and the Failure of Internal Control Systems”, Journal of Finance, vol. 48, nr. 3, pp. 831-880.
Kiel, G.C. & Nicholson, G.J. (2003), “Board Composition and Corporate
Performance: How the Australian Experience Informs Contrasting Theories of Corporate Governance”, Corporate Governance: An International Review, vol. 11, nr. 3, pp. 189-205.
Kyereboah-Coleman, A., Adjasi, C.K.D. & Abor, J. (2006), “Corporate
Governance and Firm Performance: Evidence from Ghanaian Listed Companies”,
Corporate Ownership & Control, vol. 4, nr. 2, pp. 123-132.
Larcker, D.F., Richardson, S.A. & Tuna, I. (2007), “Corporate Governance,
Accounting Outcomes, and Organizational Performance”, The Accounting Review, vol. 82, nr. 4, pp. 963-1008.
Rhoades, D.L., Rechner, P.L. & Sundaramurthy, C. (2000), “Board Composition and Financial Performance: A Meta-Analysis of the Influence of Outside
Directors”, Journal of Managerial Issues, vol. 12, nr. 1, pp. 76-91.
Schleifer, A. & Vishny, R.W. (1997), “A Survey of Corporate Governance”,
Journal of Finance, vol. 52, nr. 2, pp. 737-783.
