Effects of the Introduction of the Dutch Corporate Governance Codes of 2003 and 2009 on Dutch Stock Returns

(1)

Effects of the Introduction of the Dutch Corporate Governance

Codes of 2003 and 2009 on Dutch Stock Returns

ANTON H.G. SPRANGERS* Student number: s1555553

ABSTRACT

To investigate whether the introduction of the Dutch corporate governance codes affected Dutch stock returns, I apply event study methodology to thirty-two AEX and AMX listed companies at the Dutch stock exchange and to the AEX index as a whole. Next, I analyze whether there are differences between companies in the impact of the codes on stock returns, by regressing the cross-section of abnormal returns on five key firm-specific covariates. There is presumptive evidence that the codes caused positive abnormal returns. The key characteristics of companies do not influence the impact of the introduction of the codes on stock returns.

(2)

THE SEPARATION OF OWNERSHIP AND CONTROL of investors‟ capital leads to agency problems: how do suppliers of capital make sure that they get compensated for their investment (Shleifer & Vishny, 1996 and De Jong et al., 2005). These agency problems are mitigated by corporate governance codes. These codes should make it easier for companies to raise capital in financial markets (De Jong et al., 2005).

In December 2003 the Dutch Corporate Governance Committee (Committee Tabaksblat), headed by Morris Tabaksblat, published a new corporate governance code (hereafter: Code Tabaksblat) to improve the corporate governance of companies publicly traded in the Netherlands or with a statutory residence in the Netherlands (Akkermans et al., 2007). The committee was composed by employers‟ and shareholders‟ associations and the Dutch stock exchange, at the invitation of the Ministers of Finance and Economic Affairs (Corporate Governance Committee, 2003). On 1 January 2009 a revised code (hereafter: Code Frijns) by the Corporate Governance Monitoring Committee (Monitoring Committee), headed by Jean Frijns, came into force.

According to the mission statement of Committee Tabaksblat, the mission of the Dutch corporate governance code is, in line with Aguilera & Cuervo-Cazurra (2004), to improve the confidence that suppliers of capital have in the governance of publicly listed companies. This study investigates whether Code Tabaksblat and Code Frijns fulfilled that mission, by analyzing abnormal stock returns around key event dates.

To investigate whether the introduction of the codes affected Dutch stock returns this article applies event study methodology. First, the methodology is applied on thirty-two listed companies at the Dutch stock exchange as well as on the AEX index as a whole. Second, I analyze whether there are differences between companies in the impact of the codes on stock returns, by regressing the cross-section of abnormal returns on five key firm-specific covariates: industry category, market capitalization, leverage, liquidity, and the number of employees.

(3)

investors‟ confidence in good governance of companies (Corporate Governance Committee, 2003). Likewise, Shleifer & Vishny (1996) argue in a theoretical approach that corporate governance codes are needed. They explain that since production capital is highly specific and sunk, the suppliers of capital need to be protected against managers expropriating the return of capital after the capital provided by investors is sunk.1

How do corporate governance codes protect then suppliers of capital against expropriation by managers? In general, corporate governance codes describe “best practice provisions with respect to management, supervision, disclosure and auditing” (Akkermans et al., 2007). Code Tabaksblat contains for example recommendations on the tasks, procedures, and remuneration and composition of the directors, on the disclosure of internal control mechanisms, and it strengthens the positions of the annual meeting of shareholders and the accountant. The key element of Code Tabaksblat and Code Frijns is that the best practice provisions outlined in the codes are supposed to be applied by companies through self-enforcement (Corporate Governance Committee, 2003). This includes that when a company does not apply a provision of the code, it has to explain in its annual report why it does not. The codes also state that if the general meeting of shareholders approves such a deviation from the code, the company in effect also complies with the code (Corporate Governance Committee, 2003).

Corporate governance codes differ between countries. For example, the bankruptcy of Enron in 2001 led directly to the establishment of the U.S. Sarbanes-Oxley Act (SOX) in July 2002 (Zhang, 2007). One of the major differences between the „apply or explain‟ Dutch corporate governance code and its U.S. counterpart, the SOX, is that the SOX provisions are obligatory. One can question whether „apply or explain‟ regulations are able to change corporate governance characteristics of companies as well as obligatory regulations can do.

Code Tabaksblat was preceded by recommendations on corporate governance by Committee Peters. These recommendations were also supposed to be applied by self-regulation. Compliance was voluntary instead of required. De Jong et al. (2005) evaluate the impact of these recommendations on stock returns. The authors conclude that this self-regulation initiative was not successful, because the recommendations were not applied significantly,

1

(4)

since outsiders did not enforce the recommendations. Since the subsequent Dutch corporate governance codes are also based on self-regulation, this paper questions their effectiveness. But if Code Tabaksblat and Code Frijns (hereafter: the Dutch codes) increased investors‟ confidence as they were supposed to – decreased apparent risk of accounting scandals –, the Dutch codes should have caused positive abnormal stock returns. As opposed to possible positive abnormal returns, the introduction of the Dutch codes might also have resulted in negative abnormal returns, because prior research shows that new corporate governance legislature can also impose net (compliance) costs on companies (Zhang, 2007). This leads to the following main research question: Have the Dutch corporate governance codes of 2003 and 2009 changed shareholder value of Dutch publicly listed firms?

Corporate governance is an important subject, because also in developed economies there is a lot of discussion on how governance mechanisms should work. In less developed economies there are almost no governance mechanisms. Comprehension of the workings of governance mechanisms might not only improve minimally corporate governance of developed economies, but it might improve important institutional changes in less developed economies as well (Shleifer & Vishny, 1996).

The first section of the article contains related research, builds the hypotheses underlying the research of this possible economic value change, and presents an overview of the event history of the Dutch corporate governance code. The second and third sections show the event study methodology and data and descriptive statistics respectively. Section four and five contain the results and conclusions of the research. This is followed by the references and the appendices in section six and seven respectively.

I. Related Research and Hypotheses and Event History of the Dutch Corporate Governance Code

A. Related Research and Hypotheses

This subsection elaborates further on the fundamentals of corporate governance codes: why are they needed and how are the Dutch codes supposed to meet this need. Then this subsection builds the hypotheses. The subsection is followed by an event history of the Dutch codes.

(5)

Shleifer & Vishny (1996) explain that, in theory, there are two approaches in which investors are willing to supply capital even when they do not get any control rights in return for it. The similarity between the two approaches is that suppliers of capital simply hope that they will make good returns. The first approach is that investors are willing to supply money because they think that companies cannot afford to damage their reputations, so companies will always take care of good returns on investment. The second approach is that suppliers of capital can show excessive optimism about future returns (supplying capital to a pyramid scheme is an extreme example). Still, the authors do not think that suppliers of capital are willing, in general, to supply a lot of capital in exchange for stocks or loans that might be worth nothing in reality. They argue that suppliers of capital must get control rights in return for their supplies. They state that control rights are the main reason that suppliers of capital are willing to invest.

Corporate governance codes are issued to compensate for shortcomings in corporate governance mechanisms with respect to control rights, in particular with respect to shareholders‟ rights (Aguilera & Cuervo-Cazurra, 2004). In a perfect world a contract between suppliers of capital and management could be complete, but such a contract is not possible in the real world, because not all possible states of the future are known beforehand (Shleifer & Vishny, 1996). Corporate governance codes compensate for these shortcomings in control rights by improving the quality and transparency of management (Akkermans et al., 2007). As a result they improve the performance of companies and increase the confidence of investors in management (Akkermans et al., 2007).

Aguilera and Cuervo-Cazurra (2004) argue that countries and companies are stimulated to apply corporate governance codes based on internationally accepted best practices, since that makes them more attractive for foreign investors. The principals and best practice provisions of Code Tabaksblat and Code Frijns also represent national and international best practices (Corporate Governance Committee, 2003; Corporate Governance Code Monitoring Committee, 2009).

(6)

chapter 5 contains provisions for the external auditor (Corporate Governance Committee, 2003; Corporate Governance Code Monitoring Committee, 2009).

Akkermans et al. (2007) investigate the compliance with the provisions of Code Tabaksblat. They conclude that - the basic question - whether Code Tabaksblat changed the behavior of corporate management, consists of the following three sub-questions: the meaning of the compliance found, the meaning of non-compliance found, and what the results indicate about the content of Code Tabaksblat. They conclude one year after the implementation of Code Tabaksblat that the extent of compliance increased since the code was enacted, but they also notice that the „apply or explain‟ mechanism might not be effective since it is possible for countries to comply in a symbolic manner instead of genuinely. Regarding the third question they argue, first, that as a result of the ambiguities of the code about 20% of the provisions cannot be evaluated based on publicly available information. Second, the complexity of the code induces camouflage. They end with that to investigate the real impact of Code Tabaksblat on company behavior a longer term perspective is needed (Akkermans et al., 2007).

De Jong et al. (2005) support Akkermans et al.‟s (2007) conclusions that a self-enforcement „apply or explain‟ mechanism might not be effective. De Jong et al. (2005) do not find any positive stock return effects due to actions by the Peters Committee. They conclude that their event study results indicate that the financial market did not expect corporate governance practices in The Netherlands to change. There were negative abnormal stock returns around the release of the Peters Committee monitoring report. This report investigated whether Dutch companies applied the recommendations of Code Peters. De Jong et al. (2005) think that the financial market was disappointed with that Dutch companies were not changing their corporate governance practices as recommended by Code Peters. The authors conclude that self-regulation of corporate governance practices relying on market forces is not enough to stimulate changes that increase shareholder value.

(7)

events reducing the power of the test. She concludes that her results cannot be indisputably interpreted as evidence of SOX being costly. Expected costs of future anti-business regulation, other news, and possible investors‟ disappointment with SOX might also have influenced the abnormal returns. And, not only SOX provisions are correlated, the characteristics of companies in the cross-sectional analysis are also correlated (Zhang, 2007). Since prior researches indicate that investors‟ confidence in a corporate governance change depends on whether it is obligatory to comply with the provisions of a code, and the Dutch codes are not obligatory, I present the following hypothesis:

Hypothesis 1: The introduction of the Dutch codes did not lead to (cumulative) abnormal returns.

A.2. The Relationship Between Abnormal Returns and Industry Category, Market Capitalization, Leverage, Liquidity, and the Number of Employees

Next, I investigate whether the reaction of stock returns differs between companies with different key characteristics. To do so, I regress the abnormal returns of the event study on a number of key company characteristics. In light of this analysis, I present six more hypotheses on industry category, market capitalization, leverage, liquidity, and the number of employees. Bebchuck, Cohen and Ferrell (2009) state that it is possible that corporate governance provisions reflect the industry of the company, since companies do not adopt provisions randomly (Gompers, Ishii, and Metrick, 2003). Since the AEX index is a value weighted index that consists of companies from different industries, I want to control for industry effects by introducing hypothesis 2:

Hypothesis 2: The abnormal returns of companies’ values resulting from the introduction of the Dutch codes do not differ between industries.

(8)

abnormal returns would be detected if the financial market thinks that the benefits would exceed the compliance costs. In contrast to the general results of Zhang (2007) their event study shows that these financial services companies benefited significantly from the regulations, except for securities companies (e.g. securities brokerage and investment banking companies). They conclude, first, that these results may be ascribed to expected improvement in the transparency of the financial services companies, because these companies have a relatively poor image. Second, they also conclude that their results are in support of their compliance costs hypothesis (Akhigbe & Martin, 2006).

Taking into account that Akkermans et al. (2007) argues that Code Tabaksblat offers camouflage opportunities, and that the Dutch code is a self-enforcement „comply or explain‟ code, you would not expect Dutch financials to experience similar negative cumulative abnormal returns. But since the image of Dutch financials is also poor, especially after the credit crisis starting in the summer of 2007 and the nationalization of ABN AMRO, shareholders might have also benefited from the new corporate governance codes.

This ambivalence makes it hard to predict the market reaction with respect to Dutch financials. This leads to the following hypotheses:

Hypothesis 3: The introduction of the Dutch codes did not lead to cumulative abnormal returns in the case of financial services companies.

To build their hypotheses, Akhigbe & Martin (2006) refer also to Jorgensen & Kirschenheiter (2003) who argue that in risk disclosures the value of companies is higher for companies that disclose their risk voluntarily, and lower for companies that do not want to disclose their risk but are required to. Diamond & Verrechia (1991) show that disclosing public information to lessen information asymmetry can reduce a company‟s cost of capital. The mechanism behind this reduction is that such disclosures reduce security‟s risk. Eng & Mak‟s (2003) research confirms that larger companies have greater disclosure. Large companies will reveal more information because they gain most. Nevertheless, the authors also find situations where diminished information asymmetry increases the cost of capital, because diminished information asymmetry makes traders (that quote both a buy and a sell price to make money with the spread) to leave (Diamond & Verrechia, 1991).

(9)

reduction recommendations will have less effect on larger companies than on smaller companies. This leads to under mentioned hypothesis.

Hypothesis 4: The Dutch codes will have less effect on larger companies than on smaller companies.

According to Eng & Mak‟s (2003) article, companies with more debt reveal less information than lower leveraged companies. Jensen (1986) argues that debt is a mechanism to control the free cash flow problem, thus debt reduces the need for disclosure (Eng & Mak, 2003). As a result you would expect, that if the codes were effective, higher leveraged companies react more to the Dutch codes than lower leveraged ones. In contrast with Eng & Mak (2003) are the findings of Sengupta (1998) that companies with high disclosure quality experience lower effective interest costs when raising debt. This ambivalence does not predict clearly what the relation between debt and the introduction of the codes is. This leads to hypothesis 5:

Hypothesis 5: Abnormal returns of companies as a result of introduction of the Dutch codes are not dependent on the leverage level of companies.

Harford et al. (2008) find that cash reserves - as a percentage of total sales - of companies with weaker corporate governance mechanisms are smaller. They find that companies with weaker governance mechanisms are more likely to repurchase stocks than raising dividends when they want to distribute cash to shareholders. They do this to avoid future payout obligations. Therefore you expect that companies with more cash reserves react less to new corporate governance recommendations than companies with lower cash reserves. This leads to the following hypothesis:

Hypothesis 6: Companies with more cash reserves experience less abnormal returns as a result of the introduction of the Dutch codes than companies with less cash reserves.

(10)

2003) Strong representation rights make provision for influence of employees on the behavior of management. I expect representation rights to be more formally integrated (e.g. works councils) when companies have more employees and these formal channels to make management more susceptible to other stakeholders‟ interests. As a result of increased susceptibility I expect corporate governance structures to be more in line with recommendations proposed by corporate governance committees. This leads to hypothesis 7: Hypothesis 7: Companies with more employees will have less abnormal returns as a result of the introduction of the Dutch codes.

The next subsection presents an overview of the history of the Dutch corporate governance code. The key events of its history are selected to calculate the abnormal returns associated with the introduction of the codes.

B. Event History

Code Tabaksblat became effective on 1 January 2004. The code was preceded by corporate governance recommendations of the Peters Committee in the Netherlands, after a debate that started in 1997, and various similar national codes in other countries. The compliance of the recommendations of the Peters Committee were evaluated and turned out to be unsuccessful (Akkermans, 2007). On 18 December 2002 the Dutch Minister of Finance, Mr. Hoogervorst ascertained publicly that Dutch companies had had too little progress in the field of corporate governance. The investigated publicly listed companies would not act proactively on corporate governance and would particularly react to legislation changes (Het Financieele Dagblad, 2002).

(11)

(Het Financieele Dagblad, 2003b). The starting point of the committee were the recommendations of the Peters Committee.

On 1 July 2003 the draft version of code Tabaksblat was published. In general, the code was kindly received. The comments of politicians, business, and labour unions were various (Het Financieele Dagblad, 2003c). After an evaluation of the different comments, the committee published the final code on 9 december 2003. The reaction of the different interest groups and politicians on the final code was predominantly positive. The parties realized that the success of the code was to be determined by the application of it by all different parties (Het Financieele Dagblad, 2003d). The code was enacted on 1 January 2004. On 10 September 2004 strong rumours developed that Jean Frijns would become the successor of Morris Tabaksblat. Jean Frijns was the director investments of pension fund ABP. The new Monitoring Committee would monitor how companies complied with Code Tabaksblat. If in general companies had compliance problems, the Monitoring Committee would propose new provisions (Het Financieele Dagblad, 2004).

(12)

On 4 June 2008 the Monitoring Committee published for the first time a proposal with changes of Code Tabaksblat The Corporate Governance Code Monitoring Committee (2008b). On 4 December 2008 Het Financieele Dagblad publishes from the leaked out corporate governance code which had to replace Code Tabaksblat. It found it remarkable that the section on risk would not change very much, even with the outbreak of the credit crisis fresh in mind. One of the major proposed changes was on the remuneration of managers. Remuneration had to be linked to long term goals and bonuses had to become in proportion with fixed salaries (Het Financieele Dagblad, 2008). Code Frijns was officially published on 10 December 2008. On 1 January 2009 Code Frijns officially replaced Code Tabaksblat (Corporate Governance Code Monitoring Committee, 2009).

The above mentioned event dates and other key event dates are listed in table I. The next section of the paper contains the methodology of this article.

II. Methodology

The methodology is based on the event studies of Zhang (2007), Brown & Warner (1985), and MacKinlay (1997), and followed by a cross-sectional analysis of abnormal returns. The paper of Zhang (2007) investigates the changes of the U.S. market index around legislative events, since the changes affect every listed U.S. firm. Zhang controls for normal U.S. market returns by changes of 22 major developed foreign stock markets. I compare the returns of the Dutch AEX index and AEX and AMX stocks with the FTSE 100, DAX, CAC 40, IBEX-35, and FTSE MIB (S&P/MIB prior to June 2009). I choose these foreign indices because the time zone differences are negligible. Due to their geographical location and integration of the European financial market, the foreign indices are influenced by much of the same economic news as Dutch firms. Datastream (Thomson Reuters, 2010) provides the returns of the indices.

In line with the paper of Zhang (2007), the following regression models the relation between Dutch and foreign market returns:

(1) AEXt, FTSEt, DAXt, CACt, IBEXt and MIBt represent returns of the markets on day t.

(13)

prior to 17/12/2002. I use the OLS estimation results to determine the parameters of the equation. With the fitted parameters at hand, I calculate AEX “normal returns” in the event period in absence of the code. For example (if only the DAX is used):

(2) The abnormal returns are then calculated as the prediction errors (Zhang, 2007):

(3) T runs from day 1 to 100.

In a similar way I calculate the abnormal returns for the prediction errors of the thirty-two individual stocks in the estimation window by using the following formula:

(4)

represents the returns of the individual stock i on day t. I runs from company 1 to 32 and t

runs from day 1 to 100.

I calculate the normalized abnormal returns by dividing the abnormal returns from equation (4) by the standard deviation of the residuals belonging to equation (4). Normalized returns assign less weight to volatile stocks, so that abnormal returns originating from high returns of highly volatile stocks have a smaller influence (Edmans, García, and Norli, 2007).

Then, I calculate aggregated average abnormal returns ( ): the average of the average abnormal return of each event window day for all fifteen events. I do this for the abnormal and normalized returns of all 32 stocks combined, the AEX index, and the financials separately.

After calculating the (normalized) abnormal and cumulative [-5, 5] abnormal returns the Student‟s t-test shows whether the (normalized) (cumulative) abnormal returns are significant (Brown & Warner, 1985; Zhang, 2007).

(14)

(LIQ), and the number of employees (EMP). The influence of these variables on the abnormal returns on the event day and the day after is calculated by equation (5):

(5)

T [0] and T [+1] for all thirty-two stocks for all fifteen events.

The event windows of the firms overlap perfectly. According to Brown & Warner (1985), clustering in event dates do not drastically change the results of event studies.

The next section presents the data and the descriptive statistics of the samples.

III. Data and Descriptive Statistics

I examine the market returns around the following fifteen key events (table I): Table I

Event Dates

The key event dates belonging to the introductions of Code Tabaksblat and Code Frijns (Het Financieele Dagblad, 2002 until 2009, Corporate Governance Committee, 2003, and Corporate Governance Code Monitoring Committee, 2005 until 2009).

Date Event

18/12/2002 1. Outgoing Minister of Finance declares the desire for a new code

10/03/2003 2. Establishment of the Committee Tabaksblat

01/07/2003 3. Publication of the draft version of Code Tabaksblat

09/12/2003 4. Publication of the final Code Tabaksblat

01/01/2004 5. Enactment of the Code Tabaksblat

10/09/2004 6. Strong rumours on Frijns as the successor to Tabaksblat

05/10/2004 7. Publication of the first monitoring report on the compliance with the Code Tabaksblat (financial year 2003) by the Dutch Corporate Governance Foundation and Netherlands Insitute for Corporate Governance

(15)

20/12/2006 9. Publication of the second yearly evaluation report by the Monitoring Committee on the compliance with the Code Tabaksblat (financial year 2005)

30/05/2007 10. Publication of the advisory report by the Monitoring Committee to the Dutch government on the company-shareholder relationship and on the scope of the Code

19/12/2007 11. Publication of the third yearly evaluation report by the Monitoring Committee on the compliance with the Code Tabaksblat (financial year 2006)

04/06/2008 12. Publication of the evaluation report on the Code Tabaksblat by the Monitoring Committee (first proposal with changes of the code)

04/12/2008 13. Het Financieele Dagblad publishes from the leaked out Corporate Governance Code which has to replace Code Tabaksblat

10/12/2008 14. Publication of the fourth yearly evaluation report by the Monitoring Committee on the compliance with the Code Tabaksblat (financial year 2007) & publication of Code Frijns by the Monitoring Committee

01/01/2009 15. Enactment of Code Frijns

Datastream delivers the daily returns of the 25 AEX stocks, 25 AMX stocks and the FTSE 100, DAX, CAC 40 and FTSE MIB indices. It was not possible to download the IBEX-35 and AMX indices returns from Datastream. The estimation period consists of 100 days: from 30/7/2002 to 16/12/2002.

Because of delistings ten companies were removed from the AEX sample and eight from the AMX sample dated 30/07/2002 (appendices A & B). I do not expect these delistings to lead to a bias.

Different regression equations model the relation between the AEX index and the the FTSE 100, DAX, CAC 40 and FTSE MIB indices in the estimation period. These equations lead to the coefficients and adjusted R-squares listed in table II.

Table II

Estimation of the Market Model for the AEX Index with Respect to Other Indices

(16)

the independent variables of which the coefficients are given in the first four rows with their (t-statistics). Period: 30/7/2002 to 16/12/2002 ***, **, * indicate significance at 1%, 5% and 10% levels, respectively, two-tailed tests (Thomson Reuters, 2010).

Coefficients, constant and

adjusted R-squares

Equation 1 Equation 2 Equation 3 Equation 4 Equation 5 Equation 6

FTSE 100 0.293*** (3.226) 0.297*** (3.257) 1.344*** (24.183) DAX -0.054 (-1.093) 0.782*** (15.693) CAC 40 0.684*** (7.719) 0.657*** (7.713) 1.033*** (36.914) FTSE MIB 0.272*** (3.028) 0.232*** (2.824) 1.212*** (23.498) Constant -0.000*** (-0.153) -0.000 (-0.045) -0.000*** (-0.088) 0.000 (0.517) 0.000 (0.072) -0.000 (-0.345) Adj. R^2 0.944 0.943 0.855 0.712 0.932 0.848

The high adjusted R-squares of table II combined with the large changes in the coefficient values when adding or removing an explanatory variable indicates multicollinearity. Therefore table III lists the correlations between the different indices. They show that there is high correlation between the different indices indeed. The high correlations do not obstruct the study, since they do not reduce the predictive power of the model in total. Correlation only affects calculations of individual coefficients, however, what matters for an event study is the fit of the model, e.g. how well it can predict “normal returns”, and in an event study one is not interested in the estimated coefficients of the estimation window as such. I calculate the abnormal returns of the AEX index in the estimation window and its descriptive statistics using the FTSE 100, CAC 40 and FTSE MIB indices as independent variables. The coefficients and constant listed in the third column of table II are used to predict AEX normal returns in the estimation period.

Table III

(17)

Correlations of indices returns from 30/7/2002 to 16/12/2002 (Thomson Reuters, 2010).

AEX FTSE 100 FTSE MIB CAC 40 DAX

AEX 1 0.925 0.922 0.966 0.846

FTSE 100 0.925 1 0.872 0.923 0.811

FTSE MIB 0.922 0.872 1 0.923 0.881

CAC 40 0.966 0.923 0.923 1 0.873

DAX 0.846 0.811 0.881 0.873 1

The abnormal returns ( are then calculated as the prediction errors:

(6)

AEXt, FTSEt, CACt and MIBt represent the returns of the markets on day t. T runs from day 1

to 100.

According to table II the CAC 40 is the best individual predictor of the AEX index. Therefore I now calculate the abnormal returns as the prediction errors by using the following formula:

(7)

T runs from day 1 to 100. The third column of appendix C lists the descriptive statistics of these abnormal returns. The abnormal returns for both equations (6) and (7) are normally distributed as can be concluded by the Jarque-Bera statistics from appendix C. The CAC 40 is now also used to calculate the abnormal returns as the prediction errors of the thirty-two individual stocks in the estimation window by using the following formula:

(8)

represents the returns of the individual stock i on day t. I runs from company 1 to 32 and t

runs from day 1 to 100. appendix D lists the parameters of equation (8).

(18)

dependent variables. This results in average R-squares for the AEX and AMX stocks of 0.492 and 0.104 respectively. So I solely use the CAC 40 to calculate abnormal returns.

To investigate whether there are abnormal return differences between companies from different industries, I divide the companies in twelve industry categories, as listed in table IV. The last column of table IV lists the industry averages of from equation (8). Roughly, the

betas of table IV are as I expected. The betas of cyclical industries such as financials and technology (machinery, equipment) are high and the betas of non-cyclical industries like food, beverages and tobacco are low.

Table IV

The Industry Categories of the AEX and AMX Stocks of the Sample

The average industry parameters of the regression equations with the thirty-two AEX and AMX companies as dependent variables and the CAC 40 as independent variable. Period: 30/7/2002 to 16/12/2002. ***, **, * indicate average significance at 1%, 5% and 10% levels, respectively, two-tailed tests (Thomson Reuters, 2010).

Category Companies

1. Food, beverages, tobacco CSM, Heineken, Nutreco, Unilever, Wessanen 0.212

2. Wood, cork, paper Hunter Douglas 0.117*

3. Publishing, printing Wolters Kluwer, Reed Elsevier 0.436*** 4. Chemicals, oil, non-metallic

products

AkzoNobel, DSM, Shell 0.649***

5. Machinery, equipment ASM International, ASML, Océ, Philips 1.496*** 6. Construction BAM Groep, Boskalis Westminster, Heijmans,

Imtech

0.200*

7. Wholesale and retail trade Ahold, SDB (Super de Boer) 0.846***

8. Transport Vopak 0.206***

9. Post and telecommunications KPN, TNT 0.649***

10. Financials Aegon, ING en Van der Moolen 1.311***

11. Propery investment Corio, Vastned Retail, Wereldhave 0.072

(19)

Next, I calculate the correlations of the variables of the cross-section panels at the end of the estimation period, end 2002. Table V lists these correlations. Since there is no sign of multicollinearity equation (5) is expected to give a fair understanding of the influence of the individual variables on the abnormal returns. Table VI lists the belonging descriptive statistics.

Table V

The Correlations of the Variables of the Cross-sectional Covariates

The correlations between the natural logarithms of the following variables of the thirty-two AEX and AMX stocks: industry, number of employees, leverage in terms of total debt / total (shareholders’) equity, liquidity in terms of cash and cash equivalents / net sales, and market capitalization as of December 2002 (Thomson Reuters, 2010).

Industry Employees Leverage Liquidity Market cap. Industry 1 Employees -0.417 1 Leverage 0.129 0.222 1 Liquidity 0.280 -0.486 -0.037 1 Market cap. -0.255 0.525 0.201 -0.100 1 Table VI

Company Characteristics of Dutch Listed Firms as of 2002

The descriptive statistics of the natural logarithms of the variables belonging to the thirty-two AEX and AMX stocks: industry, number of employees, leverage in terms of total debt / total (shareholders’) equity, liquidity in terms of cash and cash equivalents / net sales, and market capitalization as of December 2002 (Thomson Reuters, 2010).

Industry Employees Leverage Liquidity Market cap.

(20)

Jarque-Bera 27.812 77.975 486.654 14.930 38.099

Probability 9.135 0 0 0.001 5.332e-09

Sum 2940 1938.022 -48.507 -506.555 3066.869

Sum Sq. Dev. 5902.5 390.968 210.559 133.566 248.765

Observations 480 465 480 435 480

The following section presents the research results. That section is followed by the conclusion and limitations of this research.

IV. Results

A. Abnormal Returns

This section presents a summary of the most important research results.2 First, the section starts with the tests on thirty-two listed companies at the Dutch stock exchange as well as on the AEX index as a whole. Second, subsection B shows whether there are differences between companies in the impact of the codes on stock returns, by regression results of the cross-section of abnormal returns on five key firm-specific covariates: industry category, market capitalization, leverage, liquidity, and the number of employees.

Of all fifteen events concerning the introduction of the Dutch codes, only event 2 and 14 lead to abnormal returns (on p-value levels of 0.1 and 0.05 with normalized values). Figure 1

2

(21)

shows the t-values of the (normalized) average abnormal returns of all thirty-two stocks for all 15 events resulting from equation (7). The critical t-values are 1.76 value 0.1) and 2.14 (p-value 0.05). Event 2 is the establishment of Committee Tabaksblat on 10 March 2003. The negative response to the establishment of the committee indicates that investors were disappointed with the committee‟s composition. The committee consisted mostly of directors and people who were closely allied to directors. Event 14 is the publication of the fourth yearly evaluation report by the Monitoring Committee on compliance with Code Tabaksblat (financial year 2007) and the publication of Code Frijns by the Monitoring Committee. The positive reaction of investors indicate that – two months after the credit crisis erupted - they appreciated the major changes of the code: the new linkage between remuneration and long term goals, and that bonuses had to become in proportion with fixed salaries.

The T-values of the (Normalized) Average Abnormal Returns

Figure 1. The t-values of the (normalized) average abnormal returns of all thirty-two stocks for all 15 events (from 2002 until 2009), calculated with the CAC 40 as predicting variable, two-tailed tests (Thomson Reuters, 2010). Events: 1. Desire new code 2. Establishment Committee Tabaksblat 3. Release draft 4. Release code 5. Enactment Code Tabaksblat 6. Rumours successor chairman 7. First monitoring report 8. First evaluation report 9. Second evaluation report 10. Advisory report 11. Third evaluation report 12. Proposal changes 13. Leakages new code 14. Fourth evaluation & release Code Frijns 15 Enactment Code Frijns.

Figure 2 presents the cumulative average abnormal returns (CAAR) for all events from day [-5] to [+[-5] for all thirty-two stocks. Here the t-values of the normalized CAAR of events 2, 3, 4, 5, and 9 and 15 are significant. Of the not normalized CAAR only event 15 is significant. It

(22)

is notable that investors‟ reactions are predominantly positive, only the reactions to events 2 and 4 are significantly negative. The positive reaction to the release of the draft version of Code Tabaksblat might be explained by the understatement of Lodewijk de Waal (chairman trade union FNV): “Given the composition of the committee, it‟s a windfall to me.” (Het Financieele Dagblad, 2003c). Nevertheless, half a year later investors‟ reaction to the release of the final code is negative (event 4), although interest groups reacted predominantly positive. The negative reaction might be explained by CNV‟s (trade union) clear commentary. CNV is disappointed because the remuneration provisions were less strict than initially proposed (Het Financieele Dagblad, 2003d).

The T-values of the (Normalized) Cumulative Average Abnormal Returns

Figure 2. The t-values of the cumulative average abnormal returns (CAAR) for all events (from 2002 until 2009) from day [-5] to [+5] for all thirty-two stocks, calculated with the CAC 40 as predicting variable, two-tailed tests (Thomson Reuters, 2010). Events: 1. Desire new code 2. Establishment Committee Tabaksblat 3. Release draft 4. Release code 5. Enactment Code Tabaksblat 6. Rumours successor chairman 7. First monitoring report 8. First evaluation report 9. Second evaluation report 10. Advisory report 11. Third evaluation report 12. Proposal changes 13. Leakages new code 14. Fourth evaluation & release Code Frijns 15 Enactment Code Frijns.

You would expect that above results, that reject the first hypothesis (that the introduction of the Dutch codes did not lead to (cumulative) abnormal returns), would be supported by the abnormal returns of the AEX index. Nevertheless, none of the (cumulative) abnormal returns

(23)

of the AEX index listed in appendix G is significant. The returns of the AEX index do support hypothesis 1.

This difference in results is striking. It might be caused by the stocks of the AMX sample that are not included in the AEX index, and this explanation could be supported (in part) by hypothesis 4 that states that the Dutch codes will have less effect on larger companies than on smaller companies. The results of hypothesis 4 will be discussed below.

B. The Relationship Between Abnormal Returns and Industry Category, Market Capitalization, Leverage, Liquidity, and the Number of Employees

I perform equation (5) twice, for day [0], and [+1]. Table VII presents the results of these calculations.

Table VII

Regression results of the Cross-section of Abnormal Returns on Five Key Firm-Specific Covariates

This table presents the results of the regression of the cross-section of abnormal returns on five key firm-specific covariates: industry, number of employees, leverage in terms of total debt / total (shareholders’) equity, liquidity in terms of cash and cash equivalents / net sales, and market capitalization. The adjusted R-square belonging to the equation of day [0] is 0.019, and of day [1] is 0.001. ***, **, * indicate average significance at 1%, 5% and 10% levels, respectively, one-tailed tests for employees, liquidity, and market capitalization and two-tailed tests for the other variables (Thomson Reuters, 2010). Variable Equation 1: [0] t-statistic Equation 2: [+1] t-statistic Employees 0.533 0.417 Leverage -0.010 -0.056 Liquidity -0.746 -0.300 Market cap. -0.696 0.387

Industry 1: Food, beverages, tobacco 0.153 -0.832

Industry 2: Wood, cork, paper -0.105 -0.197

Industry 3: Publishing, printing 0.153 -1.652*

(24)

products

Industry 5: Machinery, equipment 0.202 -0.552

Industry 6: Construction 0.868 -0.661

Industry 7: Wholesale and retail trade -0.059 -0.964

Industry 8: Transport 0.706 -0.668

Industry 9: Post and telecommunications 0.058 -1.188

Industry 10: Financials -0.0519 -0.264

Industry 11: Propery investment 0.247 -0.363

Industry 12: Other services -0.524 -0.826

According to table VII, market capitalization is not of significant influence on the occurrence of abnormal returns. The t-values for day [0] and [+1] are -0.696, and 0.387 respectively (the corresponding coefficients are -0.001, and 0.001 respectively). Hence, hypothesis 4 is not supported. Thus market capitalization does not seem to explain the difference between the test results of the AEX index as a whole and the test results of the individual companies of the AEX and AMX index discussed in subsection A.

Table VII also shows that the abnormal returns are not dependent on the leverage level of companies, since the t-statistics are -0.010, and -0.056 for day [0] and [+1] respectively (the coefficients are -1.20E-05, and -9.25E-05 respectively). Hence, hypothesis 5 is supported. Although the relevant liquidity coefficient from equation (5) is negative (-0.001, and -0.000 respectively) on day [0] and [+1] - as theory predicts that more cash reserves reduces abnormal returns - the corresponding t-statistics (-0.746, and -0.300) from table VII are not significant. Thus hypothesis 6 is not supported.

Hypothesis 7 predicts that companies with more employees will have less abnormal returns and the relevant coefficient is positive on day [0] (0.001) and [+1] (0.001), but those t-statistics are not significant (0.533, and 0.417 respectively) as table VII shows. So hypothesis 7 is not supported by the results.

(25)

significantly the possible abnormal returns of companies. The t-statistics of table VII support this conclusion. The only significant t-statistic, for industry 3 (publishing, printing) for day [+1], is interpreted as an outlier, since the equation does not contain a constant in Eviews. Hence, hypothesis 2 is supported.

Values of Industry Coefficients

Figure 3: The values of the coefficients of the twelve different industries, of the equation that regresses the cross-section of abnormal returns of all thirty-two stocks for all 15 events at event day [0] on five key firm-specific covariates industry, number of employees, leverage in terms of total debt / total (shareholders’) equity, liquidity in terms of cash and cash equivalents / net sales, and market capitalization (Thomson Reuters, 2010). The upper boundary is the coefficient + 1.96*Std.error of a coefficient, the middle is the value of the coefficient, and the lower boundary is the coefficient – 1.96*Std. error of coefficient. Industries: 1. Food, beverages, tobacco 2. Wood, cork, paper 3. Publishing, printing 4. Chemicals, oil, non-metallic products 5. Machinery, equipment 6. Construction 7. Wholesale and retail trade 8. Transport 9. Post and telecommunications 10. Financials 11. Propery investment 12. Other services.

To investigate whether the introduction of the codes did not lead to (cumulative) abnormal returns in the case of financial services companies (industry 10) specifically, I perform various extra calculations. The (cumulative) (normalized) average abnormal returns of the financials for the 15 events individually (appendix H) are not significant, except for event 2 (establishment Committee Tabaksblat). So (also) investors in financials were probably disappointed with the committee‟s composition. None of the aggregated (normalized)

(26)

(cumulative) abnormal returns of the financials from day [-10] to [+10] were significant. Hence, hypothesis 3 is (at least) not fully supported.

The next section presents the conclusions and limitations of this research. V. Conclusion & Limitations

The test results of the thirty-two AEX and AMX listed companies reject the first hypothesis: the results indicate that the introduction of the Dutch codes did lead to (cumulative) abnormal returns. However, the (cumulative) abnormal returns of the AEX index as a whole were not significant: the results of the AEX index indicate that the introduction of the codes did not lead to (cumulative) abnormal returns. The difference between these two samples is roughly threefold: the (first) sample of the thirty-two stocks consists on average of smaller companies (in market capitalization) than the (second) sample of the AEX index as a whole, the first sample is an unweighted average and the second sample is a weighted average, and there is more trading in AEX stocks than in AMX stocks.

Since market capitalization is not of significant influence on the occurrence of abnormal returns (hypothesis 4 is not supported), you would not ascribe the difference in results between above samples to market capitalization. Further research will have to make clear what caused the abnormal return difference between the two samples.

The results also show that the other key characteristics of companies (industry category, leverage, liquidity, and the number of employees) do not influence the impact of the Dutch codes on stock returns.

(27)

An important limitation of my research is that different event dates disclose information about the same introduction of a corporate governance code. As a result, higher abnormal returns might have been found, if the different disclosures would not have been spread as they are. Therefore, shareholders might appreciate the Dutch codes more than the results of this study show.

VI. References

Aguilera, Ruth V., and Alvaro Cuervo-Cazurra, 2004, Codes of good governance worldwide: what is the trigger?, Organization Studies, 25, 415–443.

Aguilera, Ruth V., and Gregory Jackson, 2003, The cross-national diversity of corporate governance: dimensions and determinants, The Academy of Management Review 28, 447-465.

Akhigbe, Aigbe, and Anna D. Martin, 2006, Valuation impact of Sarbanes–Oxley: evidence from disclosure and governance within the financial services industry, Journal of Banking and Finance 30, 989-1006.

Akkermans, Dirk, Hans van Ees, Niels Hermes, Reggy Hooghiemstra, Gerwin Van der Laan, Theo Postma, and Arjen van Witteloostuijn, 2007, Corporate governance in The Netherlands: an overview of the application of the Tabaksblat Code in 2004, Corporate Governance: An International Review 15, 1106-1118.

Bebchuck, Lucian, Alma Cohen, and Allen Ferrell, 2009, What matters in corporate governance?, Review of Financial Studies 22, 783-827.

Brown, Stephen J., and Jerold B. Warner, 1985, Using daily stock returns: the case of event studies, Journal of Financial Economics 14, 3-31.

Corporate Governance Committee (Committee Tabaksblat), 2003, Dutch Corporate Governance Code (Code Tabaksblat).

Corporate Governance Code Monitoring Committee (Monitoring Committee), 2005, Rapport over de naleving van de Nederlandse corporate governance code.

(28)

Corporate Governance Code Monitoring Committee (Monitoring Committee), 2007, Derde rapport over de naleving van de Nederlandse corporate governance code.

Corporate Governance Code Monitoring Committee (Monitoring Committee), 2008a, Vierde rapport over de naleving van de Nederlandse corporate governance code.

Corporate Governance Code Monitoring Committee (Monitoring Committee), 2008b, Rapport over de evaluatie en actualisering van de Nederlandse corporate governance code. Corporate Governance Code Monitoring Committee (Monitoring Committee), 2009, Dutch Corporate Governace Code (Code Frijns).

Corporate Governance Code Monitoring Committee (Monitoring Committee), 2011, Monitoringrapport 2004 (boekjaar 2003). Retrieved 5/7/2011 from http://commissiecorporategovernance.nl/2004

De Jong, Abe, Douglas V. DeJong, Gerard Mertens, and Charles E. Wasley, 2005, The role of self-regulation in corporate governance: evidence and implications from The Netherlands, Journal of Corporate Finance 11, 473-503.

Diamond, Douglas W., and Robert E. Verrecchia, 1991, Disclosure, liquidity, and the cost of capital, The Journal of Finance 46, 1325-1359.

Edmans, Alex, Diego García, and Øyvind Norli, 2007, Sports sentiment and stock returns, The Journal of Finance, 62, 1967-1998.

Eng, L.L., and Y.T. Mak, 2003, Corporate governance and voluntary disclosure, Journal of Accounting and Public Policy, 22, 325-345.

Gompers, Paul, Joy Ishii, and Andrew Metrick, 2003, Corporate governance and equity prices, The Quarterly Journal of Economics, 118, 107-155.

Harford, Jarrad, Sattar A. Mansi, and William F. Maxwell, 2008, Corporate governance and firm cash holdings in the US, Journal of Financial Economics 87, 535-555.

(29)

Het Financieele Dagblad, 2003b, „Code moet effectiviteit commissarissen verhogen‟, March 11.

Het Financieele Dagblad, 2003c, Reacties op nieuwe gedragscode overwegend positief, July 2.

Het Financieele Dagblad, 2003d, Praktijk nu aan zet bij implementatie „Tabaksblat‟, December 10.

Het Financieele Dagblad, 2004, ABP-belegger Jean Frijns opvolger Tabaksblat, September 10.

Het Financieele Dagblad, 2008, Frijns wil toploon koppelen aan lange termijn, December 4. Jensen, Michael C., 1986, Agency costs of free cash flow, corporate finance, and takeovers, The American Economic Review 76, 323-329.

Jorgensen, Bjorn N., and Michael T. Kirschenheiter, 2003, Discretionary risk disclosures, The Accounting Review 78, 449–469.

MacKinlay, A. Craig, 1997, Event studies in economics and finance, Journal of Economic Literature 35, 13-39.

Reed Elsevier, 2011, Lexis Nexis.

Sengupta, Partha, 1998, Corporate disclosure quality and the cost of debt, The Accounting Review 73, 459-474.

Shleifer, Andrei, and Robert W. Vishny, 1996, A survey of corporate governance, NBER Working paper 5554, Harvard University and University of Chicago.

Thomson Reuters, 2010, Datastream.

Zhang, Ivy X., 2007, Economic consequences of the Sarbanes-Oxley Act of 2002, Journal of Accounting & Economics 44, 74-115.

VII. Appendices

Appendix A. Companies Removed from the AEX Sample

(30)

Company Reason of removal

ABN AMRO In 2007 the company was bought by Fortis, Banco Santander and Royal Bank of Scotland

Corporate Express In 2008 it was bought by Staples CMG In 2002 it merged with Logica Fortis In 2009 Fortis delisted

Getronics In 2007 it was bought by KPN Gucci In 2004 it was purchased by PPR Hagemeyer In 2008 it was purchased by Rexel KPNQwest In 2002 it went bankrupt

Numico In 2007 the company was bought by Danone VNU In 2006 VNU was bought by Valcon Acquisition

Appendix B. Companies Removed from the AMX Sample

(Reed Elsevier, 2011)

Company Reason of removal

Corus In 2007 it was bought by Tata Steel KLM In 2004 KLM merged with Air France Logica In 2002 it merged with CMG

Rodamco In 2008 it was bought by Unibail Stork In 2008 it was bought by Candover Vedior In 2008 it was purchased by Randstad Royal P&O

Nedlloyd

In 2005 it was bought by A.P. Moller-Maersk

Vendex KBB In 2004 it was bought by VDXK

Appendix C. The Descriptive Statistics of the (Abnormal) Returns

(31)

Abnormal returns AEX index

Returns all thirty-two stocks Skewness -0.106 -0.042 0.150 Kurtosis 2.556 2.778 7.12 Jarque-Bera 1.008 0.235 2276.827*** Sum -0.000 -0.000 -1.471 Sum Sq. Dev. 0.001 0.001 0.678

Appendix DI. Regression Equations Individual Companies with CAC 40

The parameters and adjusted R-squares of the regression equations with the individual thirty-two AEX and AMX companies‟ returns as dependent variable and the CAC 40 as independent variable in the estimation window. Period: 30/7/2002 to 16/12/2002 ***, **, * indicate significance at 1%, 5% and 10% levels, respectively, two-tailed tests (Thomson Reuters, 2010).

Aegon Ahold Akzo Nobel ASML DSM Reed Elsevier Heineken 1.497*** 1.433*** 0.617*** 1.663*** 0.502*** 0.515*** 0.240*** 0.001 -0.001 -0.000 0.000 0.000 0.000 -0.000 Adj. R2 0.680 0.676 0.376 0.562 0.454 0.372 0.133

Appendix DII. Regression Equations Individual Companies with CAC 40

ING Groep

KPN Philips Royal

Dutch Shell

TNT Unilever Van der Moolen

1.521*** 0.735*** 1.664*** 0.829*** 0.562*** 0.438*** 0.914***

(32)

Adj. R2 0.744 0.337 0.710 0.777 0.310 0.404 0.490

Appendix DIII. Regression Equations Individual Companies with CAC 40

Wolters Kluwer

ASM Int. Boskalis West.

Corio CSM Fugro Heijmans

0.358*** 1.663*** 0.213* 0.116** 0.103 0.073 0.262***

0.000 0.000 -0.002 -0.000 -0.000 -0.000 -0.001

Adj. R2 0.125 0.562 0.025 0.046 0.015 0.003 0.095

Appendix DIV. Regression Equations Individual Companies with CAC 40

Hunter Douglas Imtech BAM Groep Super de Boer

Nutreco Océ Randstad

0.117* 0.127 0.198** 0.259*** 0.114 0.736*** 0.878***

0.000 -0.002* -0.001 -0.000 -0.002 0.000 0.001

Adj. R2 0.019 0.014 0.037 0.065 -0.004 0.264 0.238

Appendix DV. Regression Equations Individual Companies with CAC 40

Vastned Retail

(33)

0.023 0.206*** 0.079* 0.167**

-0.000 -0.000 0.000 -0.001

Adj. R2 -0.006 0.110 0.029 0.032

Appendix E. Aggregated (Normalized) Average Abnormal Returns for All Thirty-Two AEX and AMX Stocks from Day [-10] to [+10] for All 15 Events

The aggregated average abnormal return is the average of the average abnormal return of each event window day [-10, +10] for all fifteen events. 18/12/2002 and 01/01/2009 are the first and last event dates respectively. Two-tailed tests (Thomson Reuters, 2010).

Appendix F. T-values of Aggregated Abnormal Returns of the AEX Index

The aggregated average abnormal return is the average of the average abnormal return of each event window day for all fifteen events. 18/12/2002 and 01/01/2009 are the first and last event dates respectively. ***, **, * indicate significance at 1%, 5% and 10% levels, respectively, two-tailed tests (Thomson Reuters, 2010). Day T-values aggegated AR [-10, +10] T-values aggegated AR [-10, +10], normalized -10 -0.017 -1.235 -9 0.013 0.977 -0.40 -0.20 0.00 0.20 0.40 0.60 0.80 1.00 -10 -8 -6 -4 -2 0 2 4 6 8 10 T -v a lue Day

T-value aggegated AAR [-10, +10]

(34)

-8 -0.018 -1.291 -7 -0.012 -0.906 -6 0.001 0.045 -5 -0.000 -0.030 -4 0.004 0.303 -3 -0.007 -0.504 -2 0.002 0.160 -1 -0.019 -1.366 0 -0.006 -0.450 1 0.018 1.347 2 0.001 0.076 3 0.034 2.513* 4 -0.005 -0.388 5 0.011 0.806 6 0.010 0.708 7 -0.003 -0.246 8 0.004 0.265 9 0.006 0.409 10 0.001 0.100

Appendix G. T-values of the (Cumulative) Abnormal Returns of the AEX Index

(35)

5 -0.007 0.575 6 -0.215 -0.086 7 0.233 0.410 8 -0.237 -0.183 9 -0.168 0.347 10 -0.278 0.380 11 0.266 0.824 12 0.422 0.437 13 -0.448 -0.186 14 0.492 -0.157 15 -0.007 0.887

Appendix H. T-values of the (Normalized) (Cumulative) Abnormal Returns of the Financials

The (normalized) abnormal and (normalized) cumulative [-5, 5] abnormal returns of the financials calculated with CAC 40 as predicting variable. 18/12/2002 and 01/01/2009 are the first and last event dates respectively. ***, **, * indicate significance at 1%, 5% and 10% levels, respectively, two-tailed tests (Thomson Reuters, 2010). Events: 1. Desire new code 2. Establishment Committee Tabaksblat 3. Release draft 4. Release code 5. Enactment Code Tabaksblat 6. Rumours successor chairman 7. First monitoring report 8. First evaluation report 9. Second evaluation report 10. Advisory report 11. Third evaluation report 12. Proposal changes 13. Leakages new code 14. Fourth evaluation & release Code Frijns 15 Enactment Code Frijns.

(36)