• No results found

For the forced resignations group a significant positive abnormal return of 0,5 % is found on the announcement day of the management departure

N/A
N/A
Protected

Academic year: 2021

Share "For the forced resignations group a significant positive abnormal return of 0,5 % is found on the announcement day of the management departure"

Copied!
32
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Abstract

This study examines the value relevance of top management changes by analyzing their announcement effects on the share price. This is done for companies listed on the Dutch stock exchange during 2000 – 2007. For the forced resignations group a significant positive abnormal return of 0,5 % is found on the announcement day of the management departure. Furthermore, company performance (ROA) is declining for the three years before the management change.

These results are consistent with internal corporate governance mechanisms. For the voluntary resignations group significant negative CAARs of 2 % are found. Age related resignations do not cause any market reaction. Furthermore, the market does not react on successor (internal or external) announcements.

(2)

Table of contents

1. Introduction 5

2. Theoretical background 7

2.1 Efficient market hypothesis 7

2.2 Agency theory 7

2.3 Corporate governance mechanisms 8

2.4 Corporate governance in The Netherlands 9

3. Review of literature and hypotheses 11

3.1 Pre-event company performance 11

3.2 Forced management departures 11

3.3 Voluntary- and age related management departures 12

3.4 Internal versus external successor 12

4. Methodology 13

4.1 Time line of an event study 14

4.2 Measuring abnormal return 14

4.2.1 Market Adjusted Returns model 15

4.2.2 Ordinary Least Squares market model 15

4.3 Test statistics 16

4.4 Assumption Student-t test 17

4.5 Pre-event company performance 18

5. Data 18

5.1 Sample firms 18

5.2 Identifying management changes 18

5.3 Descriptive analysis and outliners 19

5.4 Increasing the dataset 20

5.5 Classification of management changes 21

5.6 Return on assets 22

6. Empirical results 22

6.1 Pre-event company performance 23

6.2 Event study results 23

6.2.1 Whole sample 23

6.2.2 Forced management departures 24

6.2.3 Voluntary management departures 25

6.2.4 Age related departures 25

(3)

6.3 Successor results 26

6.3.1 Internal successor 26

6.3.2 External successor 27

6.4 Comparison with Cools and Van Praag 27

6.5 Results including the outliners 28

7. Conclusion, discussion and further research 29

7.1 Conclusion 29

7.2 Discussion 30

7.3 Further research 31

References 32

Appendices 34

(4)

1. Introduction

In 2003 one of the largest multinationals in the Netherlands, Ahold, reported financial problems.

One billion dollar of the stated profit did not seem to exist. This ‘gap’ in the annual report was mainly caused by consolidation problems of joint ventures (US Foodservice) in the United States.

The share price of Ahold dropped with 60 % on the announcement day of this accounting fraud.

Private shareholders and institutional investors lost a great part of their investments. In addition, stakeholders in general lost their trust in the company. Ahold was once known as one of the best performing companies in the Netherlands. Since this accounting scandal, Ahold lost a great part of their goodwill.

Cees van der Hoeven (CEO) and Michael Meurs (CFO) were held responsible for this scandal.

Michael Meurs was the financial executive of the company and he should have know about the misstatements in the annual reports. Cees van der Hoeven, the highest company executive, was responsible for the whole company. After the announcement of the accounting fraud, the board of directors forced these two managers to resign. In 2006, Van der Hoeven and Meurs appeared in court because they were held ‘personal responsible’ for the problems. The main accusations covered fraud and misstatements in the financial reports of the company. The two management directors were charged to pay a fine of 225.000 euro.

This paper will be about shareholder wealth effects caused by management changes. More specific, top management change announcements will be analyzed. Top management is defined as the set of individuals being member of the board of management. A change in management may create different effects because of the arguments behind the departure. Therefore, management changes will be classified into three groups: age related resignations, voluntary resignations and forced resignations. Based on the period 2000 – 2007, 124 useful management departures are found for companies listed on the Dutch stock exchange. The research question can be formulated as follows:

What is the influence of a top management change on the share performance of a company?

For the three subgroups different hypotheses are developed and tested. The focus of this paper will be on the forced resignations group. The Ahold case makes clear that managers (control) do not necessary have to make decisions in the best interests of the shareholders (owners). This separation of ownership and control and the problem it may cause is also known as the agency problem. To deal with this agency problem corporate governance mechanisms are developed.

(5)

Examples of these mechanisms are the stock market, the board of directors and legal protection.

In the Netherlands, the board of directors is the most important controlling mechanism (Cools and Van Praag, 2003). If the board of directors is independent and monitor management in the right way, they should dismiss a manager if he or she doesn’t act in the interest of the company. This happened with Cees van der Hoeven and Michael Meurs in the Ahold case.

Attention will also be given to the pre-event company performance of a management change. If corporate governance mechanisms are effective, there should be a negative relation between the probability of a top management change and share performance. Company performance should be declining before the management change. The yearly return on assets (ROA) surrounding the announcement day of the management change will be compared with each other.

The corporate governance mechanisms are less important for the age related resignations group and the voluntary resignations group. Retirements are anticipated and voluntary resignations are in the managers’ interest. Therefore, a different share price reaction is expected compared to the forced resignations group. Attention will also be given to the successor of the management function. This successor could be an internal or external person.

This paper will use an event study to measure the impact of a management change. An event study measures the impact of an unexpected company specific event on the share price of the company. The usefulness of such a study comes from the fact that, given efficiency in the market place, the effect of an event will be reflected immediately in the share price (MacKinlay, 1997). If the event is beneficial for the shareholders, the share price will increase. On the other hand, if the event is viewed as negative for the shareholders, the price will decline. The purpose of an event study is to measure this ‘abnormal’ return, calculated as the event specific return minus an expected normal return. Abnormal returns for the different subgroups will be calculated.

Cools and Van Praag (2003) also performed research on Dutch top management changes. They analyzed changes in the board of management over the period 1991 – 2000. This paper will differ in some way from the study of Cools and Van Praag. First, this paper will focus on the period 2000 - 2007. Second, Cools and Praag only studied forced resignations. This research will add two more categories: voluntary resignations and age related resignations. Third, attention will be paid to the company performance before the management changes occurs. Fourth, the background (internal or external) of the successor will be analyzed.

(6)

The remaining part of this paper is organized as follows. Section 2 discusses the theoretical background, attention will be given to the agency theory and the different corporate governance mechanisms. Section 3 provides an overview of the existing literature and the hypotheses will be given as well. Section 4 describes the event study methodology used in this paper. Section 5 provides information about the dataset. In section 6 the empirical results will be given. The paper ends with the conclusion and a discussion in section 7.

2. Theoretical background

Warner, Watts and Wruck (1988) state that:

‘If corporate governance mechanisms are effective, and if share price performance reflects information on manager’s efficiency, there will be a negative relation between the probability of a top management change and share performance. In addition, there should be a positive share price effect on the announcement day of a forced management departure’.

Therefore, attention will be first given to the efficient market hypothesis. Second, the agency theory will be highlighted. Third, the different corporate governance mechanisms will be analyzed.

This will also happen for the corporate governance system specific for The Netherlands.

2.1 Efficient market hypothesis

Market efficiency is one important assumption that have to be made if share price movements are going to be analyzed. There are in general three levels of market efficiency known. Strong-form market efficiency implies that all available information is reflected in the share price. Thus, all information about management changes will be reflected in the share price. In the weak-form efficiency, only past prices are reflected in the share price. In the semi-strong-form, market prices reflect past prices and public available information. The semi-strong-form is assumed in event study methodology; share prices reflect past prices and public available information.

2.2 Agency theory

Since Berle and Means (1932) it has been widely recognized that a potential divergence of interests may arise between managers and shareholders. Jensen and Meckling (1976) put this so called agency problem in a theoretical framework. The shareholder, who is the owner or

‘principal’ of the company, delegates day to day decision making in the company to the managers, who are the ‘agents’. There is thus a separation of ownership and control. The problem that arises as a result of this system of corporate ownership is that the agents do not necessary make decisions in the best interests of the principal (Solomon and Solomon, 2004).

(7)

Since the 1970s a stakeholder theory has been developed. The basis for this stakeholder theory is that there is not only a relation between managers and shareholders. Companies are so large that they are responsible for a broad range of stakeholders. These stakeholders include employees, suppliers, customers, creditors, governments, shareholders and the society at large.

Managers interests do not only have to be aligned with shareholders interests but also with the interests of all the other stakeholder groups. There is a growing perception among theorists and practitioners that the agency theory and the stakeholder theory are compatible (Solomon and Solomon, 2004). It is not possible to achieve long term shareholder wealth maximization without taking the interests of all the other stakeholders into account.

Thus, the primary reason for corporate governance is the separation of ownership and control, and the agency problem it may cause. Managers are likely to follow their own interests, and this may harm the interests of the other stakeholders. For example, managers may focus on company investments that provide high short term profits (because of the related bonus), rather than selecting investments that will provide shareholder wealth maximization in the long run. Corporate governance deals with mechanisms by which stakeholders exercise control over management such that their interests are protected (John and Senbet, 1998).

2.3 Corporate governance mechanisms

There are a number of ways stakeholders can monitor the management of a company. First, investors have legal protection. For example, shareholders have the right to vote on corporate matters, such as mergers and the election of the board of directors. This voting right is an important part of the financial asset the investor owns. However, voting rights turn out to be expensive to exercise and to enforce (Shleifer and Vishny, 1997). Another form of legal protection is that shareholders have the possibility to come up with resolutions.

Second, investors can get more effective control rights by being large. Institutional investors like pension funds, life insurance companies, unit trusts an investment trusts can play an active role in monitoring the board of management (Solomon and Solomon, 2004). Investment institutions have more power and resources available to allow them to focus on corporate governance issues.

Private equity funds also belong to this category and are in upcoming since the past decennium.

Private equity funds typically control management of the companies in which they invest, and often bring in new management teams that focus on making the company more valuable.

Shleifer and Vishny (1997) find that ownership concentration is indeed an important mechanism in controlling the agency problem.

(8)

Third, the stock market is an effective controlling mechanism. If the semi-strong market hypothesis is taken into account, a share price reflects public and historical information. This information includes decisions that are made by the management board. Managers are therefore forced to make decisions in the interests of the owners.

Fourth, the takeover market is an important controlling mechanism. Investors can gain control, either by a direct offer to purchase shares (tender offer) or by an appeal for shareholder votes for directors (proxy fight) (Fama and Jensen, 1983). If the investor gains the majority of the shares, management can be fired. Takeovers can thus be viewed as a rapid fire mechanism for ownership concentration. However, there are some doubts about the effectiveness of takeovers as a corporate governance mechanism. For example, takeovers can actually increase agency costs when bidding management overpay for acquisitions that bring them private benefits of control (Shleifer and Vishny, 1997).

Fifth, the board of directors have the power to hire, fire and compensate top management and to ratify and monitor important decisions (Fama and Jensen, 1983). The stock market and takeover market are external corporate governance mechanisms, the board of directors is an important internal mechanism. It is essential that the board of directors is independent and free from any relationships with the management board. Because of their independence, the board of directors should help to reduce the notorious conflicts between stakeholders and company management.

2.4 Corporate governance in The Netherlands

The corporate governance system is different all over the world, each country has an own unique system. This unique system is determined by a wide range of factors, including corporate ownership structure, the legal system, government policies, cross border investments, culture and history.

The corporate governance system in the Netherlands is also known as the Code Tabaksblat and this code is introduced in January 2004. The Ahold scandal in 2003 was one of the reasons of the introduction of this Code. Main goals of the Code Tabaksblat are improvements in the transparency of the financial statements and more power for the shareholders. For example, shareholders must be informed if the company wants to make a significant acquisition. An other object of the Code Tabaksblat is that board members may be compensated with a maximum of 1 year salary if they leave the company (gouden handdruk).

(9)

The Dutch system of corporate governance is characterized by a two-tier board system (Solomon and Solomon, 2004). Under this structure, the management board and the board of directors (supervisory board) are separate entities. The management board is entirely composed of executive directors, while the supervisory board is entirely composed of non-executive directors.

The main responsibility of the management board is day-to-day decision making and general corporate issues. The supervisory board monitors the management board and provides advice when necessary. Thus, the primary task of the supervisory board is to make sure that the management board acts in the interest of all the stakeholders. The two-tier board system is most common in The Netherlands, however there are companies with a one-tier board system. In this system both executive and non-executive directors are combined in one board.

The two-tier corporate governance structure is part of the Structure Act. Companies with more than 100 employees, a legally installed work council and a book value of shareholders’ equity exceeding 11 million Euro are bounded to this Structure Act. In October 2004 a ‘new’ version of the Structure Act was introduced in the Netherlands. Most important change is that the supervisory directors are nominated and elected by the shareholders’ meeting1. In the ‘old’

Structure Act it was normal for the supervisory board to elect their own directors (co-optation).

The position of the shareholders with the introduction of the new Structure Act is further strength.

For example, shareholders have an influence in the compensation structure of the management board1. The key attributes of a two-tier board system are summarized in table 1.

Table 1

Key attributes of two tier-tier board system Supervisory Board

Presence Obligatory

Role Monitoring of the management board and an advisory role Nomination and

appointment of directors By the shareholders’ meeting Number of directors Minimum of three non-executives Term of membership Four years

Composition Independent of management, non executive directors only

Management Board

Presence Obligatory

Role Day-to-day decision making

Nomination, appointment

and dismissal of directors By the supervisory board. The shareholders' meeting must be notified or consulted when managing directors are dismissed.

Number of directors One or more executive directors Composition Executive directors only

Source: Maassen and Van den Bosch (1999)

1 http://www.vandiepen.com/old/pubs/corporate.php?id=corporate3

(10)

The supervisory board is an important corporate governance mechanism in the Netherlands. This because of the fact that external mechanisms for disciplining management are almost non- existent in the Netherlands (Cools and Van Praag, 2003). However, shareholders are getting more and more power in the Dutch corporate governance system. The supervisory board, an internal corporate governance mechanism, is still the most important mechanism controlling management in The Netherlands.

3. Review of literature and hypotheses

3.1 Pre-event company performance

If corporate governance mechanisms are effective, and if share price performance reflects information on manager’s efficiency, there will be a negative relation between the probability of a top management change and share performance (Warner et al., 1988). This implies that company performance should be declining prior to the management change. Therefore, the following hypothesis will be tested:

H1 Top management change is inversely related to recent-past company performance.

Dedman and Lin (2001), and Denis and Denis (1995) find indeed a declining ROA for the three years before the management departure. Furthermore, Denis and Denis (1995) report a negative abnormal return of 11,4 % for the 200 days prior to the announcement of a management change.

Kang and Shivdasani (1996) find a negative return for the same period as well. Warner et al.

(1988) document a negative abnormal return of 5,0 % during a five month period before the management change. In summary, all find that performance is declining prior to the management change.

3.2 Forced management departures

If the supervisory board is independent and monitor management in the right way, it should dismiss a manager if he or she doesn’t act in the interest of the company. Therefore, an increase in the share price is expected if the manager is forced to leave the company:

H2 Announcements of forced resignations will cause a positive share price reaction.

The findings on forced management departures are not unambiguous. Cools and Van Praag (2003), and Warner et al. (1988) find no significant market reaction when a manager is forced to leave the company. Dedman and Lin (2001) performed research on companies in the United

(11)

Kingdom and find a significant negative abnormal return of 4,0 % in the case of forced resignations. Mahajan and Lummer (1993) find a significant negative abnormal return of 0,7 %.

This is in contrary with the results found by Dherment-Ferere and Renneboog (2000), Kang and Shivdasani (1996), and Denis and Denis (1995). They all find significant positive abnormal returns when the manager is forced to leave the company. Appendix A provides a detailed summary of all these studies.

3.3 Voluntary- and age related management departures

Corporate governance mechanisms are important in the case of a forced management departure.

These mechanisms are less important if a manager voluntary decides to resign, because with a voluntary resignation the manager acts in his own interest. Dherment-Ferere and Renneboog (2000) argue that a non-conflictual resignation will create a negative price reaction, because the company loses valuable company specific human capital. Furthermore, as retirements at the normal retirement age can usually be well anticipated, there is no reason that the announcement of a retirement would cause a price reaction. Therefore, the following hypotheses will be tested:

H3 Announcements of voluntary resignations will cause a negative share price reaction.

H4 Announcements of age related departures will cause no share price reaction.

Dedman and Lin (2001), Dherment-Ferere and Renneboog (2000), and Mahajan and Lummer (1993) all find no market reaction when a manager voluntary decides to leave the company. Kang and Shivdasani (1996) observe a positive significant announcement effect of 0,4 % during the two days surrounding the announcement day. These findings are in conflict with the hypothesis where a negative market reaction is expected.

In the case of age related retirements Dedman and Lin (2001), and Mahajan and Lummer (1993) document no market reaction. This is in contrary with the results of Denis and Denis (1995) who find a significant positive abnormal return of 0,2 % and Dherment-Ferere and Renneboog (2000) who find a significant negative abnormal return of 0,2 %.

3.4 Internal versus external successor

When a company has to choose a successor for a management function, they have the opportunity to promote an insider or select someone from outside the company. Bonnier and Bruner (1989) argue that the appointment of an outsider (insider) has a negative (positive) effect on performance because (a) outsiders have none of the company specific human capital, (b) the board of directors know insiders better than outsiders, (c) internal promotions give incentives to

(12)

junior managers and (d) the appointment of an external candidate may give a signal to the market that the current financial situation of the company is bad. Therefore, the following hypotheses will be tested:

H5 Announcements of internal appointments will cause a positive share price reaction.

H6 Announcements of external appointments will cause a negative share price reaction.

Furtado and Rozeff (1987) find a significant positive abnormal return of 1,1 % when an internal successor is appointed. Kang and Shivdasani (1996) find a positive announcement effect of 0,4

%. This is contrary with the results of Dherment-Ferere and Renneboog (2000), and Reinganum (1985) who both find a negative abnormal return on the announcement day of a internal successor. The results found on external succession all show in the same direction. Dherment- Ferere and Renneboog (2000), Kang and Shivdasani (1996), Furtado and Rozeff (1987) and Reinganum (1985) all document significant positive abnormal returns. The details of these studies can be found in appendix A.

In summary, the results of the various studies about the influence of top management changes on the share price of a company are conflicting with each other. The use of various sample selection mechanisms, research designs, control variables and definitions of management departures may be causes of the inconsistencies between the studies. In this study I will combine the positive aspects of the past studies.

4. Methodology

Most of the previous studies done on management changes used event study methodology to analyze the effects of these changes. An event study measures the impact of an unexpected company specific event on the share price of the company. The usefulness of such a study comes from the fact that, given efficiency in the marketplace, the effect of an event will be reflected immediately in the share price (MacKinlay, 1997). If the event is beneficial for the shareholders, the share price will increase. On the other hand, if the event is viewed as negative for the shareholders, the price will decline. The purpose of an event study is to measure this

‘abnormal’ return, calculated as the event specific return minus an expected normal return. In this section the procedure of an event study will be explained.

(13)

4.1 Time line of an event study

The time line of an event study can be found in figure 1. T0 is the day of the event. In this study the day of the management change announcement. If the market is efficient, an immediate share price reaction would be expected. However, it is not always clear when the announcement will reach the market. Due to information leakage and post-closing market announcements, news may reach the market before or after the announcement day (AD). Therefore, different event windows will be used:

• Event Window I [ t0 ]: Market efficiency and transparency would suggest this to be the relevant event window (Cools and Van Praag, 2003).

• Event Window II [ t-1, t0 ]: There is a possibility that news will reach the market one day before the official announcement because of information leakage. This event window captures this possibility.

• Event Window III [ t0, t+1 ]: This event window considers the possibility of an indirect announcement effect on day 1. In addition, announcements made after closing of the market are also included in this event window. (Cools and Van Praag, 2003).

• Event Window IV [ t-1, t+1 ]: This event window captures the two days surrounding the announcement day. It combines event windows II and III.

Figure 1 Time line event study

The estimation window [ t-220, t-21 ] is used in the Ordinary Least Squares market model to calculate ‘normal’ returns. More attention will be paid to this in section 4.2.2. Generally the event window itself is not included in the estimation window to prevent the event from influencing the normal performance model parameter estimates (MacKinlay, 1997). The post-event window will not be used in this study.

4.2 Measuring abnormal return

The return of a share can only be considered as ‘abnormal’ if it can be compared to a normal return. Thus, it is first necessary to specify a model generating ‘normal’ returns before ‘abnormal’

returns can be measured. Brown and Warner (1985) use the following three models for calculating normal returns: (1) Mean Adjusted Returns model, (2) Market Adjusted Returns model, and the (3) Ordinary Least Squares (OLS) market model. They conclude in their study that these

(14)

simple models perform well and are relatively powerful under a wide variety of conditions. In addition, the gains of using more difficult multifactor models, like the Arbitrage Pricing Theory model, are limited. Binder (1998) and MacKinlay (1997) both confirm that these simple models are relatively powerful. Therefore, the Market Adjusted Returns model and the OLS market model will be used in this study to calculate ‘normal’ returns. The Mean Adjusted Returns model will be excluded because it has low power in case of clustering in the data (Brown and Warner, 1985). In short, there is chosen for two models calculating normal returns. This is done to increase the strength of the outcomes of this study. As will be seen in the results section are the results of the two models almost the same.

4.2.1 Market Adjusted Returns model

The Market Adjusted Returns model (Brown and Warner, 1985) takes the movements of the market into account. The abnormal return is calculated as the difference between the return on the share and the corresponding return on the market index,

ARit = Rit – Rmt (1)

where ARit is the abnormal return on share i on day t; Rit is the actual return on share i on day t and Rmt is the return on the market index on day t.

The AEX index will be used as the market index. It is composed of the 25 most traded companies on the Euronext Amsterdam over the previous calendar year. The index is value weighted, based on the market capitalization of the companies. The AEX index provides a fair representation of the Dutch economy2.

4.2.2 Ordinary Least Squares market model

The OLS market model (Brown and Warner, 1985) takes into account both market factors and the systematic risk of each share. For each event, the market model will be estimated by Ordinary Least Squares. This will be done over a 200-day estimation window (period between -220 and - 21 days before the announcement),

Rit = ai + iRmt + it t = [-220, -21] (2)

2 www.euronext.com

(15)

where Rit is the actual return on share i on day t; Rmt is the return on the AEX index on day t; ai and i are the estimated parameters of the market model during the estimation period and it is the disturbance term assumed to be normally distributed with zero mean and constant

variance.

Next, the abnormal return can be estimated with the market model, ARit = Rit – (ai + iRmt) (3)

where ARit is the abnormal return on share i on day t; Rit is the actual return on share i on day t;

Rmt is the return on the AEX index on day t and ai and i are the estimated parameters of the market model during the estimation period.

4.3 Test statistics

The next step in the event study is that the abnormal returns will be tested for statistical significance. The hypothesis to be tested is that the average abnormal return (AAR) is different from zero. It is therefore first necessary to calculate this average abnormal return for the N events on day t with the following formula,

AARt =

= N

i ARit

N 1

1 (4)

with variance,

var(AARt) =

= N

i ei

N 1 2 2

1 σ (5)

Next, the average abnormal return of a single day can be tested for significance with a Student t- test (Brown and Warner, 1985). A test statistic of this form is widely used in event studies,

t=

)

var( t

t

AAR

AAR (6)

To test over a multi day event window [t2,t3] it is necessary to cumulate the average abnormal returns (CAAR) of the specific days in the event window,

(16)

=

= 3

2 3 2, ) (

t t t

t t

t AAR

CAAR (7)

with variance,

=

= 3

2 3

2 ) var( )

var( ( , ) t

t

t t

t

t AAR

CAAR (8)

To test for significance, the Student t-test will be used,

t=

)

var( ( , )

) , (

3 2 3 2

t t t t

CAAR

CAAR (9)

4.4 Assumption Student t-test

For the statistical test used in this study the assumption is made that the abnormal returns are normal, independent (otherwise clustering possible) and identically distributed. In practice this assumption does not always holds. Brown and Warner (1985) performed research on this assumption and their main findings will be presented in short.

Brown and Warner find that the non-normality of daily returns has no obvious impact on event study methodology. Moreover, non-parametric tests like a Sign test and a Wilcoxon Rank test do not perform better than a Student t-test. It appears that these non-parametric tests themselves suffer from a problem of misspecification. Brown and Warner report furthermore that for their simulations it makes little difference whether or not clustering is taken into account. The rejection rates when there is clustering, are not markedly different from those when there is no clustering.

For the hypothesis tests over intervals of more than one day (CAAR), the failure to take into account autocorrelation could result into misspecification. However, Brown and Warner observe in their study that autocorrelation plays a minor role. They find that the benefits from autocorrelation adjustments appear to be limited.

After these findings Brown and Warner conclude that a Student t-test performs well, even when the underlying assumption of normality, independency and identically distribution does not hold.

MacKinlay (1997) confirms that deviations from the assumption will in general not lead to problems. Binder (1998) analyzed the event study methodology since 1969 and he confirms that many of the above stated problems may simply be ignored, because in practice they are quite minor. Taking these findings into account, this study will continue to use a parametric Student-t test.

(17)

4.5 Pre-event company performance

The return on assets (ROA) will be used as measure for company performance. ROA is calculated as the annual earnings divided by total assets of the company. The ROA is often used in economics to compare the yearly performance of companies with each other. In addition, Dedman and Lin (2001) and Denis and Denis (1995) both used ROA in their study as pre-event performance indicator. If the same indicator is used, it will be easier to compare the results of this study with the outcomes of Dedman and Lin (2001) and Denis and Denis (1995).

5. Data

5.1 Sample firms

The sample used in this study consists of the 100 most traded companies listed on the Euronext Amsterdam3. It was not possible to select more companies, because information/press release about top management changes in small companies is limited4. 24 companies were excluded from the sample due to one of the following reasons: share data not available, annual reports not available, no management changes, company merged or an unclear board structure. Cools and Van Praag (2003) also used trading volume as a measure of size and their sample of 100 companies represents more than 80 % of the total market value of the population of Dutch listed companies. Therefore, the remaining 76 companies should be a fair representation of the Dutch economy. The companies that are included and excluded in the sample can be found appendix B and C.

5.2 Identifying management changes

Top management is defined as the set of individuals being member of the board of management.

This includes the Chief Executive Officer (CEO) and the Chief Financial Officer (CFO). Cools and Van Praag (2003) analyzed the management changes in the Netherlands over the period 1991 – 2000. This study will analyze the changes in the board of management that took place from January 1, 2000, to December 31, 2006. For each company in the sample, the annual reports from the different years were compared with each other5. Where different names in the management board were recorded, a management departure was deemed to have occurred.

Over the seven year period, 227 changes in the management board were recorded for the 76 companies.

3 These 100 companies are based on an average seventeen day (January 2007) trading volume.

4 Originally the sample consisted of 120 companies, but it was not possible to find management change information for the 20 smaller companies.

5 More than 800 annual reports were collected from www.company.info

(18)

Next, the announcement date of each management change was identified with the use of Het Financieele Dagblad. This journal is the Dutch equivalent of the Wall Street Journal and its online version is an efficient and complete database for relevant press articles about Dutch firms (Cools and Van Praag, 2003). Besides the announcement date the following information was collected for each event: name of manager, company, exit motive and successor (internal or external). With the use of Het Financieele Dagblad it was possible to identify announcement dates for 171 (75%) of the total of 227 management changes. However, 35 press releases about management turnover were combined with other company specific news, these 35 events are excluded from the sample. Share price data was obtained form Datastream. No share price data was available for 6 events. After these exclusions the sample consists of 130 Dutch management changes.

5.3 Descriptive analysis and outliners

Figure 2 provides an analysis of the abnormal returns based on the sample of 130 management changes. These abnormal returns are based on the OLS market model. The figure shows that there are some outliners on the right side of the mean in the distribution. To test whether the abnormal returns are normally distributed a Jarque-Bera test is performed. The outcome of the Jarque-Bera test is significant; this indicates that the sample of abnormal returns is not normally distributed. The outliners may cause this non-normality. Therefore, an analysis without the outliners is also performed. 6 outliners with abnormal returns all above 15 % are excluded form the sample. Figure 3 provides a descriptive analysis of the sample without the 6 outliners.

Figure 2

Descriptive statistics of whole sample including outliners

0 5 10 15 20 25 30

0.0 0.1 0.2 0.3

Series: OW Sample 1 130 Observations 130 Mean 0.009644 Median 0.000144 Maximum 0.319183 Minimum -0.075666 Std. Dev. 0.061785 Skewness 3.126700 Kurtosis 15.08550 Jarque-Bera 1002.973 Probability 0.000000

(19)

Without the outliners the Jarque-Bera test is not significant and this means that the sample is normally distributed. This switch form non-normality to normality may indicate that the outliners could have a big influence on the results of this study6. Therefore, the 6 outliners will be excluded from the sample. The final sample will consist of 124 management changes.

Figure 3

Descriptive statistics of whole sample excluding outliners

0 4 8 12 16 20

-0.05 0.00 0.05

Series: WO Sample 1 124 Observations 124

Mean -0.001932 Median -0.000570 Maximum 0.090634 Minimum -0.075666 Std. Dev. 0.029796 Skewness -0.004959 Kurtosis 3.806750 Jarque-Bera 3.363209 Probability 0.186075

5.4 Increasing the dataset

There were thoughts to increase the dataset, however that is complicated for the following reasons. First, increasing the number of companies would not increase the number of useful management changes, because information/press release about management turnover in small companies is limited. Second, the focus of this study is the Netherlands, adding an additional country could lead to problems because of differences in board structure and corporate governance.

Other studies also had to deal with the problem of a small dataset. Bonnier and Brunner (1989) used a sample of 87 management changes. Dherment-Ferere and Renneboog (2000) analyzed sub samples with a maximum of 37 management changes. Cools and Van Praag (2003) used in their analysis small sub samples as well.

6 In the Results section it will become clear that the results based on the sample excluding outliners are significant different compared to the results of the sample including the outliners.

(20)

5.5 Classification of management changes

The focus of this study are the forced management departures. Therefore, the 124 management changes are classified in three groups: age related resignations, voluntary resignations and forced resignations. This classification is also used by Dherment-Ferere and Renneboog (2000).

They include the following exit motives in the voluntary resignations group: health problems, death, personal motives and an internal change of function7. An age related departure implies that the manager leaves the firm because of his/her pension. The forced resignations group consists of the following exit motives: scandal, bad performance, difference in opinion, external change of function and corporate restructuring8.

It was necessary to use and classify the exit motives into groups, because forced resignations are rarely described as ‘forced’ in press releases. In some of the announcements no particular reason for the management departure is given, these announcements are put in a separate category.

Table 2 provides an overview of the classification of the management changes. 48 % of the management departures are forced, 28 % is voluntary, 11% is related to age and for 13 % no reason is given.

Table 2

Classification of management changes

N %

Age related resignations 14 11,29%

Pension 14 11,29%

Voluntary resignations 34 27,42%

Health related 2 1,61%

Death 0 0,00%

Personal 24 19,35%

Internal change of function 8 6,45%

Forced resignations 60 48,39%

Scandal 3 2,42%

Bad Performance 16 12,90%

Difference in opinion 21 16,94%

External change of function 5 4,03%

Corporate restructuring 15 12,10%

Reason unclear 16 12,90%

Total 124 100,00%

7 Warner et al. (1988) classifies an internal change of function as an voluntary resignation. They classify an external change of function as a forced resignation, because the manager leaves the firm.

8 Kang and Shivdasani (1996) include corporate restructuring in the forced resignations group.

(21)

Information about the successor is also collected. Only in 37 % of the management change announcements information was also provided about the successor. 23 % of the managers are succeeded by an internal person and 14 % are succeeded by an external candidate.

Table 3 Successor

N %

Internal successor 29 23,39%

External successor 17 13,71%

No information available 78 62,90%

Total 124 100,00%

Appendix D provides an overview of the descriptive analysis of the different subgroups. Most important result is that almost all the subgroups have a sample that is normally distributed. This is important because normality is one of the assumptions of the Student t-test.

5.6 Return on assets

The return on assets (ROA) were collected for the sample of 76 companies with the use of Datastream. Yearly ROA’s from 1996 till 2006 were available for 71 companies. This resulted in a dataset of 87 management announcements with useful ROA’s.

6. Empirical Results

This section presents the empirical results of the study. In the first part attention will be paid to the accounting performance (ROA) of the company before the management change has occurred.

Next, the event study results will be presented. This will be done for the whole-, forced-, voluntary- and age related group. The results regarding the successor, internal or external, are also given. Finally, the results of the study will be compared with the results of Cools and Van Praag (2003). Note that the results are based on the sample without the outliners. The results including the outliners can be found in appendix E.

(22)

6.1 Pre-event company performance

H1 Top management change is inversely related to recent-past company performance.

Figure 4 presents the results of the average return on assets surrounding the 87 management change announcements. A change in management occurs in year 0. The figure shows that the companies suffer from declining ROA for the three years before the year of the management change. Moreover, in year 0 the ROA becomes even negative. After the change in management has occurred the ROA becomes positive and shows a positive trend.

Figure 4

ROA around management departure (n=87)

-3,00 -2,00 -1,00 0,00 1,00 2,00 3,00 4,00 5,00 6,00

-3 -2 -1 0 1 2

Year

ROA in %

ROA

This result is in line with the hypothesis. Company performance is declining prior to the management change. Dedman and Lin (2001) and Denis and Denis (1995) also find a declining ROA for the three years before the management departure.

6.2 Event study results

6.2.1 Whole sample

Table 4A presents the results of the event study based on the whole sample of 124 management changes. On the announcement day, d = 0, a small negative abnormal return of 0,20 % is presented. This holds for both the Market adjusted model and the OLS market model. This announcement effect is however small and insignificant. For the other event windows the results are similar. The cumulative average abnormal returns (CAARs) are negative, small and insignificant.

(23)

Table 4A

Results whole sample ( n = 124)

OLS model Market model CAAR (%) t-value CAAR (%) t-value

[d = 0] -0,19% -0,64 -0,20% -0,66

[d = 0, d = 1] -0,56% -0,93 -0,67% -1,11 [d = -1, d = 0] -0,37% -0,62 -0,32% -0,53 [d = -1, d = 1] -0,74% -0,82 -0,79% -0,87

Denis and Denis (1995), Mahajan and Lummer (1993) and Warner et al. (1988) find no significant market reaction as well. The results are conflicting with earlier findings by Kang and Shivdasani (1996), Bonnier and Bruner (1989), Weisbach (1988) and Furtado and Rozeff (1987) who all find a significant positive abnormal return.

6.2.2 Forced management departures

H2 Announcements of forced resignations will cause a positive share price reaction.

Table 4B provides the results of the announcement effects caused by a forced management departure. On the announcement day a positive abnormal return of 0,59 % (OLS market model) and 0,53 % (Market model) is shown. These returns are both significant. The significance level is however 10 %, this level is quite high. In addition, if the event window is two days or larger the significant returns seem to disappear. In short, only on the announcement day a significant abnormal return of 0,5 % is found, the evidence is however not that strong.

Table 4B

Results forced resignations ( n = 60)

OLS model Market model CAAR (%) t-value CAAR (%) t-value

[d = 0] 0,59%* 1,31 0,53%* 1,19

[d = 0, d = 1] 0,71% 0,80 0,49% 0,55 [d = -1, d = 0] 0,34% 0,38 0,33% 0,37 [d = -1, d = 1] 0,46% 0,35 0,29% 0,21

* Significant at 10 % level

These results of a positive market reaction are consistent with the hypothesis. If a manager doesn’t act in the interest of the company, he or she should be dismissed. Therefore, an increase in the share price should be expected on the announcement day of a forced management

(24)

departure. The results are consistent with the significant positive results found by Dherment- Ferere and Renneboog (2000), Kang and Shivdasani (1996), and Denis and Denis (1995). The results are in contrast with the results of Cools and Van Praag (2003) and Warner et al. (1988) who both find no market reaction. They are also conflicting with the results of Dedman and Lin (2001) and Mahajan and Lummer (1993) who both find a significant negative abnormal return.

6.2.3. Voluntary management departures

H3 Announcements of voluntary resignations will cause a negative share price reaction.

Table 4C shows the results of the voluntary resignations group. On the announcement day a negative abnormal return of 1,4 % (OLS market model) and 1,2 % (Market model) is shown.

These returns are both significant. The other event windows also show significant negative CAARs, in general larger than 2 %.

Table 4C

Results voluntary resignations (n = 34)

OLS model Market model CAAR (%) t-value CAAR (%) t-value

[d = 0] -1,40%*** -2,77 -1,20%** -2,37 [d = 0, d = 1] -2,22%** -2,19 -2,07%** -2,04 [d = -1, d = 0] -1,52%* -1,50 -1,20% -1,19 [d = -1, d = 1] -2,34%* -1,54 -2,07%* -1,36

*** Significant at 1 % level, ** 5 % level, * 10 % level

These results are consistent with the hypothesis. A voluntary resignation creates a negative market reaction because the company loses valuable specific human capital. The results are conflicting with the results of Dedman and Lin (2001), Dherment-Ferere and Renneboog (2000), and Mahajan and Lummer (1993). They all find no market reaction when a manager voluntary decides to leave the firm. The results also conflict with the positive significant announcement effects found by Kang and Shivdasani (1996).

6.2.4 Age related departures

H4 Announcements of age related departures will cause no share price reaction.

Table 4D presents the CAARs of the age related resignations group. The abnormal returns are insignificant. This is consisted with the hypothesis, because retirements are anticipated and

(25)

therefore no market reaction is expected. Dedman and Lin (2001), and Mahajan and Lummer (1993) find indeed no market reaction. However, the results are in contrary with the results of Denis and Denis (1995) who find a significant positive abnormal return and Dherment-Ferere and Renneboog (2000) who find a significant negative abnormal return.

Table 4D

Results age related retirements ( n = 14)

OLS model Market model CAAR (%) t-value CAAR (%) t-value

[d = 0] -0,08% -0,13 -0,39% -0,67

[d = 0, d = 1] -0,52% -0,44 -1,19% -1,02 [d = -1, d = 0] 0,07% 0,06 -0,01% -0,01 [d = -1, d = 1] -0,37% -0,21 -0,81% -0,47

6.3 Successor results 6.3.1 Internal successor

H5 Announcements of internal appointments will cause a positive share price reaction.

Table 4E shows the results when an internal successor is appointed. The CAARs for the different event windows are negative but insignificant. It seems that there is no significant movement in the share price, the hypothesis is rejected. The results are in conflict with the significant positive market reactions found by Kang and Shivdasani (1996) and Furtado and Rozeff (1987). They are also in conflict with the significant negative abnormal returns found by Dherment-Ferere and Renneboog (2000) and Reinganum (1985).

Table 4E

Results internal successor ( n = 29)

OLS model Market model CAAR (%) t-value CAAR (%) t-value

[d = 0] -0,28% -0,57 -0,40% -0,80

[d = 0, d = 1] -0,45% -0,45 -0,87% -0,87 [d = -1, d = 0] -0,23% -0,23 -0,30% -0,30 [d = -1, d = 1] -0,39% -0,26 -0,77% -0,52

Referenties

GERELATEERDE DOCUMENTEN

The spectrum shows that the interaction of electromagnetic waves with rough surfaces results in a periodic energy profile, with a periodicity close to the wavelength of the

This effect relies on the matching of the standing wave field within the multilayer stack with the structure: the minima of the wave field intensity are placed in the center

Identity extraction, merging and correlation in Tracks In- spector will be demonstrated using a working system with a case containing evidence that has already been processed..

Methicillin-resistant Staphylococcus aureus (MRSA) is a bacterium resistant against most antibiotics. It belongs globally to the most frequent causes of

(2011) European risk factors’ model to predict hospitalization of premature infants born 33–35 weeks’ gestational age with respiratory syncytial virus: validation with Italian

The socio-economic factors included as independent variables in the multivariate regressions consist of the home country gross domestic product (GDP) of the sponsoring

Appendix IV: Worldwide equity indices used as proxies for Fama and French factors R mt is the equity market’s return; R ft is the 3 month Euribor rate. SMB t is the excess

For example, a higher dividend/earnings pay out ratio would mean that firms would pay a larger part of their earnings out as dividends, showing off a sign of