Continental European countries Burak Genc, Enschede, 14-05-2012

Abstract: Globalization is rapidly and radically changing the world. Multinational firms are becoming more common as firms expand across country borders, but the firms require an international CEO. This research analyses how investors value the internationality of a CEO by measuring the investor response to CEO succession announcements. It was found that investors in British firms react positively to a CEO succession announcement, while investors in German firms react negatively. Furthermore, the German firm investors’ response is moderated by the internationality of the firm, which is not the case for British firm investors. Keywords: CEO succession, internationality, investor response, corporate governance, globalization

The Rise of the International CEO:

differences between Anglo-Saxon and

Continental European countries

2

“The performance of CEOs determines the fates of corporations,

which collectively influence whole economies. Our standard of

living depends upon excellence at the very top.”

- Charan (2009), p. 72

C

OLOFON

Document: Master’s thesis

Version: Final

Date: 14-05-2012 Author: Burak Genc Student number: 2000008

Email: burakgenc@me.com Phone number: +31644748065

Address: Hengeveldebrink 21, 7544 TL, Enschede Institution: University of Groningen / Uppsala University

Faculty: Faculty of Economics and Business / Department of Business Studies Degree programme: International Financial Management

Phase: Master

Supervisor: Dr. W. Westerman Assessor: Dr. B. Qin

© Burak Genc, Enschede, 2012

3

TABLE OF CONTENTS

§3.1SAMPLE & DATA-COLLECTION ... 19

§3.2RESEARCH METHOD ... 20

4. DESCRIPTIVE ANALYSIS ... 30

§4.1TOTAL SAMPLE ... 30

§4.2COUNTRY-SPLIT SAMPLES ... 33

§4.3REGRESSION INDEPENDENT VARIABLES ... 37

5. EVENT STUDY ANALYSIS ... 40

§5.1ABNORMAL RETURNS ... 40 §5.2SUB-SAMPLES ... 45 §5.3REGRESSION ANALYSIS ... 46 6. CONCLUSION ... 52 §6.1SUMMARY ... 52 §6.2IMPLICATIONS ... 53 §6.3LIMITATIONS ... 58 REFERENCES ... 59

APPENDIX I: COUNTRY DIFFERENCES ... 63

APPENDIX II. CORRELATIONS ... 66

APPENDIX III. DESCRIPTIVE STATISTICS ... 67

APPENDIX IV. ROBUSTNESS ... 76

APPENDIX V. SUB-SAMPLE ANALYSIS ... 84

APPENDIX VI. UNIVARIATE REGRESSIONS ... 85

4

1. INTRODUCTION

How important is the CEO for a company? Three conflicting viewpoints in current academic literature have tried to answer this question. The “common sense” approach claims that managerial succession improves firm performance, while the “vicious circle” approach advocates that changes in leadership create tensions and disruptions that deteriorate performance, which in turn again lead to managerial change. The third viewpoint regards CEO succession as “ritual scapegoating” (Davidson III, Worrell & Cheng, 1990). According to the Upper-Echelon Theory, the organization can be seen as a reflection of its top management, as their perception of problems and implementation of strategies have a direct influence on the performance of the firm (Balsmeier, Buchwald & Zimmerman, 2011; Fondas & Wiersema, 1997).

5 down, as HP’ shares fell 47% under his watch, as a result of deteriorated operational performance. After the announcement of Whitman as the new CEO3, HP’ shares fell to a 6-year low indicating that the investors do not believe that Whitman is the right person to turn HP’ fortunes around. These events highlight the strategic significance of a CEO succession for a firm, as it can cause large stock value movements and can have a direct influence on the firm’s future performance (Dedman & Lin, 2002).

It is quite surprising that almost half of the large multinational companies with revenues above $500 million have no meaningful CEO succession plan. Furthermore, the average CEO tenure shrank from 9.5 years in 1995 to 7.6 years in 13 years’ time and two out of every five CEOs fail within the first 18 months of taking office. The Corporate Leadership Council found that only 20% of the 276 interviewed HR executives of large companies were satisfied with the corporate executive succession processes within their firms. Last but not least, just three recruiting agencies control 80% of the CEO recruitment market of the Fortune-100 companies, indicating the lack of competition in this market due to low demand (Charan, 2009).

As a result of an increasingly globalizing world, international business has become more important for firms. For example, international trade increased from $95 billion in 1955 to $12 trillion in just 50 years’ time4. Trade in foreign currencies reached a volume of $1.5 trillion per day5 in the early 21st century. However, globalization is not without its challenges, as it introduces new types of risks (for example exchange rate risk) and magnifies the scale of all ‘normal’ risks (Hoffman & Gopinath, 1994). Approximately 60% of all international strategic alliances fail as a result of either incompatibility between partners or the inherent riskiness of operating in a foreign country which has different value systems and cultures (Dacin, Hitt & Levitas, 1997). Internationally experienced managers have the experience that is needed to successfully manage an international firm, making them a valuable asset to multinational firms (Daily et al., 2000). According to the Upper-Echelon Theory, a firm should be better able to handle the international challenges when its top

6

management team itself is more international, resulting in a more internationally diversified firm (Hermann & Datta, 2005; Sambharya, 1996).

This research focuses on the ‘internationality’ of the successor CEO, which has been mostly neglected in past research. The first contribution to the current academic literature is that the effects of ‘CEO internationalization’ are measured on the short-term, instead of the long-term as used by Daily et al. (2000). Secondly, it explores the importance of internationality of the CEO by integrating the concepts of globalization, Resource-Based View, Upper-Echelon Theory and the Efficient Market Hypothesis (EMH). Thirdly, it approaches CEO succession from the Upper-Echelon Theory, while focusing on internationality. The practical relevance of this research is that by knowing how investors respond to the CEO’s internationality, managers can better anticipate the investor response to a CEO succession announcement and fine-tune their search for the ‘best’ CEO successor. Furthermore investors will gain more insight into their behaviour and whether they value characteristics of the successor CEO correctly. Therefore the main research question of this research is: DOES THE INTERNATIONALITY OF THE CEO INFLUENCE THE PERFORMANCE OF THE FIRM?

7 $3,629 billion larger than the UK’s economy with a GDP of $2,481 billion6. However, the size of Germany’s stock market ($1,429 billion) is dwarfed by the size of the UK’s stock market ($3,107 billion)7. Cernat (2004) claims that the corporate governance model is an important factor in determining the economic performance of companies. One of the implications of the corporate governance model for a CEO succession is that, as stakeholders have more influence in the Continental European model, investors have less control over the appointment of the successor CEO, but on the other hand they also face less agency costs than the Anglo-Saxon investors (Cernat, 2004). The question of how the corporate governance model influences the investor response to key strategic decisions has not been answered in current academic literature. This question is answered by the following sub-question: WHAT ARE THE DIFFERENCES BETWEEN ANGLO-SAXON AND CONTINENTAL EUROPEAN COUNTRIES IN INVESTOR RESPONSES TO CEOSUCCESSION?

The companies in the sample of this research were drawn from the UK and Germany, representing respectively the Anglo-Saxon and the Continental European corporate governance models. By comparing the investor responses to CEO succession announcements in these two countries, the effects of the two models on investor expectations of future firm performance can be measured. The time frame of this research is 2004-2010, which captures an entire business cycle and includes the global financial crisis that started to hit European companies in the third quarter of 2008.

The structure of this research is as follows. The next chapter discusses the current literature and builds the theoretical framework. The research methodology is dealt with in the third chapter, by first outlining the sample selection criteria and data-collection methods and then explaining how the collected data was analysed. Chapter four presents the descriptive analysis and chapter five the statistical analysis. The statistical analysis consists of two phases: first the abnormal returns are analysed for statistical significance and then cross-sectional regressions are performed. The last chapter concludes the research by summarising the findings, drawing conclusions and providing suggestions for future research.

6

IMF World Economic Outlook Database (September 2011)

7

8

2. THEORETICAL FRAMEWORK

§2.1

F

The primary goal of business from a managerial perspective is to create shareholder value by paying dividends and/or increasing the share price (Hillier, Ross, Westerfield, Jaff & Jordan, 2010). Managers can increase shareholder value by earning profits from the firm’s operations. Shareholders value the firm on its profit generating capacity, meaning they value the expected future profits.

The next two sub-sections discuss the Efficient Market Hypothesis (EMH) and the Upper-Echelon Theory. The EMH is founded on the assumption that new information is immediately incorporated into the stock price (Malkiel, 1989). The Upper-Echelon Theory explains why and how certain CEO characteristics influence the investors’ response to a CEO succession announcement.

§2.1.1EFFICIENT MARKET HYPOTHESIS

Firm performance can be measured from the managerial and the investor perspective. The managerial perspective focuses on operational performance, measured by operational measures such as the ROA, ROI and ROE (Daily et al., 2000; Zajac, 1990). The investor perspective focuses on financial performance, measured by the stock value of the firm (Davidson III et al., 2002; Lubatkin et al., 1989). De Wet (2005) concludes that the operational performance measures have serious credibility issues, because the correlation between the accounting measures and shareholder value has been found to be almost non-existent. Furthermore, as operational performance requires long-term measurement, it results in multiple variables and events confounding with the original event (Lubatkin et al., 1989).

9 affect the firm’s performance and as the stock price aggregates all investor expectations (Eiteman, Stonehill & Moffet, 2010), it implies that the investor response to the announcement of a strategic event offers the best measure of the effects of that single strategic event for the future operational performance of the firm.

As poor CEO succession often results in poor firm performance and corporate instability (Charan, 2009), managers have to consider the CEO succession as a strategic event. The (measured) investor response to the CEO succession depends on the fit between the CEO’s characteristics, the expectations of the investors and the CEO succession context and the research methodology (Pitcher, Chreim & Kisfalvi, 2000). The EMH predicts that the fit between these factors determines the investor response, as investors forecast the future performance based on their expectations of the managerial performance of the successor CEO.

§2.1.2UPPER-ECHELON THEORY

The Upper-Echelon Theory asserts that the backgrounds, experiences and values of the corporate executives influence their strategic decisions (Tihanyi, Ellstrand, Daily & Dalton, 2000), meaning that the firm’s vision and strategic orientation can be explained by certain characteristics of the top management team (Nielsen, 2010). However, the large amount of tests required to measure the beliefs and values (psychological dimensions) of the corporate executives is one of the reasons why managers are reluctant to participate in these researches (Hambrick & Mason, 1984). The Upper-Echelon Theory on the other hand, is founded on the assumption that easily measurable characteristics can explain certain organizational outcomes, even-though these measures contain more ‘noise’ than true psychological measures. Demographic characteristics are often used as surrogates for measuring cohort behaviour, as they have the advantages of objectivity, comprehensiveness, parsimony, logical coherence, predictive power and testability (Sambharya, 1996).

10

accompanies organizational change. Firms with younger executives were found to be more internationally diversified in spite of the challenges and risks associated with it (Herrman & Datta, 2005; Tihanyi et al., 2000).

Tenure is also an important determinant for firm performance, as individuals with longer company tenures have better knowledge of the firm’s core capabilities, allowing them to better assess the capabilities of the firm in unknown situations. This in turn results in more successful international diversification (Tihanyi et al., 2000).

Educational level is associated with the executive’s cognitive orientation and knowledge base. A higher education results in more innovation, higher tolerance for ambiguity, more knowledge and an openness to change. A higher educated top management team results in a more internationally diversified firm, because they are better capable in facing ambiguous situations and have better information-processing capabilities required to manage international business (Hermann & Datta, 2005).

The Upper-Echelon Theory is founded on the assumption that not just the CEO’s characteristics, but the characteristics of all top management team members affect the performance of the firm. The managerial responsibilities are the domain of the entire management team and cannot be attributed to a single individual (Herrman & Datta, 2005). However, the CEO is the most influential employee of the firm and therefore deserves special attention for its individual influence on the firm (Balsmeier et al., 2011). Because of the CEO‘s high internal and external visibility, they are generally considered to be the key strategic decision-maker in large companies (Beatty & Zajac, 1987). Furthermore, a CEO succession is usually followed by a high senior executive turnover as a result of the reorganization of the strategic leadership of the firm by the CEO (Shen & Canella Jr., 2002). This reorganization process is the result of the successor CEO’s efforts to clear inefficiencies and to facilitate strategic reorientation Even though the amount of change that the new CEO can initiate depends on the country’s corporate governance model, this reorganization process does imply that the CEO will restructure the top management team as much as possible, to realize the best alignment between the values and beliefs of all the top management team members.

11

§2.2

C

§2.2.1CULTURAL DIMENSIONS

Appendix I.A shows the cultural dimensions for the UK and Germany (Hofstede, 2001). Though both countries are identified as individualistic societies according the Individuality index, the UK outranks Germany with 22 points, making it one of the most individualistic societies in the world. This is why the British focus on individual achievements, while in Germany the focus is on family relationships. British people are encouraged to think about their personal contribution to society while Germans prefer to work with people based on personal preference, with sense of duty and responsibility towards the employer and society.

The Uncertainty Avoidance Index (UAI) shows the largest difference between the UK and Germany. The high UAI of Germany represents a preference to control the future instead of just letting it happen. Plans are derived according to deductive reasoning and are not easily changed when new information comes to light. The British on the other hand do not mind waking up to the fact that they don’t know what will happen and plans are frequently changed and adapted. This is why in the UK plans focus on the end goal, while leaving the process of actually getting there emergent and open to change, while in Germany plans are thought out in detail, minimizing future risks and problems. This results in Germans being well prepared for problems if they emerge, but also limits their adaptability to the environment. The British on the other hand can be characterised as having a high curiosity for creativity and innovation, making everything that is ‘new’ attractive.

In this research it is assumed that investors are locally-oriented. In other words, investors choose to invest in the countries that have the closest match with their values and beliefs. British firm investors are assumed to be risk tolerant, while German firm investors are assumed to be risk-averse.

§2.2.2CORPORATE GOVERNANCE MODEL

12

corporations. This perspective results in short-term profit-oriented behaviour by market actors that compete for material success. The main characteristics of this model are dispersed shareholding, a broad delegation of corporate responsibilities to corporate managers and a strong shareholder-protection climate. Employee and non-financial shareholder influences are low. In the Continental European model, banks play an important role in the allocation of capital. This model adopts a broader perspective and considers also the interests of stakeholders. Furthermore, banks and employees have a strong influence on corporate decision-making in the Continental European model, as banks frequently own a large block of shares of their clients’ companies as a measure of control. The Continental European model reduces the possibility of agency conflicts and favours long-term profits, providing a safer environment for large-block shareholding. Investors in Continental European companies usually own larger blocks of shares to gain more control over the firm’s operations, as exiting the company is more costly and more difficult due to less liquid stock markets. Furthermore, the influence of banks results in more intensive monitoring of corporate managers, making investment less risky. The British firm shareholders tend to own smaller blocks of shares, giving them less control over the firm’s future, but also better exit options when the company is no longer in line with their strategic vision (Cernat, 2009).. Appendices I.B and I.C provide an overview of the major differences between the two corporate governance models.

In the Anglo-Saxon model, the shareholders are the only actors that appoint the board of directors, while in the Continental European model, many actors are involved in this process. This results in the expectation that the investor response to a CEO succession will be more positive in the Anglo-Saxon model, as it is more likely that the investors’ CEO of choice will get appointed. On the other hand, the Anglo-Saxon model offers the investors better exit options because of the dispersed ownership and more liquid stock markets, but they also face more agency costs due to the lack of external monitoring. This results in the expectation that the investor responses to the CEO successions at Anglo-Saxon companies will be more severe.

13 On the other hand, the Anglo-Saxon firm investors can more easily sell and buy shares on the more liquid stock markets. This should result in more severe stock-price movements, as investors can more easily enter and exit the company. The second reason is that the additional external monitoring of the corporate executives in the Continental European model, combined with the long-term profit oriented focus, should result in less agency conflicts between the shareholders and managers. This results in the expectation that investors will be less likely to enter the company (as they will prefer owning large blocks of shares to gain power) but also less likely to exit (as a result of the long-term profit orientation) in the event of a strategic change. Therefore the second hypothesis is (H2):THE COUNTRY’S CORPORATE GOVERNANCE MODEL WILL AFFECT THE RELATIONSHIP BETWEEN INVESTOR RESPONSE AND CEO SUCCESSION ANNOUNCEMENT.

§2.3

G

Globalization has become an umbrella word for a number of environmental, sociological, political and economic trends that present changes and challenges on a global scale (Smeral, 1998). In economic terms, globalization represents the increasingly international trade of goods and services and the internationalization of businesses. Many industries are increasingly dealing with stagnating growth and saturation in home markets and have to expand internationally to stimulate revenue growth (Preda, 2008). This is one of the reasons why countries with small home markets (The Netherlands, Sweden, etc.) house a relatively large degree of multinational firms (Heijltjes, Olie & Glunk, 2003).

The Upper-Echelon Theory implies that selecting a CEO successor with extensive international experience is the building block to internationalizing its top management team. In the next sub-sections the effects of globalization on the top management of firms will be discussed by approaching the need for an international top management team from a Resource-Based View of the firm.

§2.3.1RESOURCE-BASED VIEW

14

A multinational firm should have a well-planned key executive succession plan, because the top managers determine the firm’s strategies and its success (whether domestic or international), especially when considering that executives with international experience are rare. International experience is gained by accumulating cultural knowledge and results in less ambiguity with cross-cultural situations (Daily et al, 2000). Charan (2009) found that many multinational have no meaningful CEO succession planning, making it a heterogeneous resource for multinationals.

The acquired knowledge and experiences associated with international experience are tacit knowledge and are therefore difficult to transfer by language (Polanyi, 1966). The CEO has to have acquired this knowledge and experiences in his past work experiences, which limits ex post competition.-

When a CEO decides to work for one company he invests multiple years of effort in the firm and its strategy and (usually) signs a contract that prohibits joining a competitor for a period of time, resulting in imperfect mobility. In the case of a transfer to a competitor, the original firm can further increase its advantage before the transferred CEO has familiarized himself with the company (Charan, 2009).

As many companies currently do not have corporate executive succession planning high on their agenda, the ex-ante competition is limited. But with increasing globalization and the opening up of emerging markets, multinationals are recruiting more often internationally experienced managers (Nielsen, 2010). The earlier firms start with implementing CEO succession planning, the lower the costs will be of hiring top recruiters to build a pool of potential international corporate executives. The greater the number of foreign managers in a company, the greater the chance that foreign managers will enter the top management of the firm (Heijltjes et al., 2003).

From a Resource-Based View, the primary benefits of international experience for an executive are the development of an international network, which is critical for firms competing in a global environment (Daily et al., 2000), and the accumulation of knowledge of and experience in managing foreign operations (Sambharya, 1996). §2.3.2CEO INTERNATIONALITY

15 Heenan, 1974). Lin & Shuh (2008) have found that when top managers are from different nationalities, there is more flexibility in decision-making and better conflict resolution. Being a foreigner to the firm’s home country is an important indicator of managerial competence and the ability to deal with the complexities of foreign expansion (Nielsen, 2010). It also forces the managers to adopt an international view, as they have different nationalities themselves. Buckley & Brooke (1992) also found the country of origin of the top executives to be one of the most important explanations for the international expansion of companies. The inclusion of foreign members in the top management of the company ensures the availability of a solid international knowledge base at a corporate level (Heijltjes et al., 2003). The Resource-Based View argues that a foreign manager’s advantage of having a better understanding of the firm’s international environment should result in better firm performance. Reuber & Fischer (1997) found that the top manager’s country of origin had predictive power over the propensity for the firm’s success in international business. Therefore hypothesis 3A is: THE APPOINTMENT OF A FOREIGN CEO IS POSITIVELY RELATED TO THE FIRM’S STOCK-MARKET PERFORMANCE.

The second CEO internationality dimension is country of education. Having had foreign education results in a more extensive international network, more experience with other cultures and exposure to an international environment. The heterogeneity in country of education of the top management team is positively related to firm performance (Lin & Shuh, 2008). When following education abroad, companies and managers frequently choose elite institutions, as these offer higher quality education and prestige. These institutions also offer the opportunity to meet ‘elites’ from diverse international backgrounds, resulting in more awareness of international issues and more openness to other cultures (Tihanyi et al., 2000). This results in the expectation that investors will value foreign education positively, resulting in hypothesis 3B: THE APPOINTMENT OF A FOREIGN EDUCATED CEO IS POSITIVELY RELATED TO THE FIRM’S STOCK-MARKET PERFORMANCE.

16

(Sambharya, 1996). Managers lacking international experience are less certain of their abilities to manage and control foreign operations, resulting in overstating the potential risks while understating the potential benefits of international expansion (Herrman & Datta, 2005). A competent transnational manager is often regarded as the building block of any international strategy (Adler & Bartholomew, 1992), making a CEO with international business experience a major competitive advantage to a multinational firm (Daily et al., 2000). Therefore hypothesis 3C is defined as: THE APPOINTMENT OF A CEO WITH FOREIGN WORK EXPERIENCE IS POSITIVELY RELATED TO THE FIRM’S STOCK-MARKET PERFORMANCE.

§2.3.3FIRM INTERNATIONALITY

Daily et al. (2000) found that that a firm’s international exposure influences its strategic leadership, with more internationally exposed firms preferring a CEO with more international experience, because these executives will be better able to manage the complexities of an international business environment (Barney, 1992). More international firms show better long-term performance when an international CEO is appointed (Daily et al., 2000). This implies that investors should value the internationality of the successor CEO relative to the importance of international business to the firm. Therefore hypothesis four is (H4): FIRM INTERNATIONALITY MODERATES THE RELATIONSHIP BETWEEN CEO INTERNATIONALITY AND FIRM STOCK-MARKET PERFORMANCE.

§2.4

S

H Expectation 1

The stock price of a firm is related to the announcement of a CEO succession

2

The country's corporate governance model will affect the relationship between investor reponse and CEO succession announcement

3a

CEO foreign nationality is positively related to the firm's stock-market performance

3b

CEO foreign education is positively related to the firm's stock-market performance

3c

CEO foreign work experience is positively related to the firm's stock-market performance

4

Firm internationality moderates the

relationship between CEO internationality and investor response

Resource-Based View (Wernerfelt, 1984) CEO internationality (Daily et al., 2000)

Hypotheses

Literature

Efficient Market Hypothesis (Malkiel, 1989) Upper-Echelon Theory (Tihanyi et al., 2000)

Cultural Dimensions (Hofstede, 2001) Corporate governance model (Cernat, 2009)

Firm internationality (Daily et al., 2000)

3. METHODOLOGY

The first part of this chapter outlines the sample characteristics and data-collection methods. The second part discusses the research methodology and the statistical tests performed to analyse the data.

§3.1

S

&

-

This research focused on large-cap publicly listed European firms. A new dataset was created for this research as no existing datasets were suitable to test the hypotheses. As this research has the closest resemblance to Daily et al. (2000), their sample criteria were used for this research. The sample companies were drawn from the UK and Germany, as these two countries represent the two corporate governance models.

One of the sample requirements was that the firms had to be drawn from a wide variety of industries, to guarantee the heterogeneity of the business environment and the internationality of the sample firms. The second requirement was that the firms had to be publicly listed, because firm performance was measured by stock market response. The third requirement was that firms had to be drawn from the ‘largest-cap’ company stock-index of the country first and then from the smaller cap indices. The reason is that larger firms have relatively more international sales due to economies of scale and have more power to attract internationally experienced managers (Daily et al., 2000). Therefore, the sample companies were drawn from the FTSE-100 (UK) and the HDAX (Germany) indices, which contain respectively the 100 and 110 largest-cap companies of the respective countries.

A few events were contaminated with multiple CEO successions during one event period and therefore were removed from the sample. One special case was the RBS Group’s announcement of Stephen Hester as the new CEO. This announcement coincided with the British government’s announcement of cash infusion in the company, resulting in an abnormal return of -24%. Therefore this event was classified as an extreme outlier and was also removed from the sample.

20

different announcement date, the earliest date was used as the announcement date. All companies in the HDAX and FTSE-100 indices were examined for a CEO change in the time period of 2004-2010. This timespan was chosen because it captures an entire business cycle of seven years and it includes the global financial crisis. The global financial crisis started to have a major impact on European firms from late-2008 through mid-2009. Controlling for the effects of the global financial crisis allowed for the examination of the effects of an uncertain and turbulent environment on the relationship between CEO succession and investor response.

A total of 63 and 61 CEO succession announcements were identified from respectively the FTSE-100 and the HDAX, resulting in a total sample size of at 124 events. Green’s (1991) rule of thumb (N>50+8*M), with N being the sample size and M the number of independent variables, requires the sample size to be bigger than 82 (not accounting for the control variables). The total sample size of 124 events met Green’s criteria, but the country sub-samples did not. Increasing the sample size would have only been possible by expanding the research focus: by adding more countries to the sample, extending the time frame or expanding the focus to mid/small cap publicly listed companies. According to Green’s (1991) rule of thumb, the bare minimum is 58 events for a univariate OLS regression. Therefore any conclusions that were drawn from the country sub-samples were rigorously tested for validity.

§3.2

R

§3.2.1EVENT STUDY

To measure the investor response to a CEO succession announcement, the event study research methodology was used (MacKinlay, 1997). This method relies on the assumption of (semi-strong) EMH, implying that the investor response to an event can be measured by the movement of the firm’s stock price to the release of new information. The event day,

t

21 windows were tested: [0], 1,+1], 2,+2], 1,0], [0,+1], 15,-1], [+1,+15] and [-15,+15]. This resulted in the following research timeline:

FIGURE 2:RESEARCH TIMELINE

§3.2.2THE MARKET MODEL

MacKinley (1997) recommends using the Market Model for calculating the abnormal returns of stocks. By removing the portion of the return that is related to the variation in the market return, the variance in the abnormal returns can be reduced, allowing for better detection of event effects. The abnormal returns were calculated with:

̂ ( )

where is the actual return and ( ) is the expected return of firm i on day t.

The formulas for the actual firm and market returns were as follows:

where is firm i’s total return index on day t, is firm i’s total return index on day t-1. The FTSE-ALLSHARE index was used as the UK market index, as it represents 98-99% of the British market capitalization, making it the benchmark for the UK market. The HDAX index was used as the German market index, as it represents over 95% of all German market capitalization.

The expected returns were calculated with the CAPM model: ( ) ̂ ̂

22

This resulted in the following equation for the calculation of the abnormal returns: ̂ ̂ ̂

The expected value of the abnormal returns is 0. The residuals’ variance (

) is

∑

̂

where T is the number of days in the estimation window. MacKinley (1997) showed that for a large estimation window the sampling error will converge to 0. The estimation window of 175 days used in this research was large enough to assume that sampling errors would be cancelled out, making the variance of the abnormal returns equal to the variance of the residuals.

By averaging the abnormal returns across firms in common event time, the average abnormal returns ( ̅̅̅̅ ) were obtained, calculated by:

̅̅̅̅

_{∑ ̂}

By aggregating the average abnormal returns through the days of the event window of [t1,t2], the cumulative average abnormal returns ( ̅̅̅̅̅̅) were obtained (Serra, 2002):

̅̅̅̅̅̅

_∑

The Student’s t-test was performed to examine whether the average abnormal returns on the event days were significantly different from zero. The t-values were calculated by (Dedman & Lin, 2002):

̅̅̅̅

√

∑

The t-statistic for the ̅̅̅̅̅̅ was calculated by (MacKinlay,1997):

23 where N is the number of events and ̅̅̅̅ is the cross-sectional abnormal returns

variance on day t, calculated with:

̅̅̅̅ ∑

The Student’s t-test is a very naïve test procedure, as it implicitly assumes that the mean effect of the event is identical through all observations. Furthermore, it does not allow unequal variances through observations and cross-correlation through abnormal returns. Another problem with the standard t-test is that observations with large standard deviations distort the t-value significantly (Harrington & Shrider, 2002). In the presence of heteroskedasticity or cross-sectional dependence, the estimated standard errors will be biased or the estimated mean abnormal returns will be inefficient. Therefore Strong (1992) recommends using Patell Standardized Residual Test (PSR), which assumes that the residuals are not correlated and that there is no significant event-induced variance, but does not suffer from the problems associated with the Student’s t-test. The PSR requires the abnormal returns to be standardized, which results in the standardized abnormal returns :

̂

where the numerator is the abnormal return of firm i on day t and the denominator is the standard deviation of the residuals of firm i during the estimation window. The Z- score was calculated to test whether the standardized average abnormal returns were significantly different from zero (Strong, 1992):

∑ √∑

where N is the number of events on day t and Ti is the number of days in the

estimation window.

By aggregating the average abnormal returns of N events through the event window of [t1,t2], the standardized cumulative average abnormal returns ( ) were obtained:

∑ ∑

24

The Z-score of the SCAR were computed by (Strong, 2002):

√

where L is the number of observations in the event window of [t1,t2].

Boehmer, Musumeci & Poulsen (1991) found that events could change the risk of securities, resulting in a temporary increase of variance of the abnormal returns surrounding the event. Ignoring event-induced variance can result in incorrectly rejecting the null hypothesis of zero abnormal returns. Cowen & Sergeant (1996) showed that in the presence of event-induced variance, standard parametric tests and non-parametric tests become misspecified. Therefore Serra (2002) recommends Boehmer et al. (1991)’s cross-sectional standardized abnormal return test (Z-cross), as this test is correctly specified in the presence of event-induced variance:

∑

√ _∑ ̅ ⁄

where N is the number of observations on day t.

The Z-cross of the SCAR was calculated by (Cummins & Lewis, 2002):

√

where N is the number of events. The standard deviation of SCAR is calculated with:

√ ∑ ̅̅̅̅̅̅

§3.2.3EQUALITY OF VARIANCES

25

∑

∑

∑

where N is the number of events, k is the number of sub-groups, Ni is the number of

samples in group i, ̅ , is the mean of for group i and is the

mean of all . W is tested against F(a, k – 1, N – k), where a is the significance level with k – 1 and N – K degrees of freedom.

§3.2.4ROBUSTNESS

To test the stability of the research findings through methodological differences, the abnormal returns were also estimated with the mean adjusted returns model and the market adjusted returns model (Strong, 1992). The following formula was used to calculate abnormal returns with the constant mean return model:

̂ ̅

where ̅ represents the average stock returns during the estimation window of [-200,-26].

The market adjusted return model was calculated by simply subtracting the market return on day i from the firm return on the same day:

̂ §3.2.5SUB-SAMPLE COMPARISON

The unequal sample sizes, independent two-sample t-test was used to test whether the abnormal returns of the UK and Germany samples were significantly different from each other. As unequal sample variances can severely bias this test, a Levene’s test of equality variances was performed first. Based on the result of the Levene’s test, either the equal or the unequal variance t-test was performed.

The formula for the equal variance t-test is: ̅ ̅

√

26

√

The formula for the unequal variance t-test is: ̅ ̅

√

§3.2.6MULTIVARIATE REGRESSION

The next step was to test which characteristics of the successor CEO had explanatory power over the investor response to the announcement of CEO successions.

Dependent

The dependent variable, the firm’s stock market response ( ̅̅̅̅ ̅̅̅̅̅̅), was measured by DataStream’s total return index (RI). The RI is defined by DataStream as: ‘The theoretical growth in value of a shareholding over a specified period, assuming that dividends are re-invested to purchase additional units of an equity or unit trust at the closing price applicable on the ex-dividend date’. The formula for RI is:

Where: RIt (RIt-1) = return index on day t (t-1), PIt (PIt-1) = price index on day t (t-1), DYt = dividend yield in % on day t and N = number of working days in the year

Independent

Foreign nationality (FN), foreign education (FE) and foreign work experience (FWE) were retrieved manually from the CEO’s biography on the company website or yearly reports. Additional details were retrieved from news articles about the CEO. All three variables were dummy coded with 0 for local and 1 for foreign.

The country corporate governance model was represented by the firm’s country of origin, retrieved from the Orbis database. This variable was also dummy coded, with 0 for Germany and 1 for the UK.

Moderator

27 firm internationality was measured by two variables; percentage foreign sales of total sales (FIPERC), as recommended by Jorion (1990) and the monetary amount of foreign sales (FIMON). The total monetary amount of foreign sales (FIMON) was calculated by taking the natural logarithm of total foreign sales. FIMON is a hybrid variable, controlling for both firm internationality and firm size.

The data for total and foreign sales were retrieved manually from the latest annual report of the firm prior to the announcement date. When necessary, the amount of sales of British firms were translated from Pounds to Euros, with the exchange rates on the announcement dates retrieved from the ECB’s website.

Control

Zajac (1990) recommends adding firm size as a control variable, as it can influence its performance due to economies of scale, market power and access to resources. More importantly, is can also affect the volatility and the liquidity of the stock, as stock of larger firms exhibit less volatility and more liquidity. Maug (1998) found that it is cheaper and easier for investors to sell or buy more liquid shares. Lubatkin et al. (1989) found that firms with larger market capitalizations on average hired more outsiders, as a result of greater attracting power of managerial talent. Furthermore, larger firms are also less able to initiate strategic change, resulting in diminishing growth (Evans, 1987). Two control variables for firm size were used: total firm sales and total market capitalization.

Total sales (TOTSALES), as calculated by the natural logarithm of total sales, is frequently used as a proxy for firm size (Daily et al., 2000).

Zajac (1990) used average assets as a proxy for firm size, as his measure for performance was the ROA, creating a direct link between the firm size and the firm performance variable. This research followed the methodology of Zajac (1990), but instead of average assets, market capitalization was used as a proxy for firm size, in line with Lubatkin et al. (1989). This variable was calculated by taking the natural logarithm of total market capitalization (MCAP). The total sales and total market capitalization on the day of the announcement were retrieved from DataStream.

28

a contender succession occurring when the previous CEO was forced to step down. An outsider succession occurs when the new CEO comes from outside of the firm. A follower succession has no effect on firm performance, a contender succession a positive effect and outside succession a negative effect (Shen & Canella Jr., 2002).

However, Davidson III et al. (2002) found a positive shareholder reaction to an outside CEO succession from a related industry. The reason is that the more related the outsider CEO is to the industry of the firm, the less time it will need for familiarizing with the firm’s business. This results in immediate organizational change, boosting short-term performance of the firm. The industry relatedness was measured by comparing the successor’s previous and current firms’ primary SIC codes. Secondary SIC codes do not relate to the test variables and were therefore ignored (Davidson III et al., 2002). The more digits of the SIC code match, the more related the industries are (with a 4-digit match indicating the highest possible match). The outsiders were divided into two groups, the ‘high match’ group (with a 3- or 4-digit match) and the ‘low match’ group (with 0-, 1- or 2-4-digit match).

The successor origin variables were dummy coded, with the followers as the baseline group. The information about the CEO’s origin and industry relatedness were manually determined from the publicly available information such as news articles on the CEO succession, the firm’s yearly reports and the CEO’s biography. The Orbis database was used to retrieve the SIC codes of the companies.

Manufacturers usually have longer-term strategies than service providers, as manufacturing is more capital intensive (Brouthers & Brouthers, 2003), which could influence the investor response to a strategic event such as a CEO succession. The SIC code of the CEO’s current firm was also used to determine the industry type (IT), which was dummy coded with 0 for service providers and 1 for manufacturing firms. SIC codes in the range of 2000-3999 were classified as manufacturers8.

The CEO’s age (AGE) is a proxy for the executive’s experience, propensity for risk taking and openness to change and has been found to influence the investor response to strategic decisions (Tihanyi et al., 2005). The age of the CEO was retrieved from the Orbis database.

8

29 A crisis period dummy (CRISIS), set for the time period of 01/09/2008 – 31/08/2009, was inserted into the equation to test for the effects of the global financial crisis on the investor responses to CEO succession announcements. The main effects of the global financial crisis were heavy losses for financial firms and the freezing of credit lines of firms from their banks.

The variables above result in the multivariate cross-sectional regression as represented by the following formula:

⁄

The moderator analysis was be performed by adding an interaction term to the regression by multiplying the FN, FE and FWE variables with the FIPERC or the

FIMON variables. When the interaction term was significant, the moderation was

further analysed with a simple slope graph. §3.2.7TESTS OF ASSUMPTIONS

30

4. DESCRIPTIVE ANALYSIS

§4.1

T

The average age of the successor CEOs (mean and median) was 49 years. Women were underrepresented, as only 3 out of 124 CEOs were female. No major differences between CEO successor internationality across age or gender were observed.

FIGURE 3:INTERNATIONALITY OF THE SUCCESSOR CEO(YEARLY)

Figure 3 shows that on average 30% of the successor CEOs were foreigners, 31% have had foreign education and 69% have had foreign work experience.

FIGURE 4:INTERNATIONALITY OF THE SUCCESSOR CEO(SUCCESSOR ORIGIN)

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 2004 2005 2006 2007 2008 2009 2010 Foreign nationality Foreign educated Foreign work experience

0% 10% 20% 30% 40% 50% 60% 70% 80% 90%

Follower Contender Outsider

31 On average, 65% of the new CEOs were insiders, which can be further split into 73% followers and 27% contenders. Of the 35% outsiders, 74% had low industry-match and 26% had a high industry-match.

Figure 4 seems to further indicate a correlation between the successor CEO internationality and successor origin, with outsiders being more international. This correlation also holds when controlling for the industry relatedness of the outsider successors, with more unrelated outsiders being more international. An explanation could be stricter selection criteria when the successor CEO is more out of line with the firm’s current strategy, to increase the chances that the new CEO will be able to successfully change and adapt to the firm’s strategy and culture. It could also be an effect of the firm having a greater pool of CEO candidates to select from.

The crisis period was set from 01/09/2008 – 31/08/2009, suggesting that on average 1/7th (14%) of the 124 CEO successions should originate from the crisis period.

FIGURE 5:%-OF TOTAL CEO SUCCESSIONS (YEARLY)

Figure 5 shows a slightly higher share of 19% (23 CEO successions) during the crisis period. The number of CEO successions started increasing in 2006, peaking in 2008/2009. The global financial crisis started to unfold during mid-2007 in the US and spread to Europe during late-2008. The increase in CEO succession could have been caused by more shareholder pressure for strategic change as the performance of many firms deteriorated when the crisis started to hit European firms. Combining figure 3 with figure 5 shows that the global financial crisis also had an influence on the importance of the internationality of the successor CEO. The amount of foreign

32

nationals hired in the crisis period decreased from 32% to 23%, foreign educated successors decreased from 32% to 28% and successors with international work experience decreased from 70% to 59%. One of the reasons for firms preferring a more locally-oriented CEO during the crisis period could be that firms chose to (temporarily) focus on their home market to reduce their risk exposure, because of the turbulent environment caused by the crisis. As nationality is the most ‘visible’ internationality characteristic of the successor CEO, it could explain why it responded faster and fluctuated more dramatically when the global financial crisis started.

The descriptive statistics of the firm characteristics reveal that 54% were from the service industry and 46% from the manufacturing industry. Financial services firms had the largest share (13%). No major differences between CEO internationality across industries were observed.

FIGURE 6:SUCCESSOR CEO INTERNATIONALITY (MARKET CAPITALIZATION)

Firm size measured by total sales did not show a clear relationship with CEO internationality. However, figure 6 suggests a positive correlation between firm size (measured by total market capitalization) and CEO successor internationality, as firms with higher market capitalizations hired relatively more often CEOs with foreign nationalities and foreign work experience. Foreign education showed a more ambiguous relationship with market capitalization. Prestige of leading a larger firm could explain the positive correlation between the firm’s market capitalization and the

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% €100 - €1000 €1000 - €10000 > €10000 Foreign nationality Foreign Education

33 internationality of the CEO, although it surprising that total sales did not show the same positive correlation.

FIGURE 7:SUCCESSOR CEO INTERNATIONALITY (INTERNATIONAL SALES)

The firm internationality (%-of total sales) did not show a clear correlation with the internationality of the CEO. However, the monetary amount of international sales did suggest a positive correlation (figure 7). The more revenues a firm generates from international sales, the more international the successor CEOs were (ignoring the 100% local firms, which had an 11% share of the total sample). According to the Resource-Based View perspective, more international CEOs offer a competitive advantage to international firms and therefore firms with higher international sales will have more incentives to attract more international managers.

§4.2

C

-

The average ages of the successor CEOs were nearly the same in both countries: 48.9 in Germany and 49.1 in the UK. In Germany there were no female CEOs appointed, while in the UK three female CEOs were appointed, which amounts to 5%.

Figure 8 shows that the percentage of CEO successors with foreign nationalities in the UK was 40%, the double of Germany’s 20%. Foreign educated CEO successors were 29% in the UK and 33% in Germany. Foreign work experience showed a larger difference, 59% in Germany and 78% in the UK. British firms appointed on average more international CEOs than German firms.

34

FIGURE 8:CEO SUCCESSOR INTERNATIONALITY (PER COUNTRY)

In both countries, CEOs are more often appointed from the inside than from outside of the firm (figure 9). 71% of German successor CEOs and 60% of British successor CEOs were insiders. Of these insiders, 30% of the German and 24% of the British insiders were classified as contenders. The outsiders in Germany had an average industry relatedness of 1.5, while in the UK it was 1.2. German firms seem to have preferred to initiate strategic change by appointing a contender CEO, while British firms chose to appoint an outsider. The Uncertainty Avoidance Index could explain this, as appointing an insider who is familiar with the board members and the firm itself is less risky than appointing an unfamiliar outsider.

FIGURE 9:CEO SUCCESSOR ORIGIN (PER COUNTRY)

20% 33% 59% 40% 29% 79% 0% 20% 40% 60% 80% 100% Foreign nationality Foreign education Foreign work experience

UK DE 49% 21% 30% 46% 14% 40% 0% 10% 20% 30% 40% 50% 60%

Follower Contender Outsider

35

FIGURE 10:%-OF TOTAL CEO SUCCESSIONS (PER CRISIS & COUNTRY)

Figure 10 shows that the amount of CEO successions in the UK during the crisis period was equal to the yearly average of 14%, while in Germany it was with 23%. Furthermore, figure 11 shows that the percentage of CEO successions in Germany were more concentrated around the crisis period, while in the UK this was more stable. Before the crisis, the percentage of CEO successions per year was lower in German than in the UK, but when the crisis started to unfold in late-2007, the amount of CEO successions in Germany doubled, while remaining fairly stable in the UK.

FIGURE 11:%-OF TOTAL CEO SUCCESSIONS (PER YEAR & COUNTRY)

36

FIGURE 12:CEO SUCCESSOR INTERNATIONALITY (PER CRISIS & COUNTRY)

German firms preferred a local market focus during the crisis by hiring less international CEOs, while for the British firms the opposite was true (figure 12). The amount of CEO successors with foreign nationalities in Germany decreased from 23% to 8% during the crisis, while in the UK it increased from 39% to 44%. Foreign educated successor CEOs decreased from 38% to 15% in Germany and increased from 26% to 44% in the UK. Successor CEOs with foreign work experience decreased in Germany from 64% to 39% during the crisis period and increased from 76% to 89% in the UK.

FIGURE 13:CEO SUCCESSOR ORIGIN (PER CRISIS & COUNTRY)

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Normal Crisis year Normal Crisis year

DE UK

Foreign nationality Foreign education Foreign work experience

0% 10% 20% 30% 40% 50% 60%

Normal Crisis year Normal Crisis year

DE UK

37 During the crisis year, there was an 8% shift from contenders to outsiders in Germany, while the percentage of followers remained constant at 50% (figure 13). In the UK there was large shift from followers to outsiders. German firms reacted more conservatively to the crisis, while the British firm reacted more aggressively.

The descriptive statistics of the firm characteristics show that the German manufacturing industry had a 64% share of the total German sample, while in the UK the manufacturing industry had a 43% share. The ‘diversified manufacturing’ industry was the largest in Germany (12% share), while in the UK the financial services industry was the largest (19% share). The average British firm market capitalization of €35,984 (£26,940) million was more than three times larger than the average German firm market capitalization of €8,800 million. Average British firm sales were €42,797 (£32,041) million and €14,967 million in Germany. This difference can be explained by the country corporate governance models, as Continental European firms are more often privately held, while in Anglo-Saxon firms usually have a full IPO as soon as possible (Cernat, 2009). Average firm internationality was almost identical (56%). Average total international sales however did show a large difference: €25,640 (£19,196) million in the UK and €9,195 million in Germany.

§4.3

R

§4.3.1DESCRIPTIVE STATISTICS

TABLE 2:DESCRIPTIVE STATISTICS INDEPENDENT VARIABLES (TOTAL SAMPLE) Variable Mean Median Std. Dev. Minimum Maximum

FN 0.29839 0 0.45941 0 1 FE 0.30645 0 0.46289 0 1 FWE 0.68548 1 0.46621 0 1 FIPERC 0.56265 0.65499 0.30783 0 0.98676 FIMON 14275.9 3992.3 28604.4 0 179166.4 MCAP 18023.4 6199.3 30084.2 26.0 159937.4 TOTSALES 23641.6 6635.5 44940.4 31.0 314173.9 CONTENDER 0.18548 0 0.39027 0 1

OUTSIDER (HIGH MATCH) 0.09677 0 0.29685 0 1

OUTSIDER (LOW MATCH) 0.25807 0 0.43935 0 1

IT 0.53226 1 0.50098 0 1

AGE 49.02 49 4.45659 39 63

CRISIS 0.18548 0 0.39027 0 1

38

TABLE 3:DESCRIPTIVE STATISTICS INDEPENDENT VARIABLES (UK SAMPLE)

TABLE 4:DESCRIPTIVE STATISTICS INDEPENDENT VARIABLES (DE SAMPLE)

Table 2, 3 and 4 represents the descriptive statistics for the independent variables of respectively the total, UK and Germany samples. The CEO successor internationality was measured by the FN, FE and FWE variables, dummy coded with 0 for local and 1 for foreign. FN measures the nationality dimension, FE the educational dimension and FWE measures the work experience dimension. FIPERC and FIMON measure

Variable Mean Median Std. Dev. Minimum Maximum

FN 0.39683 0 0.49317 0 1 FE 0.28571 0 0.45538 0 1 FWE 0.77778 1 0.41908 0 1 FIPERC 0.55829 0.67519 0.34088 0 0.98676 FIMON 19195.7 5664.6 36674.6 0 179166.4 TOTSALES 26939.7 8790.7 37395.1 1462.3 159937.4 MCAP 32041.2 12193.0 57577.4 124.6 314173.9 CONTENDER 0.14286 0 0.35274 0 1 OUTSIDER (HIGH MATCH) 0.09524 0 0.29590 0 1 OUTSIDER (LOW MATCH) 0.30159 0 0.46263 0 1 IT 0.42857 0 0.49885 0 1 AGE 49.14 49 4.67960 39 63 CRISIS 0.14286 0 0.35274 0 1

UK Sample (N = 63)

Variable Mean Median Std. Dev. Minimum Maximum

FN 0.19672 0 0.40082 0 1 FE 0.32787 0 0.47333 0 1 FWE 0.59016 1 0.49589 0 1 FIPERC 0.56717 0.61444 0.27229 0 0.91030 FIMON 9194.9 2494.0 15356.2 0 74338.99 MCAP 8814.7 2184.0 15513.7 26.0 82631 TOTSALES 14966.5 3712.0 23749.0 31.0 102114 CONTENDER 0.22951 0 0.42401 0 1 OUTSIDER (HIGH MATCH) 0.09836 0 0.30027 0 1 OUTSIDER (LOW MATCH) 0.21312 0 0.41291 0 1 IT 0.63934 1 0.48418 0 1 AGE 48.90 49 4.24933 40 59 CRISIS 0.22951 0 0.42401 0 1

39 the internationality of the firm. FIPERC is the %-of total sales measure and FIMON is the monetary amount of international sales measure. MCAP and TOTSALES are the firm size measures, with MCAP representing the market capitalization and

TOTSALES representing the total sales. The CEO successor origin is measured by the

dummy variables CONTENDER, OUTSIDER (HIGH-MATCH) and OUTSIDER

(LOW-MATCH). IT is a dummy variable for the industry type, AGE measures the age

of the successor CEO and CRISIS is a dummy variable representing the CEO successions during the crisis period of 01/09/2008 – 31/08/2009.

The variables FIMON, MCAP and TOTSALES showed very large differences between their minimum and maximum values. To reduce these ranges, the natural logarithm transformation was applied to these variable, in line with Daily et al. (2000).

§4.3.2CORRELATIONS

Appendix II shows that several independent variables have correlations in excess of the critical correlation value of 0.7 as recommended by Cooper & Schindler (2006). These are the correlations between:

 FIPERC - FIMON: both measure the internationality of the firm

 TOTSALES - FIMON: FIMON is a function of TOTSALES

 TOTSALES - MCAP: both measure the size of the firm

40

5. EVENT STUDY ANALYSIS

The first section of this chapter presents the analysis of the abnormal returns and the second section presents the differences between the country sub-samples. The third section presents the results of the cross-sectional regressions.

§5.1

A

Figure 14 graphs the (cumulative) average abnormal returns for the total sample. Appendix III.A and appendix III.B show the descriptive statistics of respectively the average abnormal returns (AARs) and the standardized average abnormal returns (SAARs). Appendix III.C presents the test statistics for the AARs in the event window [-15,+15].

FIGURE 14:(C)AAR- TOTAL SAMPLE

Table 5 presents the statistical tests for the CAARs during the various test windows. The z-scores are the Patell’s Standardized Residual (PSR) test-scores, the z-cross is the Boehmer’s cross-sectional standardized abnormal returns tesscore and the t-value is the standard t-test score. Event-induced variance is indicated by a significant Levene’s test. Z-cross is the correct test in the presence of event-induced variance, as the PSR and the t-test are misspecified when the event causes a significant increase in the variance of the abnormal returns.

41

TABLE 5:STATISTICAL TESTS OF THE (C)AARS (TOTAL SAMPLE)

The total sample showed a slightly positive stock market response to the CEO succession announcements with an average abnormal return (AAR) of 0.083% on the event day. However, all three test scores show that this AAR was not significantly different from zero. The increase of the standard deviation of the AARs from the average of 0.0229 to 0.0308 on the event day (appendix III.A), combined with the significant Levene’s test-score, indicates the presence of event-induced variance, sugesstint that thhis AAR was related to the CEO succession announcements.

The CAAR during [-2,+2] was 0.65% for the total sample. The z-cross statistic was not significant, but the z-score statistic was. Therefore, [-2,+2] window provides some support for H1 for the total sample. However, the Levene’s test indicates the possibility of event-induced variance in the event window of [-2,+2], suggesting possible misspecification of the z-score test.

The [-15,-1] CAAR of 1.84% received strong statistical support, suggesting information about the CEO successions leaked to the market before the official announcements. The CAAR during the event window of [+1,+15] was 1.20%, which was not significantly different from zero.

In the event window of [-15,+15] the CAAR was 3.13% (p<0.05). Combined with the results of the [0] and [-2,+2] event windows, H1 is weakly supported; the CEO succession announcements were positively related to the stock-price of the firms in the total sample. The CAAR-curve shows that this positive stock-market response was spread out over a longer event period and not concentrated on the event day as was predicted by the semi-strong Efficient Market Hypothesis (EMH).

Period CAAR Levene's test z-Score z-cross t-value

[0] 0.00083 13.0057*** 1.5668 0.9623 0.4360 [-1,+1] 0.00423 2.0564 1.5619 1.1815 1.2827 [-2,+2] 0.00653 2.3631* 2.0172** 1.4148 1.5324 [-15,-1] 0.01843 0.8018 3.2313*** 2.7945*** 2.4988** [-1,0] 0.00292 3.8276* 2.0970** 1.4868 1.0849 [0,+1] 0.00214 2.0216 0.6732 0.4470 0.7944 [+1,+15] 0.01200 1.8504* 0.6824 0.6402 1.6268 [-15,+15] 0.03126 1.9703*** 2.9401*** 2.5512** 2.9480*** Total sample (N=124)

42

Figure 15 graphs the (cumulative) average abnormal returns of the UK sample and table 6 presents the statistical test results for the various test windows.

FIGURE 15:(C)AAR-UK SAMPLE

TABLE 6:STATISTICAL TESTS OF THE (C)AARS (UK SAMPLE)

The flat CAAR-curve after the event day suggests that the information about the CEO successions was immediately incorporated into the stock price, in line with the semi-strong EMH. Prior to the event day, the CAAR-curve is sloped upwards, indicating information about the CEO succession leaking to the market. In addition, the CAAR of [-15,-1] showed an abnormal return of 2.94% (p<0.01). Levene’s test is negative

-1.00% -0.50% 0.00% 0.50% 1.00% 1.50% 2.00% 2.50% 3.00% 3.50% 4.00% 4.50% -15 -13 -11 -9 -7 -5 -3 -1 1 3 5 7 9 11 13 15 AAR CAAR

Period CAAR Levene's test z-Score z-cross t-value

[0] 0.00824 11.3717*** 3.7871*** 2.2046** 3.7327*** [-1,+1] 0.01293 5.7758*** 3.2153*** 2.5255** 3.3818*** [-2,+2] 0.01695 4.0445*** 3.3130*** 2.0643** 3.4322*** [-15,-1] 0.02941 0.8710 3.8454*** 3.1050*** 3.4389*** [-1,0] 0.01172 12.7460*** 4.1420*** 2.7864*** 3.7532*** [0,+1] 0.00946 4.7300** 2.4895** 1.6180 3.0280*** [+1,+15] -0.00099 1.3299 -0.2924 -0.3017 -0.1160 [-15,+15] 0.03666 2.3729*** 3.1556*** 2.7045*** 2.9818*** UK sample (N=63)