Netherlands and the impact of successor type, age and
tenure
Robert Wagenaar
University of Groningen
MSc Business Administration Finance
Master Thesis
July 2011
Abstract
This study examines the stock market reaction on the announcement of the replacement of the current CEO, using a database of 102 CEO turnovers in the Netherlands between 1997 and 2010. The results reveal that the announcement of a successive CEO results in a significant positive stock price reaction for the firm. When an outside successor is appointed, an even larger market reaction is found. Furthermore, for successors older than 55, a significant positive market reaction is found.
1. Introduction
1.1 Background
The chief executive officer (CEO) is the highest executive in an organization and in charge of the total management. When a CEO leaves the firm, a successor has to be appointed. The market reaction, in terms of the stock price of the firm, as result of the choice of the successor may differ significantly and depends on several variables, for example, it may matter whether he or she leaves the company due to retirement or is forced to leave. A forced CEO turnover may be interpreted as a bad signal about the company’s performance. One can also argue that the background of the successor is also important. The new CEO can be an insider who already works for the firm, or an outsider who has no past in the firm, and we may expect that the stock market values these backgrounds differently.
This study focuses on the relation between the appointment of a new CEO, called the successor and the market reaction as a result of the announcement of this succession. The event study methodology is used to test whether there are statistically significant abnormal returns on the day of the announcement. In addition, this study examines to what extent the variation in abnormal returns as a result of the new CEO announcement can be explained by several succession characteristics. To do so, the cross section of abnormal returns on the announcement days is regressed on a number of CEO and external characteristics, in particular, the successor type, the age of the successor and the tenure of the departing CEO.
The analysis uses a database consisting of 102 CEO succession announcements of Dutch listed firms between 1997 and 2010.
1.2 Literature review
In the literature CEO turnover is often discussed, particularly the market reaction on the announcement of the departure of the CEO and the performance the period after the announcement. Other studies discuss several succession characteristics of the successor like successor type. Cools and Van Praag (2004) found positive abnormal returns as result of the announcement of unanticipated forced departures of top executives in the Netherlands. However, the stock market reaction as result of the announcement of a successive CEO for Dutch listed firms and the influence of several succession characteristics is not discussed earlier.
returns (Davidson III et al., 1990; Murphy and Zimmerman, 1992). In this study the abnormal returns are measured using event study methodology.
Studies dealing with CEO succession found that the announcement of succession leads to abnormal stock returns. Lubatkin et al. (1989) found that in general succession conveys negative information to investors. Davidson III et al. (1990) found evidence for favorable stock market reaction to the announcement of CEO succession. This result is supported by Rhim et al. (2006), Borokhovich et al. (1996) and Dedman and Lin (2002). Furtado and Rozeff (1987) state that a change of top management signals shifts in the firm’s policy and therefore raise shareholders wealth. In addition, Cools and Van Praag (2004) report positive abnormal returns for the Netherlands, this is also the only paper dealing with CEO turnover in the Netherlands. Thus, the announcement of a successive CEO is expected to have a positive market reaction as result.
1.3 Succession characteristics
A number of successor characteristics have been found to affect the stock price reaction in the literature. These are listed below.
1.3.1 Insiders
In the literature, a distinction is made between successive CEOs that are insiders and those who are outsiders. An insider is usually defined as having at least a 5 year tenure, an outsider is considered to be no longer than one year in the company (Lubatkin et al., 1989). Although new CEOs are generally considered as insiders when they have at least a 5 year tenure in the company, an exception is made for new CEOs who are outsiders according to this definition, but have a long-‐lasting past in the firm, they are still seen as insiders.
Furthermore, recent research points out that inside succession occurs more often than outside succession (Rhim et al., 2006). Helfat and Bailey (2005) confirm that most firms select successors from inside the firm in order to minimize disruption.
When the firm appoints an insider as successor, there are two advantages: the insider has better knowledge about the company, its processes and its relations; the second advantage is that they already have the social network via which they acquire specific internal information (Dherment-‐Ferere and Renneboog, 2000). Cannella and Shen (2001) mention as disadvantage that insiders are limited by their ability to initiate strategic change because they are often selected and groomed by the departing CEO. Another disadvantage mentioned by Sheifler and Summers (1988) is that insiders are significantly constrained by their within-‐firm social networks.
In the case of inside succession, a distinction is made between followers and contenders, as used by Shen and Cannella (2002). These two types are discussed in the following sections.
1.3.1.1 Followers
In this study followers are formulated as successive CEOs who are appointed after the retirement of the former CEO, the friendly departure of the former CEO to another company or the case in which the former CEO stays active in the firm, but not in the position of CEO. Dherment-‐Ferere and Renneboog (2000) found declining stock prices in case of internal succession in times of bad performance of the firm. However, in most research it is assumed that inside succession occurs under conditions of good company performance and reflects the intention to maintain strategic continuity (Shen and Cannella, 1990). Indeed, Khurana and Nohria (2000) found that an insider who succeeds a CEO who left naturally (i.e. not forced) had little effect on the company performance. So since little strategic change is expected, the expectation is that when the successive CEO is a follower, there will be no abnormal returns.
1.3.1.2 Contenders
On the other hand, contenders are inside successors who are appointed after dismissal of the former CEO. Khurana and Nohria (2000) did not found any results for contenders, they point out that forced turnover creates a condition for change, but that an insider is futile to affect change. Shen and Cannella (2002) however state that some contender successions after CEO dismissal may reflect the outcome of power struggles within top management rather than an intention to initiate strategic change. Since the former CEO is dismissed, it is expected of the contender that he will follow another strategy than his predecessor. On the other hand, Borokhovich et al. (1996) found that shareholders are harmed when an insider replaces a fired CEO.
However, because the contender is likely to be charged with a mandate to initiate strategic change, it is expected that appointing a contender will results in a positive abnormal stock return.
1.3.2 Outsiders
fresh perspective and initiate strategic change (Shen and Cannella, 2002), it is expected that the announcement of a successive outside CEO leads to a positive abnormal stock returns.
1.3.3 Age
It is expected that the age of the successive CEO is of importance for the market reaction of the firm. The age of the departing CEO is often mentioned in the literature, however, few studies discuss the age of the new CEO. Davidson III et al. (1990) state that the appointment of a young person is often related with change and intention to grow, where more experienced and older successors may reflect the expectation of continuity of operations. They found that the market appears to react more positively to the announcement of a young successor. Therefore it is expected that the appointment of a young CEO will result in a positive abnormal stock return.
1.3.4 Tenure
Studies found that the tenure of the departing CEO has impact on the stock reaction of the firm. Shen and Cannella (2002) state that a long tenure can be associated with strong organizational inertia, which can cause difficulty when the successors wish to initiate strategic change. However, when the tenure of the departing CEO is too short, it is possible that the firm did not sufficiently recover from the disruption of the previous succession. In this study a long tenure will be defined as a tenure of more than 10 years. So it is expected that the announcement of a successive CEO who follows a CEO which had a long tenure will result in a negative abnormal stock return.
2. Methodology
2.1 Event study methodology
The market reaction as results of the announcement of the appointment of a new CEO can be measured using the event study methodology (Brown and Warner, 1980). This methodology relies on the assumption that expectations are incorporated in the stock price, so stock prices reflects all available information. Unexpected announcements, such as the announcement of the name of the new CEO, result in stock price movements since the information is not yet incorporated in the stock price. So the effect of the new information on the stock price can be measured.
In the event study methodology described by Brown and Warner (1980), Brown and Warner (1985) and MacKinlay (1997), the event day is the day that information about the announcement is available on the market. In this study the event day is the day the announcement of the successive CEO is made by the firm. If the announcement is made after the closure of the stock market, the next trading day is used as event day. If the announcement is made in the weekend, the Monday thereafter is used.
The period prior to the event is called the estimation window and is used as a benchmark to measure if the event period is surrounded by statistically significant abnormal returns. The estimation window used in this study is from day [-‐200; -‐2], where day 0 is the event day. The event window is the period surrounding the event for which is measured if any abnormal returns are observed. The event window used is from day [-‐1; 2]. By testing for abnormal returns on day -‐1 it is possible to see if there is any indication for information leakage. In the case the information of the announcement is leaked the day before the announcement, day -‐1 will show abnormal returns. In order to check for any abnormal returns on the days after the event, day 1 and 2 are also included in the event window.
In order to check if there are any statistically significantly returns over the whole event window, the cumulative average abnormal return (CAAR) is examined (Brown and Warner 1985). The CAAR tests for abnormal returns for periods in the event window. In this study the periods of days [-‐1;2], [-‐1;0], [0;2], [1;2] are tested. By testing multiple periods, it can be observed in which period of the event window abnormal returns are present. If the results of the CAAR are statistically significantly, it can be stated that the event is surrounded by abnormal returns.
2.2 The market and risk adjusted model
(1980). The market and risk adjusted model takes into account both market-‐wide factors and the systematic risk of each sample security.
As market index, the MSCI Netherlands is used for all events, since all the firms in the database are listed in the Netherlands. The beta of the stock reflects the volatility of the return of the stock compared with the volatility of the return of the market index in the estimation window. The beta is calculated by taking the covariance between the stock returns and the market returns and dividing that covariance by the variance of the market return. The alpha gives the specific or individual risk of the stock and is obtained by taking the average of the daily returns of the stock in the estimation window and subtract from that the beta times the average of the daily returns of the market index in the estimation window.
The returns of the stocks in the estimation window are considered to be the normal returns. These normal returns are compared with the stock returns in the event window in order to search for abnormal returns in the event window.
The abnormal return for a specific day is obtained by subtracting the alpha and the beta times the stock market return from the stock return of the security for that day. Averaging the returns for all events for each day in the event window gives the average abnormal returns (AARs) for each day in the event window.
The next step is to test if these AARs are statistically significant from zero. In this way it can be measured whether the announcement of a new CEO results in an abnormal return in the event window. To test if the AARs in the event window are statistically significantly different from zero, Student’s t-‐test is used (Newbold et al., 2007).
The robustness of the results using the market and risk adjusted model are checked by comparing the results with the results using the mean adjusted model and the market adjusted model, as described by Brown and Warner 1980 and 1985.
2.3 Ordinary least squares approach
To examine to what extent variation in the abnormal returns is determined by the several succession characteristics, the ordinary least squares (OLS) method is used as described by Brooks (2008). For the three types of successor types, dummy variables are created for the regression. Also for the announcements for which no other information is available, and for the simultaneous announcement of the departure of the old CEO and the appointment of the new CEO, dummy variables are created.
3. Data and descriptive statistics
3.1 Data
The event examined in this study is the announcement of the name of the new chief executive officer (CEO) of a Dutch listed firm. The database consists of 102 of these announcements in the period 1997-‐2010. Since 2005 Dutch firms are required to publish information which can be sensitive to the firms stock. This information is published at the website of AFM, Autoriteit Financiële Markten is the financial services regulatory in the Netherlands. So CEO changes are easy to find using this information. Before 2005 press releases of CEO changes are available at the website of Euronext. The exact date of the announcement relevant for the study can be checked by searching for the announcement at the website of Het Financieele Dagblad. Usual the announcement day used is the day the announcement is published in the newspapers, since this is the day the stock market reacts on the news (Cools and Van Praag, 2004). The succession characteristics can usually be found in the official company announcement. When this is not the case, the website Managementscope is useful, since it contains information about top executives of Dutch companies.
In appendix 1, the list of events can be found. It includes the company, the date of the event, the name of the new CEO, the successor type, the age of the new CEO, the tenure of the old CEO and the abnormal return on the announcement day.
The stock prices of all the relevant companies and the market index (MSCI Netherlands) are found at Datastream.
A distinction is made between events for which the announcement of the departure of the old CEO is at the same day as the announcement of the new CEO and announcements for which only the information about the new CEO is released. When both announcements are made on the same date, the market reaction of solely the new CEO announcement cannot be measured since the market reaction on the departure of the old CEO is incorporated in the stock price on the same day. Therefore these events are labeled with a dummy to examine if the AARs are different for such events.
Table 1 shows the main characteristics of the 102 events. For most of the announcement of the new CEO, no other information about the firm has come into the market. This is the case for 90 events, which have an AAR for the announcement day of 1.04%. When the announcement of the new CEO comes on the same day as the announcement of the departure of the old CEO, the AAR is 2.63%. However, this is only the case for 12 events.
who found that inside succession occurs more often than outside succession. The AAR on day 0 is the highest for outsiders, namely 3.1%. The AAR of followers and contenders seems to be more neutral, respectively 0.2% and -‐0.5%.The AAR for insiders, so both followers and contenders, is 0.0%.
The vast majority of the CEOs (69) are between the age of 45 and 55 when they are appointed; the AAR on day 0 for this group is 1.1%. For CEOs older than 55 the abnormal return is 2.5%. News of the appointment of relatively young CEOs with an age of below 45, results in a negative market reaction of -‐0.6%.
In order to see the distribution of tenures of the departing CEOs, three tenure groups are created; 1-‐5 years, 6-‐10 years and departing CEOs which had a tenure of more than 10 years. Most departing CEOs had a tenure of in between 1 and 5 years (52.9%), this is shown in table 1. All tenure groups have positive AARs on day 0 in the event window, namely respectively 1.0%, 1.4% and 1.7%. So it can be observed that for CEOs whose predecessor had a tenure of more than 10 years, the return on the announcement of the new CEO is the highest (1.7%).
Table 1. Main characteristics of the events and the relevant successive CEO.
Number of
events 102
Available information Percentage Return on day 0 Departure old CEO 12 11.8% 0.026 No other information 90 88.2% 0.010 Successor type Outsider 42 41.2% 0.031 Follower 40 39.2% 0.002 Contender 20 19.6% -‐0.005 Insiders (followers and contenders) 60 58.8% 0.000 Age distribution <45 11 10.8% -‐0.006 45-‐55 69 67.7% 0.011 >55 22 21.6% 0.025
Tenure old CEO in years
1-‐5 54 52.9% 0.010
6-‐10 33 32.4% 0.014
>10 15 14.7% 0.017
In table 2, the number of successor types per tenure group is shown. It is striking that of 18 of the 20 departing CEOs had a tenure of 5 years of less, in case the successor was a contender. Since a contender is the successor of a dismissed CEO, it is not completely unexpected that the departing CEO had a relatively small tenure. The outsider and followers groups seem to have approximately the same distribution in tenure groups.
Table 2. Number of successor types per tenure group.
Tenure distribution per type Type Outsiders Followers Contenders Total
Tenure 1-‐5 17 19 18 54 6-‐10 16 15 2 33 >10 9 6 0 15 42 40 20 102 3.2 Descriptive statistics
Table 3 shows the descriptive statistics of the returns in the estimation window, using the market and risk adjusted returns model. The mean return in the estimation window is -‐0.02%, which is slightly negative, the standard deviation of the mean returns is 0.3%. The reason for the negative mean return is the presence of the company Ahold in the database. In 2003 it was announced that Ahold was guilty of significant internal accounting and reporting failures and poor public disclosure of market-‐sensitive information (Clark, Wójcik & Bauer 2005). This information resulted in large negative returns on certain days in the estimation window, due to this the overall mean return is slightly negative. The mean returns in the estimation window are all between -‐1.06% and 0.74% per day.
Table 3. Descriptive statistics of the returns in the estimation window.
Mean -‐0.0002 Median 0.000 Standard deviation 0.003 Minimum -‐0.011 Maximum 0.007 Skewness -‐0.276 Kurtosis 0.101 Jarque-‐Bera 44.991 *
* denotes that the distribution is not normal.
CEO age of 52.5. The youngest successive CEO in the database was 36 years old, the oldest 68. As expected most CEO’s (69 of the 102) were between 45 and 55 years old, as is shown in table 1.
The tenure of the departing CEO was on average 7.32 years. This is comparable with the tenure of 6.7 years which Kaplan and Minton (2006) found for their sample of US listed companies. Weisbach (1988) found higher tenures: 11.7 years for insider-‐dominated boards and 9.7 years for outsider-‐dominated boards, Zajac (1990) found an average tenure length of 8 years.
The longest tenure in the database was 41 years and the shortest one year, noting that all tenures less than one year are rounded upward.
Table 4. Descriptive statistics of age and tenure.
Age Tenure Mean 50.93 7.32 Median 51.00 5.00 Standard deviation 5.95 7.07 Minimum 36.00 1.00 Maximum 68.00 41.00
Table 5 shows the descriptive statistics for the age of the new CEO and the tenure of the new CEO per successor type. For the variable age it is notable that the average age of a contender is 53.5 years old, which is slightly older than for the other two successor types (on average 50.6 and 50.0 years old). For the variable tenure it is striking that the average tenure of the departing CEO is only 3.6 years in case the successor is a contender, while the average of all the events is 7.3. A logical explanation for this is that contenders succeed dismissed CEOs, who of course have a tenure less long than CEOs who for example retire. The maximum tenure for contenders is 10 years, a lot shorter than for outsiders (41 years) and followers (31 years). The average tenure for departing CEOs with outside succession is with 8.9 years relatively long. A possible explanation can be that a firm appoints an outsider after the departure of a CEO with a long tenure in order to initiate strategic change after a long period of the same policy.
Table 5. Descriptive statistics per successor type: age and tenure.
Age and successor type Tenure and successor type Outsider Follower Contender Outsider Follower Contender
Figure 1 gives the number of events in each year over the period of 14 years (1997 to 2010) and the return of the AEX index, which is representative for the performance of the Dutch economy. However there is no clear evidence, there seems to be relatively more CEO turnovers in the years surrounding a crisis; namely the dot-‐com bubble (2000-‐ 2002) and the credit crunch (2008). This is supported by Erkens et al. (2010), who found an increase in CEO turnover during the credit crisis for especially financial firms, but also for non-‐financial firms. Figure 1 shows that relatively a lot of CEO succession took place in years in which the AEX index performed poor.
Furthermore, in the first 7 years 45 announcements were made, in the subsequent 7 years the number of announcements made was 57.
In appendix 2, 3 and 4, additional descriptive statistics are available.
Figure 1. Distribution of the events over time and per year.
4. Results
4.1 Average abnormal returns
In this section the results of the event study and the regressions are reported. Also, the robustness of the results is checked by comparing the results with different methods.
The AAR on the event day for all events is 1.2%, which is statistically significant (p < 0.05). This indicates that in general the announcement of a new CEO is perceived as positive news, which has an increase in the stock price as result. The literature is not completely clear on the stock market reaction after a new CEO announcement. Lubatkin et al. (1989) found negative abnormal returns, where Davidson III et al. (1990) and Rhim et al. (2006) found positive results. Since Cools and Van Praag (2004) also found positive abnormal returns for the Netherlands, the results of this event study are as expected.
Another method would be to use a test statistic that aggregates standardized abnormal returns, which means each event is weighted in inverse proportion of the standard deviation of the estimated abnormal return (Kothari & Warner, 2004). So for each event, the normalized abnormal return on day 0 in the event window is estimated. This method confirms the results of the originally used method and also gives a highly significant positive AAR on day 0. The results are available on request.
The returns for the other days in the event window are shown in table 6. All the days in the event window show positive results, however only day 0 in the event window is statistically significant. This suggests that the announcements were unexpected and there is no material information leakage which causes significant returns the day before the announcement.
Table 6. Average abnormal returns in the event window.
Day AAR P-‐value -‐1 0.003 0.261 0 0.012 0.000 1 0.003 0.332 2 0.005 0.093
Table 7 shows the results of the cumulative average abnormal returns (CAAR) in the event window for different time spans. For all the examined periods the results were positive and statistically significant. Over the whole event window, the cumulative average abnormal return is 2.3%. So in general it can be stated that the new CEO announcements of Dutch listed firms are surrounded by positive stock returns and investors see an announcement of a new CEO as a positive event.
Table 7. Cumulative average abnormal returns in the event window. CAAR P-‐value [-‐1;2] 0.023 0.000 [-‐1;0] 0.016 0.000 [0;2] 0.020 0.000 [1;2] 0.008 0.009
4.2 The relationship between succession characteristics and the abnormal returns
In table 8 the results of the regression of abnormal returns on a number of CEO characteristics are presented. In particular, the abnormal return on day 0 of the event window is the dependent variable in the regression, since this seems to be the relevant variable according to the event study in the previous section. The independent variables are the successor type (outsider, follower and contender), age, tenure, and the situation where no other firm specific information is available on the event day and one where the departure of the old CEO is announced on the same day.
As can be seen in table 8, the dependent variable is relatively poorly described by the independent variables. The adjusted R² for model 8 is only 11.9%, indicating that 11.9% of the total variability of the return on day 0 is explained by the model.
Furthermore, for model 8, the only independent variable which is statistically significant on a 5% level is outsider. This indicates that the new CEO being an outsider in the firm has a significant impact on the return on day 0 of the event window. The appointment of an outsider as successive CEO causes an average abnormal return of over 3%.
Table 8. OLS Regression model for determinants of the abnormal returns on the announcement day. The coefficient is given for each variable; between braces the p-‐values are showed.
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Outsider 0.031 0.034 0.038 (0.001)*** (0.004)*** (0.003)*** Follower 0.006 0.002 0.005 0.003 (0.724) (0.910) (0.815) (0.868) Contender -‐0.007 -‐0.013 -‐0.004 -‐0.008 -‐0.014 (0.687) (0.488) (0.828) (0.718) (0.479) Age 0.000 (0.126) <45 -‐0.006 -‐0.004 -‐0.020 (0.718) (0.874) (0.357) 45-‐55 0.011 0.013 0.015 -‐0.021 (0.105) (0.404) (0.435) (0.131) >55 0.025 0.026 0.027 (0.044)** (0.141) (0.224) Tenure 0.000 (0.680) 1-‐5 0.010 -‐0.004 -‐0.006 0.008 (0.193) (0.842) (0.751) (0.664) 6-‐10 0.014 0.000 -‐0.004 0.001 (0.178) (0.983) (0.826) (0.973) >10 0.017 -‐0.001 (0.269) (0.949) No other information 0.010 -‐0.026 0.008 (0.088)* (0.111) (0.651) Departure old CEO 0.026 0.014 0.030 (0.113) (0.452) (0.214) R² 0.078 0.022 0.002 0.008 0.106 0.024 0.030 0.119
4.3 Successor type
In this section the AARs on day 0 are discussed for the different successor types, the results are given in table 8, model 1.
4.3.1 Insiders
Both followers and contenders are inside successors, these groups are separately discussed in the next sections. For the two groups together, so for all the inside successions, the average return on day 0 is 0.000% (see table 1). So there is no statistically significant abnormal return for inside succession. This does not support the founding’s of Davidson III et al. (1990), who found a relation between inside succession and increased firm performance and Borokhovich et al. (1996), who found positive stock returns for inside succession. However, Lubatkin et al. (1989) found that inside succession has negative stock returns as result.
4.3.1.1 Followers
The announcement of a follower as new CEO has little effect on the stock price (0.6%) and is not statistically significant. This result is in line with the expectations. A possible explanation for this result is probably that followers are expected to follow the policy of their predecessors. Another possible explanation is that it often occurs that the former CEO left the firm due to retirement, which is expected, and so it can be the case that the succession of the announced follower was already expected. This result is in line with that of Khurana and Nohria (2000), who found that an insider who succeeds a CEO who left natural (so not forced) had little effect on the company performance.
4.3.1.2 Contenders
The announcement of a contender results in a small negative return of -‐0.7%. However, this result is not statistically significant. Since a positive abnormal return was expected, this result is not line with the expectations. Contenders follow after dismissal of the former CEO and so they are appointed in difficult times. A possible explanation for the (small) negative reaction can be that investors prefer an outsider after dismissal of a CEO who is likely not to follow the policy of the former CEO. Since a contender was already in the firm, the change of strategic change is likely to be smaller with the appointment of a contender than with the appointment of an outsider. Another explanation given by Shen and Cannella (2002) is that some contender successions after CEO dismissal may reflect the outcome of power struggles within top management rather than an intention to initiate strategic change. This null result however is in line with the findings of Khurana and Nohria (2000).
4.3.2 Outsiders
appointed after dismissal of the old CEO, when firms are likely to perform poor. Shen and Cannella (2002) point out that appointing an outsider is an indication that the firm has the intention to initiate strategic change. This is an explanation for the positive AARs as a result of announcing an outsider as new CEO. Davidson III et al. (1990) and Rhim et al. (2006) found that inside succession results in higher abnormal returns than outside succession. This results contradicts my founding’s. However, Dherment-‐Ferere and Renneboog (2000) found a positive abnormal return of more than 2% for outsiders. Lubatkin et al. (1989), Kang and Shivdasani (1996) and Borokhovich et al. (1996) also found positive results in case of outside succession. So the results are in line with the existing literature.
4.4 Age
Model 8 of table 8 shows that in the regression, the age of a new CEO is not statistically significant. However, model 2 of table 8 shows that when the age group is split up in three groups, some statistically significant results are obtained. The successive CEOs with an age of over 55 have a statistically significant result on day 0 in the event window, the average return of this group is 2.5% (p < 0.05). The groups of CEOs under the age of 45 and the group CEOs who were between 45 and 55 years on the time of their announcements show no statistically significant results. This is not the same result as Davidson III et al. (1990) found. They found that the market appears to react more positively to the announcement of a young successor. Table 8 shows that for the group of new announced CEOs with an age of less than 45 years old, a negative abnormal return of -‐0.6% is found. This is not in line with the expectation that the appointment of a young CEO results in positive abnormal returns. A possible explanation for this result can be that younger CEOs have less experience, resulting in a negative market reaction as result of the announcement of a young CEO.
4.5 Tenure
It was expected that the departure of a CEO with a long tenure would result in negative AARs on the announcement day of the new CEO. This was expected due to the statement of Shen and Cannella (2002) that a long tenure can be associated with strong organizational inertia, which can cause difficulty when the successors wish to initiate strategic change. The results are not in line with this expectation. Model 3 of table 8 shows that for all the three groups positive AARs of more than 1% on the announcement day are found. However, none of these results are statistically significant. Since the results for the different tenure groups do not differ much and are also statistically insignificant, it can be stated that the tenure of the former CEO has no significant impact on the market reaction of the newly announced CEO.
4.6 Information type
A distinction is made between announcement for which no other relevant company information is made public on the announcement day of the new CEO, and events for which the announcement of the departing CEO and the successive CEO are made on the same day. When both announcements are made on the same day, the effect of only the announcement of the new CEO is not measurable. Model 4 of table 8 shows that the AAR of the announcements where also the departure of the old CEO is announced is 2.6%, which is higher than the announcement for which no other information is available (1%). However, only 12 of the 102 events are double announcements, and the AAR of 2.6% is not statistically significant. The results for the announcement for which no other relevant company information is made public on the announcement day of the new CEO (which is the case for 90 of the 102 events) are statistically significant (p < 0.10).
4.7 Robustness
In order to check if the used risk adjustment method is appropriate, the robustness of the model is checked by comparing the results of the market and risk adjusted model with the results of two other models used in event studies. Brown and Warner (1980) point out that it is useful to compare your results with other risk adjustment measures used in event studies.
The compared methods are the mean adjusted model and the market adjusted model, described by Brown and Warner (1980), Brown and Warner (1985) and MacKinlay (1997).
The first model used to check the robustness is the mean adjusted model. The abnormal returns of the mean adjusted model are calculated by subtracting the average return in the estimation window from the every day returns. So basically it looks at the daily deviation from the average return of the stock in the estimation window. In table 13 it is shown that the model gives an AAR of 0.9% on day 0 in the event window, which is statistically significant.
The market adjusted model determines it average abnormal return by subtracting the relevant daily market return1 from the daily return. In this way the daily stock returns are compared with the market movement. Table 9 shows an AAR of 1.2% on day 0 in the event window, which is statistically significant.
Comparing the results of the mean adjusted model and the market adjusted model with the results of the market and risk adjusted model (see table 9), it becomes clear that the AARs are of the same order of magnitude. So it can be stated that the results of the used model are robust.
Table 9. Average abnormal returns in the event window for the mean adjusted model, the market adjusted model and the market and risk adjusted model in order to check the robustness of the results.
Day Mean adjusted model Market adjusted model Market and risk adjusted model
AAR t-‐test P-‐value AAR t-‐test P-‐value AAR t-‐test P-‐value
-‐1 0.002 0.505 0.002 0.627 0.003 0.261
0 0.009 0.001*** 0.012 0.000*** 0.012 0.000***
1 0.000 0.971 0.003 0.373 0.003 0.332
2 0.006 0.033** 0.005 0.127 0.005 0.093*
5. Conclusion
This study examines CEO successions in the Netherlands from 1997 until 2010. In particular, new CEO announcement of 102 Dutch listed firms are used to check its impact on the average abnormal return on the day of the announcement of the successive CEO. On the day of the announcement the results are statistically significant with an AAR of 1.2%. So when a Dutch listed firm announces a new CEO, their stock price increases with 1.2% on average on that day, indicating that shareholders interpret the announcement as good news. No statistically significant AARs are found for the other days in the event window, indicating no material information leakage. For the entire event window (day -‐1 to day +2), a cumulative average abnormal return of 2.3% is found. This result shows that the announcement of a new CEO is surrounded by positive stock price returns, which is in line with the existing literature. Several studies suggest that a change in top management signals positive shifts in the firm’s policy and therefore raise shareholders wealth.
The results are robust to several different specifications.
In addition, the successive CEOs are split up in three successor types: outsiders, followers and contenders. For the CEO announcements of outsiders, a statistically significant AAR of 3.1% is found. For the followers and contenders, the AARs are respectively 0.2% and -‐0.5%. These results are however not statistically significant. So the announcement of an outsider as successive CEO results in a market reaction of 3.1%. This result is in line with the literature, where also positive results are found for outsiders. The reason for this positive AAR is probably that appointing an outsider is an indication that the firm has the intention to initiate strategic change, which is received as positive news by the market. One could say that contenders signal negative news based on the negative sign of the abnormal returns. However, as the result is statistically insignificant, no hard conclusion can be drawn with respect to this finding. Other succession characteristics which are investigated are the tenure of the former CEO and the age of the new CEO. For the tenure of the former CEO, no statistically significant results are found. However, for the group successive CEOs with an age of 55 years and older, a statistically significant AAR of 2.5% is found. So the market reacts more favorable if an old CEO is appointed, this indicates that successive CEOs who have more experience are preferred over a younger, more inexperienced CEO.
References
Berry, T.K., J.M. Biziak, M.L. Lemmon and L. Naveen, 2000. CEO turnover and firm diversification, Working Paper, Portland State University, Portland, OR.
Borokhovich, K.A., R.P. Parrino and T. Trapani, 1996. Outside directors and CEO selection,
The Journal of Financial and Quantitative Analysis, 31-‐3, 337-‐355.
Brooks, C., 2008. Introductory Econometrics for Finance. Cambridge University Press, New York.
Brown, S.J. and J.B. Warner, 1980. Measuring Security Price Performance, Journal of
Financial Economics, 8, 205-‐258.
Brown, S.J., and J.B. Warner, 1985, Using Daily Stock Returns, Journal of Financial
Economics 14, 3-‐31.
Cannella, A.A., Jr. & W. Shen, 2001. So close and yet so far: promotion versus exit for CEO heirs apparent, Academy of Management Journal, 44, 252-‐270.
Clark, G.L., D. Wójcik and R. Bauer, 2005. Corporate Governance, Cross-‐Listing and Managerial Response to Stock Price Discounting: Royal Ahold and Market Arbitrage – Amsterdam and New York, 1973-‐2004, Economic Geography Working
Papers, Working Paper, 05-‐07. School of Geography, Centre for the Environment,
University of Oxford.
Cools, K. and C.M. Van Praag, 2004. The value relevance of forced management departures: Dutch evidence, working paper,
ftp.tinbergen.nl/discussionpapers/03051.pdf.
Davidson III, W.N., D.L.Worrell and L. Cheng, 1990. Key executive succession and stockholders wealth: the influence of successors origin, position, and age, Journal of
Management 16(3), 647-‐664.
Dedman, E. and S.W.J. Lin, 2002. Shareholder wealth effects of CEO departures: evidence from the UK, Journal of Corporate Finance, 8, 81-‐104.
Dherment-‐Ferere, I. and L. Renneboog, 2000. Share price reactions to CEO resignations and large shareholder monitoring in listed French companies, Center for Economic
Research, 70, 1-‐32.
