1
The impact of reputation on the
relationship between performance and CEO
turnover
Name: René Koning Student number: 6148158 Date: August 13, 2015 Master Thesis
Name supervisor: A. Sikalidis Word count: 10,439
2 Abstract
This paper examines the effect of CEO reputation on the relationship between firm performance and CEO turnover. This is examined by an empirical archival study with a regression sample of 4,128 observations of US companies between 1998 till 2012. The hypothesis of this paper stated as: poor performing firms are less likely to replace a CEO with a good reputation. The performance measures were addressed by Return on Assets (ROA) as accounting performance and Stock Return as market performance. The results showed only a positive significance for the industry-adjusted ROA, which contradicts to prior literature negative between firm performance and CEO turnover. Only the coefficient of the interaction between Stock Performance and CEO reputation shows a negative significant relation to CEO turnover. This suggests that high CEO reputation weakens the relationship between market performance and CEO turnover. The paper found a negative moderating effect of reputation on market performance and CEO turnover.
3 Table of content
1. Introduction ... 4
1.1 Background ... 4
1.2 Research question ... 6
1.3 Motivation & Contribution... 6
1.4 Structure of the paper ... 7
2. Literature review and hypothesis ... 8
2.1 Agency theory ... 8
2.1.1 Separation of ownership and control ... 8
2.1.2 Moral Hazard and Adverse Selection ... 9
2.1.3 CEO performance and compensation ... 10
2.2 CEO turnover ... 11
2.2.1 Managerial ability and scapegoat theory ... 11
2.2.2 Performance and CEO turnover ... 12
2.3 Other factors influencing CEO Turnover ... 13
2.4 CEO Reputation ... 14
3.1 Sample selection and Research methodology ... 17
3.1.1 Identifying Reputation ... 18
3.1.2 Identifying Turnover ... 18
3.1.3 Identifying Performance ... 21
3.2 Regression model and proxies ... 21
3.2.1 Expectations regarding hypotheses ... 23
4. Results ... 24 4.1 Descriptive Statistics ... 24 4.2 Regression results ... 26 4.3 Sensitivity analysis ... 28 5. Conclusion ... 29 6. References ... 30 Appendix ... 32
4 1. Introduction
1.1 Background
According to the agency theory there’s a conflict of interests between the principal and the agent. This conflict is due to the separation of ownership and control, which leads to two different parties with their own interests, the principal and the agent. In this case the principal are the shareholders and the agent is the CEO. This study doesn’t make a distinction between male and female CEOs and will use the male term he for both male and female CEOs. The CEO, who has an information advantage over the principal, can act in his own interests (reaching targets, maximal reward with minimal effort) instead of the interests of the
shareholders (to maximize firm performance). This goal incongruence or conflict of interests and behavior causes the agency problem. For the organization costs are involved to get goal congruence by stirring the agent in the desired direction. These costs to solve the agency problem are agency costs. Jensen and Meckling (1976), describe multiple forms of agency costs like monitoring, bonding, residual loss and incentives. One example of agency costs is incentivizing a CEO. If an organization wants to give the CEO an incentive, due to a bonus, it has to set up performance measures tied to that bonus (Conyon, 2006 and Bebchuck and Fried, 2006). For a performance measure to be informative, Indjejikian (1999) argues, it has to reflect the actions of the CEO. If this is the case organizations can effectively evaluate and monitor the performance of the CEO. Performance measures are therefore used for two purposes: to incentivize the CEO and to evaluate the CEO. Organizations set performance measures because when an organization appoints a new CEO, his talents are unknown (Bushman et al. 2010). With these measures organizations can figure out if the new CEO is a good one for the firm. Therefore organizations use compensations which are attached to performance measures to evaluate the competence of the CEO. (Conyon 2006 and Bushman et al. 2010). It seems that organizations use the performance of the CEO to evaluate the overall performance of the firm. This assumption is consistent with Puffer and Weintrop (1991) who argued that the expectations of the corporate performance are used by judging the CEO performance. After evaluating the performance and actions of the CEO the organization will make a decision about the future of the CEO at the firm. Performance seems to be an important aspect for the CEO turnover decision.
Prior literature suggests a relationship between firm performance and CEO turnover (for example Warner et al. (1988), Puffer and Weintrop (1991), Kato and Long (2006) and Hartono (2011)). Warner et al. (1988) studied the outcome of performance (measured by
5 stock market performance) on the decision to replace or retain the top management. They found an inverse relation between the probability of a top management change and Stock Performance. Furthermore Puffer and Weintrop (1991) found support for their hypothesis. They hypothesized and found that there is a negative relationship between corporate
performance (measured by stock and accounting returns) and CEO turnover. This implies that when the firm has poor performance, the more likely it is that a CEO gets fired. Thereby effect on turnover was stronger when the expectations of the board were more reflected in the performance measures. Kato and Long (2006) also suggest that the increased executive turnover in China could be a mechanism of infusing new blood into the firm to turnaround a firm’s poor performance. They only found limited evidence but they point out to the study by Xu et al. (2005) that after a replacement of the CEO, companies experience performance improvements. Hartono (2011) tested the relationship in Indonesia (a non-western country). He showed that accounting performance has a significant negative influence over the turnover decision. However he didn’t found that accounting performance was influenced by turnover. In summary, most prior literature indicates a negative relationship between performance and turnover.
Moreover several researches are set up with the assumption that other factors, the so-called moderators, could influence the prior negative relationship between performance and CEO turnover. An example is the study of Weisbach (1988), which examines the
performance-turnover relationship with the factor corporate governance. For corporate governance he uses the independence of the board of directors as a proxy. He determines boards with more outside directors to be more independent from the CEO. The paper
concludes that the greater the proportion of outside directors on the board, the more likely it is for a CEO to get fired when the firm performance is bad. This is due to the higher
independency of the board, which makes it harder for the CEO to exercise his power over the board. But also the decision about the future of a CEO will be taken more objectively by an independent board. For the factor corporate governance, Weisbach (1988) documented that the fraction of outsiders on the board weakens (i.e., negatively moderates) the negative relationship between performance and CEO turnover.
6
1.2 Research question
The previous section described the relationship between performance and CEO turnover. Furthermore it was discussed that other factors can influence the performance-turnover relationship. An interesting factor that might influence the relationship between performance and CEO turnover could be CEO reputation. CEOs are claimed to build reputation following a series of favorable performance achievements which might insulate them from dismissal when performance is poor. Or alternatively stated, are CEOs with little reputation dismissed sooner when there is poor performance. This study will aim at CEO reputation as a potential factor that might influence the relationship between performance and CEO turnover. To examine the influence of CEO reputation on the relationship between performance and CEO turnover, this research focuses on the following research question:
How does CEO reputation affect the relationship between firm performance and CEO turnover?
1.3 Motivation & Contribution
The subject of this research is not extensively studied yet. Though, many studies examined the relationship between performance and CEO turnover, while looking at alternative measures of performance. Next to that a few studies looked at variables influencing the relationship between performance and CEO turnover, the so-called moderators. But only little research looked at CEO reputation as a moderator for the relationship between performance and turnover. This study will contribute by examining how CEO reputation affects the relationship between performance and turnover.
Allgood and Farrell (2000) previously examined the moderating factor tenure on the performance-turnover relation. They examined if the duration of working within the firm (tenure) by a CEO affects the turnover decision when there’s poor performance. Likewise the study of Allgood and Farrell (2000), this research will contribute by adding more insights to existing literature about a possible moderator. Incorporating the factor CEO reputation will give this research an extra dimension, because a moderating effect on the performance-turnover relation will be examined.
From a societal point of view it is interesting to see what the effect of CEO reputation on the performance-turnover relationship is. In one case a CEO might get fired and in the other case a CEO continues his work despite the same poor performance. It is interesting to examine how CEO reputation influences the turnover decision. Henderson et al. (2006) argue
7 that during the tenure of a CEO the evaluation of the annually performance will become less important. This can be due to a good relationship between the CEO and the organization or with the board. But also characteristics of the CEO can play a role by the dismissal. So besides performance, an organization might look at other aspects by making decisions about the future of the CEO. For the outside world it’s strange that in one case the CEO gets fired and in the other not. This is because the public doesn’t know the characteristics of the CEO and what’s going on in the organization behind the scenes.
1.4 Structure of the paper
This paper uses the following structure. The second chapter describes the literature and the hypothesis. The third chapter covers the research methodology and the sample selection that is provided. The fourth chapter covers the empirical results. The fifth chapter covers a conclusion and the limitations of this research.
8 2. Literature review and hypothesis
2.1 Agency theory
2.1.1 Separation of ownership and control
The agency theory describes the separation of ownership and control in organizations. This separation leads to two parties and to an information asymmetry. On the one hand are the shareholders who represent the capital and the ownership. On the other hand are the
executives who are in control of the firm activities. The separation of ownership and control has advantages and disadvantages. According to Fama and Jensen (1983) organizations can survive due to the separation of decision and risk-bearing functions. This might be because the CEO can focus only on business activities and doesn’t have to take his risk into account. Thereby a CEO is willing to take riskier (i.e. better) decisions for the company because when it turns out to be a bad decision, the risk is shared among the shareholders. Fama and Jensen (1983) claim that these organizations benefit of specialization of management and risk-bearing. Though the agency problem caused by the separation of decision and risk-bearing functions need to be controlled effectively. One important factor is that management has to deal with the specific knowledge of the agents at all levels. By delegating decision instead of allocating decision to the agents the organization benefits from better decision making. It is the task of the management to collect valuable information for decision making. Collecting relevant specific information is costly to transfer. Most complex organizations are risk-bearing due to the diffused knowledge among all agents in the organization. Therefore, according to Fama and Jensen (1983), it has advantages to have many residual claimants in complex organizations because the total risk which is large can be shared among them. In summary, the separation of ownership and control can lead to better decision making (due to specialization) and can be supported due to risk-sharing.
Though there are also disadvantages in the separation of ownership and control. In their paper, Jensen and Meckling (1976), describe the managerial behaviour, agency costs and ownership structure in firms. The authors argue that an agency problem can arise when two parties have different goals in an organization. These two parties are called the principal and the agent. In the setting of this research the principal-agent relationship is the relationship between the shareholders (principal) and the CEO (agent). The CEO has an information advantage over the shareholders due to the fact that the CEO is more entrenched in the organizations’ operations and therefore has more specific knowledge. Because the actions of the CEO has consequences for the shareholders (i.e. they bear the risk). The shareholders are
9 concerned if the CEO acts in their interests. Therefore the managerial behaviour of the CEO needs to be controlled by the shareholders.
2.1.2 Moral Hazard and Adverse Selection
As discussed above there is an information asymmetry between the principal and the agent within the agency theory. As Picard (1987) describes this information can take two different forms: Adverse Selection and Moral Hazard. He argues that Moral Hazard has to do with monitoring the agent’s actions and that Adverse Selection has to do with observing the agent’s private information. Similar, Guesnerie et al. (1989, p.807), argue that by designing a contract for incentivizing a CEO, the hidden knowledge (i.e. Adverse Selection) and hidden actions (i.e. Moral Hazard) need to be taken into account. With respect to Adverse Selection, the stakeholders must verify whether the abilities of the incumbent CEO matches the
requirements of the firm. With respect to Moral Hazard, the stakeholders must verify whether the CEOs take those actions that are in the best interest of the firm. A well designed contract will lead to less room for hidden knowledge and actions and thus reduces the agency problem. Eisenhardt (1988, p.58) stated that there are two possibilities why the agency relation can be distorted and cause an agency problem. The first explanation is a conflict of interest between the agent and the principal. He assumes that in a good relation (ideal situation) the interests of both parties are aligned and there is goal congruence. But in most situations there is a conflict and this causes differences in interest. The second possibility is that the principal is not completely able to monitor the agents’ activities. In a transparent and not complex
organization the principal will be perfectly able to evaluate the actions of the agent. But in most situations the organization is not transparent and even complex. This causes difficulty in evaluating the actions of the agent. For the first explanation of Eisenhardt (1988, i.e. conflict of interests) the organization will come up with contracts, to stir the agent in the right
direction. These contracts ensure that the agent is willing to work for the principal. According to Prendergast (2002) contracts are optimal when there is a good trade-off between the risks and the incentives. Organizations will have to make the trade-offs in the contract to reduce the agency problem because it is too costly to incentivize the agent completely. Contracts are made to incentivize the CEO, so he acts in the best interest of the organization (Bebchuck en Fried, 2006). For the second explanation of Eisenhardt (1988, i.e. unknown action) the organization has to monitor the agent, for example with performance measure. The
performance measures have to be set up by the organization in such a way that the agent can be evaluated and monitored. This allows the principal to easily assess if the agent has reached the targets. For both explanations there are some costs involved to overcome or reduce the
10 agency problem. These agency costs are defined by Jensen and Meckling (1976, p.308) as the sum of (1) the monitoring expenditures by the principal, (2) the bonding expenditures by the agent and (3) the residual loss. An organization cannot entirely stir the CEO in their desired direction, because it’s assumed that there will always be room for lacking behavior. Therefore the organization has to make a trade-off between risk and incentives when designing a
contract.
2.1.3 CEO performance and compensation
By designing a contract the organization has to take multiple considerations in to account. For example, CEOs need to be incentivized in such a way that they act in the interest of the organization (Bebchuck and Fried, 2006). It is assumed that the shareholders want to
maximize their shareholders’ value (Eisenhardt, 1988). The performance measures set by the organization must be clear and aligned to maximize shareholders’ value. Fama and Jensen (1983) argue that control of the agency problem can be achieved by the separation of decision control (monitoring) and decision management (implementation). Thereby the use of
incentive structures are important for the efficiency of the decision systems. A CEO can be incentivized due to compensation. The compensation package can be fixed or variable or a combination of both. Conyon (2006) gives an example of fixed compensation as salary, examples of variable compensation as annual bonus, stock options and additional
compensations. To get an annual bonus, the organization will set targets for the CEO which are in line with the goals of the organization. So the variable compensation is in this example dependent on the performance of the CEO. To evaluate the activities of a CEO the
organization will look for performance measures, like for example Return on Assets (ROA) or Stock Return. Banker and Datar (1989) argue that the agency costs can be mitigated by using efficient performance measures. An organization can set up efficient performance measure by using some criteria. According to Indjejikian (1999) performance measures are useful when the organization can see what the effort of the CEO is in that measure (Holstrom (1979) informativeness principle). Performance is influenced by more factors than the effort of the CEO only, for example a financial crisis can have influence on the performance. According to Gibbons and Murphy (1990), the organization will evaluate the performance on the hand of relative performance measures, which will be more aimed at shareholders’ value. Then the organization makes a decision about the future of the CEO. Furthermore organizations make use of performance measures to incentivize and evaluate the CEO (Bebchuck and Fried, 2006) and to overcome the two components of the agency problem: Moral Hazard and Adverse Selection.
11 Overall the organization has to take some steps to come to effective performance measures. First well designed performance measure provide information about the abilities of the agent (Bushman, 2010). If the abilities of the agent can be measured then the problem of Adverse Selection will be reduced. Second, performance measure provide information about the actions taking by the agent (Indjejikian, 1999). Moreover, if incentives are tied to
performance measures, this means that the agent provides incentives towards actions
measured by those performance measures. If this is the case then the unknown actions can be seen by the principal and the Moral Hazard will be reduced. Though, it is assumed that the CEO will always have more specific knowledge then the organization thus the Adverse Selection (hidden knowledge) phenomena will always be there. Thus the organization has to keep in mind that performance measure will never be perfect, as Jensen and Meckling (1976) stated that there is a residual loss.
2.2 CEO turnover
2.2.1 Managerial ability and scapegoat theory
Kim (1996) and Huson et al. (2004) make the assumption that firm performance is the sum of manager ability and the random component chance. By chance can be thought of industry-specific or firm-industry-specific shocks. The ability of the CEO and the chance play a role by the outcome of firm performance. Huson et al. (2004) come up with two hypotheses: the
managerial ability hypothesis and the scapegoat hypothesis. For the managerial ability theory, Huson et al. (2004) stated that managers differ in quality. When a firm has poor performance, this can be a sign of low ability, bad luck or both. They expect future performance to increase for two reasons: due to a more able manager or reversal of bad luck to normal. When this is applied to the prior described negative relation between performance and turnover, the
decision to dismiss a CEO arises from either bad luck, low ability or both as described above. Though it can be assumed that a CEO with a high ability will be more able to mitigate the bad luck factor. Then the quality factor will be the most important by the evaluation. According to Huson et al. (2004) the scapegoat theory is based on agency models. In this theory the quality of managers is equal. The random component chance plays a big role in this theory, poor performance is not caused by bad management but by bad luck alone. Furthermore, Huson et al. (2004, p.242) stated that the dismissal of managers is to induce other managers to provide the desired level of effort. Because managers have the same quality a dismissal itself does not increase managerial ability or firm performance. Though the management change is expected to result in a positive firm performance. The authors, Huson et al. (2004) state that a fired
12 manager can be seen as an example. This example is set to make sure the other managers, who dislike effort but are threatened by dismissal, will demonstrate the desired level of effort. Thus, basically there are two reasons to fire a CEO, scapegoat and managerial ability. With respect to the scapegoat theory, this can be linked with the Moral Hazard phenomena. With respect to the managerial ability theory, this can be linked with the Adverse Selection
phenomena. The most important of the two reasons for firing a CEO is the managerial ability which addresses the Adverse Selection. Because this aims on the differences in quality among CEO and this gives a ground for the possible dismissal of the CEO. The standard economic theory argues that factor beyond the CEO’s control should be ignored by the board of directors when assessing the quality of its CEO (Jenter and Kanaan, 2006). Though they stated it is difficult to fully separate performance variation due to CEO skill from performance variations due to luck and thus that the CEO dismissal is a mix of the effect of skill and the effect of luck.
2.2.2 Performance and CEO turnover
Above is described that organizations use performance measures to evaluate the actions of the CEO and the turnover decision. In short, performance measures play an important role in addressing the problems of Moral Hazard and Adverse Selection. That is, performance
measure outcomes provide information about the action choices made by executive managers. Also, incentives tied to performance measure outcomes leads to a situation where executive managers are upfront motivated to take those actions that are in the best interest of the firm as a whole. But performance measure outcomes also provide information about the abilities of the incumbent CEO. That is, poor performance could represent a signal that the abilities of the CEO do not match the requirements by the firm. The turnover could have to do with a lack of abilities of the CEO or to set an example for the organization, in order to turn poor
performance.
Warner et al. (1988) studied the association between a firm’s Stock Return and
subsequent top management changes. By looking at the top management changes, they looked at CEO, president and chairman of the board. Basically, they study the outcome of
performance measure (stock market performance) on the decision to replace or retain the CEO. Warner et al. (1988) show an inverse relation between the probability of a top
management change and Stock Performance. This implies the poorer the Stock Performance the more likely it is a top management change will occur. The paper provides new evidence that managers are more encouraged to act in the shareholders’ interests due to management changes when other manager perform poor. Thus the treat of being fired by performing poor
13 will increase the managers’ effort to reach the targets set by the organization. Warner et al. (1988) also find some evidence for several internal monitoring mechanisms. This implies that the turnover-performance relationship might be monitored by other factors.
Furthermore, Farrell and Whidbee (2003) examine the role of performance expectation by analysts’ forecasts on the turnover decision. They show that boards will focus on deviation from expected performance rather than performance alone in making a decision about CEO turnover. This means that the board will not only take the performance number in
consideration. The board takes into account analysts’ errors by evaluating their expectations about the CEO, because errors can capture a large part of the CEO performance. Farrell and Whidbee (2003) find an inverse relation when there is agreement among analysts between the likelihood of CEO turnover and analyst forecasts errors. This implies that when there’s a higher error in the forecast of the analysts the CEO will be less likely be fired. These findings suggest that the relationship between bad performance and CEO turnover can be increased by several other factors.
Jenter and Kanaan (2006) examined the factors beyond CEO’s control when CEOs were fired after bad performance. They expect and find limitedly that boards do not fully filter out exogenous shocks and that the CEO is dismissed when his quality fall below some
threshold (often the expected quality of the replacement). Jenter and Kanaan (2006) think boards by assessing CEO quality, systematically impute errors caused by factors beyond the CEO’s control and make use of rule-of-thumb relative performance evaluation. The error can have a positive or a negative consequence for the CEO. Furthermore the article points to other variables which can affect a forced CEO turnover like for instance CEO tenure and board composition. Another factor that might influence the CEO turnover is CEO reputation which will be discussed in the next section.
2.3 Other factors influencing CEO Turnover
In his paper, Weisbach (1988), studies the effect corporate governance on the relationship between firm performance and turnover. Weisbach (1988), shows that firms with more outside directors on the board will be more likely to fire a CEO by poor performance. He finds that CEO power can affect this relationship. This implies that CEO power can lead to less dismissal of more powerful CEO when the firm has poor performance. The power will become stronger when there are less outside directors in the board or during the tenure of the CEO.
Another factor influencing CEO turnover decision is accounting information. Engel et al. (2003), investigate to what extent properties of accounting information can have for
14 influence on the performance-turnover relationship. They find that when accounting-based measures are more precise and sensitive, these have a greater weight in the turnover-decision than market-based measures. Thus, when accounting information is more timely or when market returns are noisier, firms will set higher weights on accounting-based measures like sales, earnings per share (EPS) and return on assets (ROA) than on market-based measures like total shareholder return.
2.4 CEO Reputation
According to Henderson et al. (2006) CEO tenure is roughly defined as the life cycle of a CEO in which he learns during his time in the office. The life cycle is determined by the organization who provides the CEO a contract and evaluates his performance. So in the context of this research the CEO is held by the organization. The CEO is hired and paid (compensation) by the organization, and the organization has a kind of possession (under contract) of the CEO. When a CEO gets appointed by an organization he starts to work on his career in the organization. It is necessary for the CEO to build on his career because his talents are unknown (Bushman et al., 2010). It is likely that a CEO tries to show that he is competent enough to do his job for the organization. The organization will evaluate the CEO every year on his activities in the organization. Because this is a yearly evaluation you would expect that results of previous years aren’t taken in to consideration. In this research there will be examined whether building on reputation insulates CEO from dismissal following poor performance. In the next part will be looked at how reputation can influence performance and CEO turnover apart.
Several studies (Henderson et al. 2006; McClelland et al. 2012) investigate the role of tenure on the turnover. According to Henderson et al. (2006, p.458), the evaluation on the annually performance of the CEO will become less important. Therefore Shen and Cannella (2002, p.730) argue that a board of directors carefully needs to manage CEO tenure. Although Wang et al. (2010) argue that SOX weakens the board monitoring on CEO tenure. Also according to Jian and Lee (2011) the CEO power grows with the reputation of a CEO in an organization during his career. The papers are concluding that CEO tenure needs to be monitored because it can have an impact on the turnover decision.
On the other side, Davies and Chun (2009), stated that when a leader is the
spokesperson of the company his image reflects that company. So a good reputation leads to a good image of the company and its employees. The concept of reputation building seems to be contradicted because of the pros and cons. A better reputation might lead to better results but also might lead to more costs. Also Jin and Yeo (2011) note that a positive CEO
15 reputation and the success of a company are associated. This is consistent with Jiang and Lee (2011) who examine the association between CEO reputation and capital investments. They stated that the more reputable the CEO is the more he has to lose, in terms of credibility and future compensation when his firm invests in negative projects. Thereby Jiang and Lee (2011, p.930) stated that more reputable CEOs are more likely to achieve successful project
outcomes.
Regarding to tenure and performance, Allgood and Farrell (2000), also mention: “the longer a CEO works for a company, the more able he is to ‘circumvent monitoring and
incentive alignment techniques’ but is also able to make better decisions. The study of Kaplan
and Minton (2012) expect that the shorter the CEO tenures, the greater sensitivity to Stock Performance. So there are two assumptions about the role of CEO tenure regarding to performance. It can be argued that new CEOs are more motivated to do their job well but on the other hand incumbent CEOs have more experience. So it is questionable what the
influence on tenure is. Furthermore, Kaplan and Minton (2012), argue that the longer tenure, the more the CEO is focused on maintaining relations within the organization. But the authors also stated that the recent tenure is declining compared with earlier studies.
This area will provide the hypotheses of the research. In the previous sections the agency theory is described. To mitigate the agency problem there are agency costs involved (Jensen and Meckling, 1976). Especially the Adverse Selection phenomena are addressed. When the CEO starts working for the organization his abilities are unknown (Guesnerie, 1989). Because the organization doesn’t completely know what the activities of the CEO are, there’s a need to incentivize and monitor the CEO. When a CEO is appointed by the
organization he starts his career. Over series of time a CEO can build on his reputation in an organization. According to Jiang and Lee (2011) there’s a positive relationship between the reputation of the CEO and the corporate wealth of investments.
H1: Poor performing firms are less likely to replace a CEO with a good reputation.
This hypothesis will examine whether CEO with a good reputation (measured by his tenure) will be fired less quickly compared to other peers with a bad reputation. This study expects to find that CEOs with a good reputation will be fired less quickly. As Henderson et al. (2006) argues that the evaluation on the annually performance of the CEO will become less
important later in a CEO’s career. The board might be looking at time series of past
performance of the CEO. So it’s assumed that when a CEO has series of good performance the board can consider the current bad performance as an incident. Thereby other papers state
16 that the CEO power increase during his tenure and this needs to be monitored by the board (Shen and Cannella (2002) Jiang and Lee (2011)). With increasing CEO power it might be possible for a CEO to intimidate the board or to get a more preferable treatment of the board.
On the other hand Allgood and Farrell (2000) argue that a CEO with a long tenure is better able to circumvent the situation and better able to make decisions. So poor performance might imply that the CEO didn’t make the right decisions which could be a sign of lacking effort. Thus if a CEO with long tenure has poor performance it could be that the CEO is not motivated and this seems to be a good reason to fire the CEO despite or perhaps because of his good reputation.
17
3.1 Sample selection and Research methodology
The research will be examined by using an empirical archival study like other studies about the relationship between performance and CEO turnover (like Puffer and Weintrop (1991) and Weisbach (1988)). This is because performance is in most cases expressed in numbers, so it’s measurable quantitatively. Also the study of Allgood and Farrell (2000) who examine the effect of CEO tenure on firm performance and turnover, make use of quantitative data in their research. The quantitative data will provide enough information to test the hypothesis.
Therefore using quantitative data is a suitable way to execute this research. The data will be focused on US companies, because of the data availability and the extensive database of US Compustat companies. The data will be collected from the S&P 500 companies. The S&P 500 companies represent the overall US stock market. The dataset will be conducted by a 15-year time period and is ranged from 1998 till 2012.
Information about the CEO can be derived out of the Execucomp database. The Execucomp data is ranged from 1998 till 2013. The data about the appointment of a CEO and of if he left is in the Execucomp database. For the performance information about firm
financial numbers is used. This information is also derived from the Compustat database. With this database the fundamental annuals of the companies can be collected. There has been a financial crisis during 2007-2008, this need to be taken into account. There’s taken control for this by using industry adjusted return on assets, so companies are compared with their peers in the industry. For the CEO turnover this study makes use of the Plan based awards in the Execucomp database to see if the CEO name is changed in a certain year which implies a turnover. The two databases must be matched and merged to have a combined dataset for statistical analyses. The Execucomp data will be the starting point for the matched and merged dataset.
After importing the Execucomp and Compustat data, these data must be paneled and combined. The data must be paneled because it is observed cross sectional and over time series. The initial sample consists of 171,391 observations, after combining the data. Then the data is matched and merged, which leaves a matched merged data of 27,636 observations (143,755 observations deleted). Furthermore missing observations of financial information and of the characteristics of the CEO (like age and when he became CEO) will be eliminated. Besides that, Allgood and Farrell (2000) make a distinction between unregulated and
regulated firms, this study will also make that distinction. This is because unregulated firms and regulated firms can be different and can have different policies regarding to corporate governance (Smith and Watts, 1992; Gaver and Gaver, 1993). Financial institutions and
18 public utilities will be scrapped out of the original sample. All these eliminations lead to a sample of 21,236 observations. This is reflected in the table below.
Table 1. Sample Selection
Observations
Matched merged data 27,636
Less: observations with missing financial data on ROA or Return 621
Less: observations with missing data on CEO Tenure (becameCEO) 397
Less: observations with missing data on the age of the CEO 797
Less: financial institutions (SIC codes 6000-6799) 4465
Less: public utilities (SIC codes 9100-9999) 120
Final sample 21,236
3.1.1 Identifying Reputation
The reputation variable will be measured through tenure. This is done because it is assumed that the CEO builds up his reputation over the years. This implies that a longer tenure probably leads to a higher reputation. This assumption is made on the argument of Bushman et al (2010), that it is necessary for the CEO to build on his career because his talents are unknown to the organization. The CEO reputation variable will be calculated by the tenure of a CEO, this means the difference in the moment when he becomes a CEO and the moment when he leaves the company. This will be a static variable and if there is no information available on when the CEO is left the date of leaving will be set on this year (2015). The study doesn’t treat the reputation variable as a continuous variable because Allgood and Farrell (2000) don’t expect that there is a continuous function of interaction of CEO
reputation and firm performance. Reputation will be coded as high reputation when tenure is higher than the median tenure in the sample and tenure that is equal or lower than the median tenure in the sample will be coded as low reputation.
3.1.2 Identifying Turnover
A turnover will be noted when the CEO’s full name changes among a certain year. The turnover variable will be a dummy variable. When a turnover takes place this will have the value of 1, and zero otherwise (no turnover). After analysing the sample with regard to the turnover I come to the following results. The sample consists of 2,385 unique firms
19 experiencing 2,185 CEO turnovers and 21,236 CEO years. This implies a 10.29 percent turnover rate per year. This is comparable to the 9.0 percent of Allgood and Farrell (2000). The numbers are displayed in the table 3.1 CEO Turnover below.
Furthermore the turnovers will be classified. The classification will be between forced and voluntary turnover. Likewise Allgood and Farrell (2000) I exclude turnovers which reports no reason for the turnover. According to Kaplan and Minton (2012) there are two types of turnover: internal and external turnover. This implies that the total turnover is the sum of the internal and external turnover. In their article, Kaplan and Minton (2012, p.58), stated that internal turnover is board initiated and external turnover is related to mergers and acquisitions. The paper also gives that the external turnover is in most regressions unrelated with Stock Performance, while internal turnover is related with Stock Performance (2012, p.58). This research won’t take into account if a turnover is external or internal, it will only look at if the turnover is forced or voluntary.
Vancil (1987) stated that many CEO changes are due to the normal succession process. Also, according to Warner et al. (1988), it is possible to identify reason for the management change in many cases. They identify the reason for management change by first looking at the Poor’s Register of Corporations, Directors and Executives. Secondly, they look at the Wall Street Journal (WSJ) Index. Though they stated that it is difficult to exactly identify forced turnovers because press releases are rarely describe them. Puffer and Weintrop (1991) give reasons for voluntary turnover like age and increasing retirement benefits.
Although the reason why a CEO turnover has taken place is hard to measure. This is because the organization doesn’t have to give a report on why the turnover has taken place. According to Allgood and Farrell (2000) a CEO turnover could be forced or voluntary. The authors stated that a forced turnover is the consequence of poor performance (or dead) and a voluntary turnover has to do with: retirement, sickness and career movement.
The Execucomp database gives five possible reasons for turnovers. These reasons are: deceased, resigned, retired, unknown and no reason. The last two will be excluded by classifying forced and voluntary turnover. A table about the reason for turnover is in the table 3.2. In this research the following classifications will be made. For the voluntary
classification, this includes CEO changes as a result of retirement, normal succession, death or illness and planned succession. Death or illness is from the CEO point of view a forced way to leave the company, though the company didn’t force the CEO to leave. The turnover can be seen as one who was necessary due to the circumstances but not initiated by the board of directors. Forced turnovers will be classified for CEOs who were resigned. Furthermore for
20 the retirements of CEO there will be looked at the age of the CEO. Likewise Warner et al. (1998) the age will be examined by its median and mean. The median age of retired turnovers is 62 years, the mean of retired turnover is 61.2 years. For all the turnovers other than
retirement the median age is 57 years and the mean is 57.3 years. If only the forced turnovers (reason is deceased or resigned) are examined then the median age is 54 years and the mean is 54.5 years. Just as the results of Warner et al. (1998) retirements are generally older. Likewise Parrino (1997) all reported retirements will be forced departments when the CEO is under a certain age. Parrino (1997) uses the age 60 to determine a forced turnover, this research will use the median age of all turnovers (i.e. 57 years). The classification of forced and voluntary turnover is displayed in table 3.3.
Table 3.1 CEO turnover
CEO Turnover Frequency Percentage
No turnover | 19,051 89.71
Turnover | 2,185 10.29
Total | 21,236 100.00
Table 3.2. Reasons for turnover
Reason Frequency Percentage
Deceased | 27 1.24 Resigned | 346 15.84 Retired | 489 22.38 Unknown | 121 5.54 No Reason | 1,202 55.01 Total | 2,185 100.00
Table 3.3: classification of turnover
Turnover classification Frequency Percentage
Forced | 457 53.02
Voluntary | 405 46.98
Total | 862 100.00
(Excluded from sample due to classification difficulties 1,202+121=1,323 turnovers, this represents 65.1 percent of the total turnover sample of 2,185)
These figures of the tables come from the 21,236 observations. The CEO reputation has two possibilities, either a CEO has a low reputation or a high reputation. This is determined by using the median tenure of CEOs. CEOs with high reputation will be coded as 1, and zero
21 otherwise (low reputation). The test of the interaction between CEO reputation and each performance measure using tenure gives the following results. For the turnover sample there are 405 voluntary turnovers and 457 forced turnovers. From the voluntary (forced) turnovers there are 320 (396) CEOs with a low reputation and 85 (61) CEOs with a high reputation.
Though the dataset that will be used for the logistic regression models consists of 4,128 observations after deleting duplicates and missing values. In this dataset there are 2,208 turnovers and 1,920 no turnovers. So there are some differences in the turnovers, this is due to the duplicates and missing values.
3.1.3 Identifying Performance
The performance will be measured in two ways, by accounting performance and Stock Performance. This measures for performance will also be set relatively to the industry. For accounting performance this will be executed by using industry-adjusted ROA, calculated as the total assets over the earnings before interest and taxes, this will be subtracted by the mean of the measure of the same two-digit Standard Industrial Classification (SIC) code. This is similar to the study of Parrino (1997). For the Stock Performance stock price will be used. Likewise Puffer and Weintrop (1991) who use change in stock price as a measure of corporate performance. The calculation of Stock Performance will be the end of the year stock price minus the begin of the year stock price divided by the begin of the stock price. To adjust it to the industry the mean of the measure of the same two-digit SIC code will be subtracted. The industry classification by two-digit SIC code is in the appendix.
3.2 Regression model and proxies
Prior literature doesn’t show a regression model. But a regression model can be derived from the research question. The research question is how CEO reputation affects the relationship between firm performance and CEO turnover. So turnover is influenced by the performance (prior literature gives a negative relationship). In this research the CEO turnover is the dependent variable. This variable will be dependent on a constant variable (b0). Furthermore as described above this research will use two measures for performance: accounting
performance (b1) and market performance (b3). Thereby, in this research the interaction between CEO reputation and the performance is an important factor (e.g. interaction is b2 for accounting performance and b4 for market performance). The remaining variable which is important for the research question is CEO reputation (b5). There will also be some control variables in the model, like controlling for size (b6) which will be the natural log of sales and
22 age which will be the age of the CEO (b7). Thereby there will be an error in the model. The regression model for this research is written out and showed below:
CEO_turnover = b0 + b1AccPerf + b2AccPerf*CEO_Reputation + b3MarketPerf + b4MarketPerf*CEO_Reputation + b5CEO_Reputation + b6CEO_Age + b7Size + error
Whereby:
CEO_turnover = Dummy variable; 1 for turnover, 0 otherwise (no turnover) AccPerf = Accounting performance; measured by ROA
CEO_Reputation = Higher than median tenure is coded as high reputation, Equal or lower than median tenure is coded as low reputation Dummy variable; 1 for high reputation, 0 otherwise (low reputation)
MarketPerf = Market performance; measured by stock price at end of fiscal year minus stock price at beginning of fiscal year, divided by stock price at beginning of fiscal year
Control variables:
(CEO age) = Age of the CEO
(Size) = Natural log Sales
With regard to the accounting performance, b1 gives the relationship between accounting performance and the likelihood of CEO dismissal for CEOs with low reputation (i.e,, CEO_reputation=0), while the sum of coefficients (b1+b2) gives the relationship between accounting performance and the likelihood of CEO dismissal for CEOs with high reputation (i.e,, CEO_reputation=1). So, b2 gives the difference between CEOs with low versus high reputation with respect to the relationship between performance and dismissal. Prior literature showed a negative relationship between CEO turnover (dependent variable) and accounting performance (b1). So I expect that b1 is less than 0, which will give a negative relationship between turnover and accounting performance and is in compliance with prior literature. Regarding to the hypothesis, I expect that b2 will be greater than 0. This implies that
23 turnover. In other words, CEOs with high tenure (b1+b2) are less likely to be fired than CEOs with low tenure by poor performance (measured by accounting performance).
With regard to the market performance, b3 gives the relationship between Stock Performance and the likelihood of CEO dismissal for CEOs with low reputation (i.e,, CEO_reputation=0), while the sum of coefficients (b3+b4) gives the relationship between Stock Performance and the likelihood of CEO dismissal for CEOs with high reputation (i.e,, CEO_reputation=1). So, b4 gives the difference between CEOs with a low versus high reputation with respect to the relationship between performance and dismissal. Regarding to the hypothesis, I expect that b4 will be greater than 0. This implies that reputation will weaken the negative relationship between Stock Performance and CEO turnover. In other words, CEOs with high tenure are less likely to be fired than CEOs with low tenure by poor performance (measured by Stock Performance). Further explanation about the performance measures is in the appendix.
3.2.1 Expectations regarding hypotheses
For the hypothesis where CEO reputation is the main subject, it’s expected to find a moderating effect. Because it is necessary for the CEO to build on his career because his talents are unknown (Bushman, 2010). This study expects that CEO who have a higher tenure have built on their career and are thus familiar with the company. This makes CEOs with high reputation better in decision making. Also according to Allgood and Farrell (2000) the longer a CEO works for a company, the more able he is to make better decisions. Thus the
expectation is that the reputation will weaken the prior literature showed relationship between turnover and performance.
24 4. Results
4.1 Descriptive Statistics
Based on the sample of 21,236 observations the mean (median) age of the CEO is 55.3 (55) years. The average tenure of the CEO is 12.6 years with a median of 11 years. The youngest CEO in the sample is 28 years and the oldest CEO in the sample is 96 years. When there’s only looked to the turnover data the youngest CEO in the sample is 29 years and the oldest is 90 years. Overall, there’s a big spread in the age of CEOs in the sample. Furthermore, table 2 shows that the mean and the median tenure of the CEO are 12.6 years and 11 years,
respectively. Insiders account for 30.86 percent of the sample and outsiders account for 69.14 percent of the total sample.
Table 4.1 describes the CEO and other characteristics of the sample (n=4,128) used for the logistic models. This table shows an average CEO age of 56.9 years with a median of 57 years. The mean CEO tenure of 10.3 and a median of 8 years. Thus for the reputation variable a tenure of 8 years or lower means low reputation and a tenure higher than 8 years means high reputation. The control variable size is calculated by sales. The mean of the firm sales is about $5.3 billion and the median is about $1.3 billion. The mean (median) ROA is 6.9 percent (8.0 percent). For the Stock Return the mean is 20.2 percent and the median 2.7 percent. The performance variables (ROA and Stock Return) have high standard deviations. This means that the performance observations are variable and volatile. Especially the standard deviation of Stock Return is very high. This is caused by some very high value changes in the sample. For example in the firm Warnaco Group Inc. the Stock Return is very volatile. In 2002 when there’s a turnover, the stock price falls from $0.055 at the beginning of the year to $0.003 at the end of the year. This implies a decrease of 94.54 percent in the stock price. Even more volatile is the stock price change in the following year when there is again a turnover. In 2003 the stock price rises from $0.003 to $15.95, this is an increase of factor 5315.67. These figures create high volatility and thus a high standard deviation.
25 Table 4.1. Descriptive Statistics of CEOs
Sample Characteristics
CEO Characteristics Mean Median Std. Dev. Freq. Percentage
CEO age 56.9 57 7.5 (in years) CEO tenure 10.3 8 8.0 (in years) Insiders 6,553 30.86 Outsiders 14,683 69.14 Sales 6139 1448 19173 (in millions) ROA 6.9% 8.0%* 21.3% Stock Return 20.2% 2.7% 383.7%
*median is bigger due to more small values which are farther from the mean than the high values.
Table 4.2 below shows the descriptive statistics of the logistic regression model. Per variable the mean, standard deviation, the 25th and 75th percentile and the median is given. Table 4.2 shows a CEO turnover mean of about 46.51 percent with a median of 0. This means that about 46.51 of the CEOs in the sample don’t experience a turnover during their tenure. For the ROA and Stock Performance the means (medians) are 0.0691 (0.0796) and 0.2025 (0.0269).
Furthermore, the results indicate that 48.86 percent of the CEO have a low reputation (median = 0) during their tenure at the firm. About 62.03 percent of the CEOs are older than or equal to 57 years. At last the size variable has a mean of 7.3246 and a median of 7.2777.
Table 4.2. Descriptive Statistics of variables of the logistic regression model Sample Characteristics
Variable N Mean Std. Dev. 25% Median 75%
CEO Turn 4128 .4651 .4988 0 0 1 ROA 4128 .0691 .2134 .0388 .0796* .1307 Stock Perf 4128 .2025 3.8370 -.2277 .0269 .2662 CEO Repu 4128 .4886 .4999 0 0 1 CEO Age 4128 .6203 .4854 0 1 1 Size 4128 7.3246 1.6842 6.2276 7.2777 8.4142
*median is bigger due to more small values which are farther from the mean than the high values.
In table 4.3a and 4.3b the differences between turnover and no turnover is displayed. There are 1920 observations of no CEO turnovers and 2208 observations of CEO turnover.
Comparing the ROA of no turnover versus turnover, the mean ROA for no turnover (.0702) is little higher than for turnover (.0681). This is in accordance to the expectation that poor performing firms experience more turnovers. Also the mean of the Stock Performance is lower for turnover (.1967) compared to no turnover (.2091), but the median Stock
26 Performance of no turnover is lower than the median Stock Performance of turnover. Though as stated before there is a high volatility (standard deviation) in Stock Performance. Thus it might be that there are other factors involved by determining turnovers with Stock
Performance. Furthermore the turnover data of the CEO reputation variable shows that about 58.24 percent of the CEO who have no reputation are dismissed. But this is only 38.07 percent in the no turnover data. So this might suggest that CEOs with no reputation a have a higher probability of getting dismissed. Regarding to the control variable CEO age, it shows in the no turnover sample 62.03 percent of the CEOs are younger than 57 years and 44.70 percent in the turnover sample are younger than 57 years. This might suggest that older CEOs are more quickly face a turnover. This can be because for example the CEOs are too old and retire or older CEO have more trouble with the performance expectations. The table shows for size that in smaller companies (7.2407 versus 7.4211) there are more CEO turnovers. Table 4.3a. Descriptive Statistics no CEO turnover
Sample Characteristics
Variable N Mean Std. Dev. 25% Median 75%
ROA 1920 .0702 .1634 .0348 .0779 .1332
Stock Perf 1920 .2091 5.1333 -.3146 -.0459 .2183
CEO Repu 1920 .3807 .4857 0 0 1
CEO Age 1920 .6203 .4854 0 1 1
Size 1920 7.4211 1.6995 6.3125 7.3735 8.5608
Table 4.3b. Descriptive Statistics CEO turnover
Sample Characteristics
Variable N Mean Std. Dev. 25% Median 75%
ROA 2208 .0681 .2490 .0409 .0820 .1284 Stock Perf 2208 .1967 2.1492 -.1401 .0778 .2973 CEO Repu 2208 .5824 .4932 0 0 1 CEO Age 2208 .4470 .4973 0 1 1 Size 2208 7.2407 1.6667 6.1646 7.2033 8.3170 4.2 Regression results
To test whether CEO reputation affects the relationship between turnover and performance a logistic regression is executed. Using ROA and Stock Performance as measures for
performance to test the effect of tenure to examine if CEOs with little reputation are dismissed sooner when there is poor performance. Table 4.4 shows the coefficients, standard errors and significance levels of the variables. In table 4.5 the variables b2 and b4 are incorporated.
27 In table 4.4 the ROA coefficient is negative (-.0248) but insignificant (p>0.05). The negative coefficient is in accordance with this study’s expectations and prior literature. Though there is no support for the prior literature negative relationship between accounting performance and turnover, due to the insignificance. Also the Stock Performance has an insignificant positive coefficient, which doesn’t support the Stock Performance turnover relationship. The coefficient of CEO reputation is negative and significant. This means that there’s a negative relationship between CEO reputation and CEO turnover. Thus the less the CEO reputation the more likely it is that a CEO turnover occurs. CEO age shows a positive significant coefficient, meaning that it seems it’s more likely for an old CEO to get fired. For Size there is an insignificant positive coefficient.
Table 4.5 shows a ROA coefficient which is negative (-.4038) but also insignificant (p>0.05). Likewise the previous outcome of table 4.4, the negative coefficient is in
accordance with this study’s expectations and prior literature. Though due to the insignificance there is no support for the negative relationship between accounting performance and turnover. The b2 coefficient is positive (.7528) as expected, though not significant. Regarding to the Stock Performance it has an insignificant positive coefficient, which doesn’t support the Stock Performance turnover relationship. Coefficient b4 shows a negative significance, meaning that when a CEO has high reputation this weakens the negative market performance turnover relationship. The coefficient of CEO reputation is again negative and significant. Meaning the likelihood of a CEO turnover increases when there is little CEO reputation. Likewise table 4.4 CEO age shows a positive significant coefficient and Size is an insignificant positive coefficient.
Table 4.4. Logistic Regression using ROA and Stock Performance Sample Characteristics
CeoTurnover Coefficient Std. Err. z P>│z│ [95% Conf. Interval] ROA -.0247737 .159836 -0.15 0.877 -.3380465 .22884992 Stock Perf .0012465 .0082904 0.15 0.880 -.0150025 .0174954 CEO Repu -.9454506 .066773 -14.16 0.000 -1.076323 -.8145779 CEO Age .8431201 .06737 12.51 0.000 .7110774 .9751628 Size .0123929 .0205416 0.60 0.546 -.0278679 .0526537 Constant -.2246534 .1549751 -1.45 0.147 -.528399 .0790923
28 Table 4.5. Logistic Regression using ROA and Stock Performance with b2 and b4
Sample Characteristics
CeoTurnover Coefficient Std. Err. z P>│z│ [95% Conf. Interval] ROA -.4037729 .2936856 -1.37 0.169 -.9793862 .1718404 b2 .7528148 .4116824 1.83 0.067 -.0540679 1.559698 Stock Perf .0103332 .013641 0.76 0.449 -.0164028 .0370691 b4 -.2950302 .090727 -3.25 0.001 -.4728519 -.1172085 CEO Repu -.9718178 .073472 -13.23 0.000 -1.11582 -.8278153 CEO Age .8470588 .0675944 12.53 0.000 .7145732 .9795385 Size .011362 .020871 0.54 0.586 -.0295444 .0522684 Constant -.1953346 .1565291 -1.25 0.212 -.5021259 .1114568 4.3 Sensitivity analysis
Table 4.6 shows the industry-adjusted accounting and Stock Performance. Hereby the
variables are adjusted to the two-digit Standard Industry Classification code. It’s interesting to see that the adjusted ROA is a positive significant coefficient. This is contrary to the prior literature which provided a negative relationship between performance and turnover. For Stock Performance there’s also a positive coefficient but no significance. The b2 is quite the same as the previous, positive insignificant. The b4 coefficient is again negative significant. CEO reputation and CEO age show a significant coefficient like the previous test. Size turns into a negative but an insignificant coefficient.
Table 4.6. Logistic Regression using industry-adjusted ROA and Stock Performance Sample Characteristics
CeoTurnover Coefficient Std. Err. z P>│z│ [95% Conf. Interval] Adj ROA .6785463 .2970733 2.28 0.022 -.0962932 1.260799 Adj b2 .7005947 .5174109 1.35 0.176 -.313512 1.714702 Adj Stock Perf .0134878 .0138014 0.98 0.328 -.0135624 .0405379 Adj b4 -.1627456 .0759456 -2.14 0.032 -.3115962 -.0138951 CEO Repu -.9711693 .0673574 -14.42 0.000 -1.103187 -.8391512 CEO Age .8414959 .0676939 12.43 0.000 .7088183 .9741735 Size -.0119687 .0205509 -0.58 0.560 -.0522477 .0283103 Constant -.0417626 .1580758 -0.26 0.792 -.3515854 .2680603
29 5. Conclusion
This paper examines the effect of CEO reputation on the relationship between firm performance and CEO turnover. Prior literature shows a well documented negative relationship between firm performance and CEO turnover. Thereby other prior literature investigates possible factors that can influence this relationship. But only little research looks at CEO reputation as a moderator for the relationship between performance and turnover. This study contributes by examining how CEO reputation affects the relationship between
performance and turnover. This paper expected to find a negative effect of CEO reputation on CEO turnover. This implies that the expectation is that the higher the reputation the lower the turnover.
The hypothesis of this paper stated as: poor performing firms are less likely to replace a CEO with a good reputation. This is investigated with an initial sample of 21,236
observations and a logistic regression model sample of 4,128 observations of US companies between 1998 till 2012. The performance measures were addressed by ROA as accounting performance and Stock Return as market performance. The CEO reputation was treated as a dummy with 1 for high reputation (> median tenure) and 0 for low tenure (≤ median tenure). The results showed only a positive significance for the industry-adjusted ROA, which
contradicts to prior literature negative between firm performance and CEO turnover. Only the coefficient of the interaction between Stock Performance and CEO reputation shows a
negative significant relation to CEO turnover. This suggests that high CEO reputation
weakens the relationship between market performance and CEO turnover. The paper found a negative moderating effect of reputation on market performance and CEO turnover.
This paper has some limitations. One limitation is the measurement of the CEO reputation variable. This variable is constructed on a simple and static manner by using the median of CEO tenure. Another limitation is that the CEO turnover could be investigated more deeply regarding to the reason for the turnover. Also the performance measure is only measured by two variables: ROA and Stock Return. Future research might measure CEO reputation as a continuous variable. Furthermore, regarding to the classification of the CEO turnovers interviews can be conducted. Also other variables for performance could be used or more control variables to examine the CEO reputation effect.
30 6. References
Abe, Y (1997). Chief executive turnover and firm performance in Japan. Journal of the
Japanese and International Economie , 11 (1), 2-26
Allgood, S. Farrell, K.A. (2000). The effect of CEO tenure on the relation between firm performance and turnover. Journal of Financial Research , 23 (3), 373-390
Banker, R.D., Datar, S.M., (1989). Sensitivity, precision, and linear aggregation of signals for performance evaluation. Journal of Accounting Research , 27 (1), 21-39.
Bushman, R. Dai, Z. Wang, X. (2010). Risk and CEO turnover. Journal of Financial
Economics , 96 (3), 381-398.
Bebchuk, L. en Fried J. (2006). Pay without performance: overview of the issues. Academy of
Management Perspectives , 20 (1), 5-24
Davies, G. Chun, R. (2009). The leader’s role in managing reputation. Reputation capital , 311-323
Conyon, M.J. (2006). Executive Compensation and Incentives. Academy of Management
Perspectives , 20 (1), 25-44
Engel, E. Hayes, R. M., and Wang, Xue. (2003). CEO turnover and properties of accounting information. Journal of Accounting and Economics , (36), 197–226.
Fama, E.F. and Jensen M.C. (1983). Separation of Ownership and Control. Journal of Law
and Economics , 26 (2), 301-325
Gaver, J.J. Gaver, K.M. (1993). Additional evidence on the association between the investment opportunity set and corporate financing, dividend, and compensation policies. Journal of Financial Economics , 16 (1-3), 125-160
Guesnerie, R., Picard, P. and Rey, P. (1989). Adverse Selection and Moral Hazard with risk neutral agents. European Economic Review , 33 (4), 807-823
Hartono, J. (2011) the relationship between accounting performance and CEO turnover: Evidence from Indonesia. Accounting & Taxation , 3 (2), 97-107
Henderson, A.D., Miller, D. and Hambrick, D.C. (2006). How quickly do CEO’s become obsolete? Industry dynamism, CEO tenure, and company performance. Strategic
Management Journal , 27 (5), 447-460
Hoitash, R., Hoitash, U. and Johnstone, K.M. (2012). Internal Control Material Weaknesses and CFO compensation. Contemporary Accounting Research , 29 (3), 768-803
Holmstrom, B. (1979). Moral Hazard and observability. The Bell Journal of Economics , 10 (1), 74-91
Huson, M.R. Malatesta, P.H. Parrino, R. (2004). Managerial succesion and firm performance.
Journal of Financial Economics , 74 (2), 237-275
Indjejikian, R. J. (1999). Performance Evaluation and Compensation Research: An Agency Perspective. Accounting Horizons , 13 (2), 147-157
Jensen, M.C., and W.H. Meckling (1976). Theory of the firm: managerial behavior, agency costs and ownership structure. Journal of Financial Economics , 3 (4), 305-360. Jian, M. Lee, K.W. (2011). Does CEO reputation matter for capital investments? Journal of
Corporate Finance , 17 (4), 929–946.
Jin, C.H. and Yeo, H.C. (2011). Satisfaction, corporate credibility, CEO reputation and leadership effects on public relationships. Journal of Targeting, Measurement &
Analysis for Marketing , 19 (2), 127-140.
Kaplan, S.N. and Minton, B.A. (2012). How Has CEO Turnover Changed? International
Review of Finance , 12 (1), 57–87
Kato, Takao and Long, Cheryl. (2006). Executive Turnover and Firm Performance in China.
31 Kesner, I. F. and Dalton, D. R. (1994). Top management turnover and CEO succession: An
investigation of the effects of turnover on performance. Journal of Management Studies
, 31 (5), 701-713
McClelland, P.L., Barker, V.L. and Oh, W. (2012). CEO career horizon and tenure: Future performance implications under different contingencies. Journal of Business Research , 65 (9), 1387-1393
Nassirzadeh, F. Saei, M.J. Salehi, M. Varnosfaderani, F.R. (2013). CEO's Reputation : Corporate Performance. SCMS Journal of Indian Management , 10 (3), 5-12.
Parrino R. (1997). CEO turnover and outside succession: A cross-sectional analysis. Journal
of Financial Economics , 46 (2), 165-197
Picard, P. (1987). On the design of incentive schemes under Moral Hazard and Adverse Selection. Journal of Public Economics , 33 (3), 305-331.
Puffer, S.M. and Weintrop, J.B. (1991). Corporate performance and CEO turnover: The role of performance expectations. Administrative Science Quarterly , 36 (1), 1-19.
Samuelson, B.A., Galbraith, C.S. and McGuire, J.W. (1985). Organizational performance and top-management turnover. Organization Studies , 6 (3), 275-291.
Shen, W. and Cannella Jr., A.A. (2002). Revisiting the performance consequences of CEO succession: The impacts of successor types, postsuccession senior executive turnover, and departing CEO tenure. Academy of Management Journal , 45 (4), 717-733
Smith Jr., C.W. and Watts, R. L. (1992). The investment opportunity set and corporate financing, dividend, and compensation policies. Journal of Financial Economics , 32 263-292.
Wang, H., Davidson, W.N. and Wang, X. (2010). The Sarbanes-Oxley Act and CEO tenure, turnover, and risk aversion. The Quarterly Review of Economics and Finance , 50 (3), 367-376
Warner, J.B. Watts, L.R., Wruck, K.H. (1988). Stock prices and top management changes.
Journal of Financial Economics , 20, 461-492.
Weisbach, M.S. (1988). Outside directors and CEO turnover. Journal of Financial Economics
32 Appendix
Appendix 1: two-digit SIC codes per industry Code Industry Title
01-09 Agriculture, Forestry & Fishing 10-14 Mining
15-17 Construction 20-39 Manufacturing
40-49 Transportation, Communications, Electric, Gas & Sanitary Services 50-51 Wholesale Trade
52-59 Retail Trade
60-67 Finance, Insurance & Real Estate 70-89 Services
91-99 Public Administration
Appendix 2: logistic regression model Explanations for accounting performance
Looking to the regression model for accounting performance, the following model is applicable:
CEO_turnover = b0 + b1AccPerf + b2AccPerf*CEO_Reputation + controls
Hereby CEO_turnover which is the dependent variable, b0 which is a constant and controls which are multiple variables are not explained further. In this explanation is only looked at the b1 and b2 part. As prior literature shows a negative relationship between CEO turnover (the dependent) and accounting performance (b1AccPerf) it’s expected that the b1 is a negative number. This implies that when there is less performance there is more CEO turnover (b1 gets more negative). Now for the b2AccPerf*CEO_Reputation part consist of a dummy variable. In the study is stated that when CEOs have a high reputation this will be noted as 1 and 0 otherwise. So when a CEO has a low reputation the b2AccPerf*CEO_Reputation will expire. Meaning that when CEOs have a low reputation the CEO_turnover will be dependent on only b1. When CEOs have a high reputation (CEO_Reputation = 1), then also b2 will be taken into account. Because I expect that reputation will weaken the negative relationship between
