The effect of CEO tenure on the relationship
between CEO turnover and performance
Joris Tanke
5947448
Master thesis: Accountancy and Control 23th June 2014
Thesis supervisor: P. Kroos Second examiner:
2
Abstract
This paper empirically investigates the effect of a CEO’s tenure on the negative relationship between a company’s performance and CEO turnover. It might be interesting to see if there is a difference between bad performing CEO’s with a long tenure compared to those with a short tenure. This, because a CEO’s ability is unknown when a company hires one and CEO’s build their reputation over time. This might insulate them from experiencing a turnover after bad performance. So, the hypothesis examined in this paper is: CEO tenure negatively moderates the negative relationship between firm performance and CEO turnover. The results support the hypothesis for the accounting measures (ROA and EPS) of performance but not for the market measure (Stockperformance) of performance.
3
Table of content
1. Introduction...4
1.1 Background...4
1.2 Motivation and contribution...5
1.3 Research question...5
1.4 Structure...6
2. Literature review and hypotheses...7
2.1 Agency theory, moral hazard and adverse selection...7
2.2 Relationship between performance and CEO turnover...8
2.2.1 Performance and CEO turnover...8
2.2.2 Measures used as indicator of performance in the relationship between...9
performance and CEO turnover... 2.3 Moderators in the relationship between performance and CEO turnover...11
2.3.1 Current moderators in the relationship between performance and CEO...11
turnover... 2.3.2 CEO tenure as moderator in the relationship between performance and CEO...12
turnover... 3. Research methodology...14
3.1 Sample...14
3.2 Variable measurement...15
3.2.1 Measures of firm performance...15
3.2.2 CEO turnover...15 3.2.3 CEO tenure...15 3.2.4 Control variables...16 3.3 Empirical model...16 3.4 Descriptive statistics...18 4. Empirical results...19
5. Summary and conclusion...24
4
1. Introduction 1.1 Background
The agency theory describes the relation between the principal and the agent. Where the principal is the one who owns the company and the agent the one who manage the company on behalf of the principal (Ross, 1973). This theory builds on the assumption of information asymmetry where the principal does not know which actions the agent has selected (moral hazard), and also does not know the abilities of the agent (adverse selection). Therefore performance measures are used to learn about the actions taken by the agent and to encourage managers to take certain actions. In addition, performance measures are used to learn about the ability of the manager. This is to address adverse selection problems. This thesis will take a closer look at the way performance measures are used to learn about the ability of the agent.
An extensive literature documents an inverse relation between the likelihood of chief executive officer (CEO) turnover and firm performance (Weisbach 1988; Murphy and Zimmerman 1993). This literature is predicated on the idea that firm performance reveals information about a CEO’s ability to create value for shareholders. When firm performance is poor, a CEO is replaced because the firm’s owners infer that he is ineffective at formulating and implementing strategies and policies that enhance firm value. Since owners’ beliefs about their CEO’s ability are revised over time based on periodically observing firm performance, their beliefs of CEO ability become increasingly precise over the employment relationship (Hermalin and Weisbach, 1998).
So, shareholders and other external parties are not completely sure about the abilities of the manager. From the outcome on performance measures, they infer whether the CEO is a good match with the firm. Over time, the firm’s beliefs about the CEO’s are updated, and the firm has a more precise estimate on the abilities of the CEO. This, however, also implies that the firm has little knowledge about the abilities of a CEO that just taken office. Indeed, the ability of the CEO is unknown when they hire and they update their knowledge about his or her ability on the basis of the achievements along the way (Bushman et al., 2010). Other research states that CEO reputation is build over time with achievements (Jian and Lee, 2011). This raises the question to what extent prior performance achievements (i.e., establishing a reputation) insulates a manager
5
to some extent from the threat of dismissal of substandard performance. Formulated differently, are managers that are new to the firm and did not build a track record yet more susceptible of being ousted when performance does not meet the expectations. This is the main focus of the thesis.
1.2 Research Question
On the basis of the aforementioned, in this study I look at the relationship: How does tenure affect the relationship between performance and CEO turnover?
As mentioned in the previous paragraph, a CEO builds his reputation over time and his or her ability is unknown when they are hired. (Bushman et al., 2010; Jian and Lee, 2011). Therefore I expect that tenure negatively moderates the negative relationship between performance and the likelihood of dismissal. So, the main effect is that if performance decreases, the likelihood of dismissal increases. I expect this relationship to be weaker for CEO’s with greater tenure. That is, a CEO will less likely to be replaced if he/she performs bad when the respective CEO has build a reputation in the past relative to a manager who is new and has no reputation to rely on.
1.3 Motivation
This paper will contribute to the emerging body of literature regarding the relationship between CEO turnover and performance. Prior literature has focused on documenting the negative relationship between performance and the likelihood of dismissal (Warner et al., 1988; Coughan and Schmidt, 1985).
Prior research has looked into different performance benchmarks. For example, Warner et al. (1988) and Coughan and Schmidt (1985) use stock market performance as a benchmark for performance. If the firm exhibits a weak stock market performance this indicates poor performance and should be associated with a higher probability of a management change. Other research use analyst forecast errors (the deviation of realized earnings from expected earnings) as a performance benchmark. The extent that earnings forecasts proxy for the board of director’s earnings expectations, forecast errors may be capture the component of firm
6
performance that the board attributes, in large part, to CEO performance (Farrell and Whidbee, 2003).
Yet another proxy used as a performance benchmark is Return on Assets, calculated as the ratio of accounting earnings before interest and taxes to total assets. This measure have been commonly used in prior literature that identifies a negative relation between firm performance and the likelihood of CEO turnover (Allgood and Farrel, 2000; Dikolli and Mayew, 2013).
More recent literature has focused on firm-specific variables moderating the relationship between performance and dismissal. For example, Weisbach (1988), include corporate governance as a variable in this relationship, and find that the relationship is stronger for firms with more outside directors on the board. Another factor that has been tested is the properties of accounting information. The results of this research show that when information is of a high quality it will be used more rigorously when deciding about whether to retain or dismiss a CEO ( Engel et al., 2003).
However, little research up to now paid attention to the potential role of CEO tenure or reputation as a variable that may moderate the relationship between performance and dismissal of the CEO. So, this thesis aims to contribute by examining the role of CEO tenure in moderating the relationship between performance and CEO turnover.
1.4 Structure
The remainder of this paper will be structured as follows. Section two will discuss the prior literature and the hypothesis development. In section three the research methodology is given, it will contain a description of the sample selection, the empirical models, and the variables will be explained. Section four provides an overview of the empirical results. Finally, section five will present the conclusion and limitations of the research.
7
2. Literature review/hypothesis
This chapter reviews existing literature on CEO turnover and performance. First I explain the agency theory and two problems (moral hazard and adverse selection) that arise from this theory. Afterwards prior literature about CEO performance and turnover is reviewed. And finally, the hypothesis will be developed.
2.1 Agency theory, moral hazard and adverse selection
Most large firms nowadays are publicly held and therefore characterized by the separation of ownership and control. We will say that an agency relationship has arisen between two (or more) parties when one, designated as the agent, acts for, on behalf of, or as representative for the other, designated the principal (Ross, 1973). This yields some distinct benefits such as that investors have the possibility to diversify their investments across different firms such that the risk profile of their aggregate investment portfolio is lower. Furthermore, specialized managers can be recruited for managing large firms on a day-to-day basis. The agency problem occurs when cooperating parties have different goals and different vision of labor. One party (the principal) delegates work to another party (the agent), who performs that work. (Jensen and Meckling, 1976). However, the most prevalent problem that follows from the separation of ownership and control is information asymmetry. The managers are now the corporate insiders and important stakeholders such as investors have difficulties monitoring the firm. Two problems specifically emerge in publicly held firms, i.e., moral hazard and adverse selection. Moral hazard refers to the problem that the interests between managers and investors are not perfectly aligned. Therefore, investors need to obtain information to verify whether managers have acted in their best interest. Performance measures and the disclosure of the outcomes on those performance measures play an important role. That is, on the basis of performance measure outcomes, investors can infer whether the managers have worked hard and beforehand incentives can be tied to performance measures to provide incentives to take those actions, which are in the best interest of the firm. Adverse selection refers to the problem that investors have little information about the quality of the firm and the respective executives running the firm. Still, information about the quality of the firm and executives is important to infer what the future prospects of the firm are, and subsequently to make informed decisions on their equity portfolios
8
(buy, sell, hold) (Darrough et al., 1986). In addition, investors can make decisions such as replace corporate managers to provide them with incentives to work hard (scapegoat explanation) and to improve the quality of the executive team running the firm (managerial ability explanation). These explanations will be more elaborately discussed in section 2.2.1. So, performance measures outcomes will be used to infer whether the abilities of the managers match the requirements of the firm. Bad performance measure outcomes may prompt stakeholders to replace incumbent managers. This will be discussed in greater detail in section 2.2.
2.2 Relationship between performance and CEO turnover
Section 2.2.1 discusses the theory that explains the relationship between performance and turnover. Section 2.2.2 will subsequently review the empirical literature on the relation between performance and turnover.
2.2.1 Performance and CEO turnover
There are two alternative explanations for the negative relationship between performance and CEO turnover, that is, the managerial ability hypothesis and the scapegoat hypothesis. The improved management hypothesis or managerial ability hypothesis holds that forced management turnover tends to increase managerial quality and therefore expected firm performance. Under this hypothesis, quality, which is not directly observable, varies across managers. Firm directors attempt to infer quality from realized performance. If performance is sufficiently poor, the board infers that the incumbent manager is of low quality and that the expected benefit of replacing him exceeds the expected cost. Another manager is installed whose expected quality exceeds that of his predecessor. Moreover, poor performance tends to coincide with bad luck as well as low manager quality. Thus, future performance is expected to increase following the change in management for two reasons: the expected increment in manager quality is positive and manager luck is expected to revert to normal (Huson et. al., 2004). The scapegoat hypothesis is based on the agency models of Hölmstrom (1979), Shavell (1979), and Mirrlees (1976). The hypothesis holds, that quality does not vary across managers. Poor performance under the scapegoat hypothesis arises from chance alone rather than low managerial
9
quality. In other words, poor performance results from bad luck, not bad management. Under the scapegoat hypothesis, managers dislike effort so they must be threatened with dismissal if performance is low. In equilibrium, all managers supply the same effort (quality) and only those who are unlucky are fired. Boards of directors understand that all managers are alike, but must fire managers of poorly performing firms to induce other managers to provide the desired level of effort. Since replacement candidates only equal the quality of the outgoing manager, the turnover itself does not increase managerial quality or expected firm performance. Consequently, a manager who is fired for poor performance can be viewed as a scapegoat. Even though turnover does not increase managerial quality, the expected change in firm performance following turnover is positive. This is because turnover is triggered by improbably small performance outcomes arising from chance. Subsequent performance should revert to mean levels (Huson et. al., 2004).
So, the scapegoat hypothesis is about addressing the moral hazard problem, while the managerial ability hypothesis is about addressing the adverse selection problem. Formulated differently, even when the new manager is not of better quality than the old one, you may still have a reason to fire, that is, to keep dismissal as a credible threat for the other managers so that they know they will be fired when performance is low. As a result they will be working hard to keep the jobs.
2.2.2 Measures used as indicator of performance in the relationship between performance and CEO turnover
The paper of Warner et al. (1988) is one of the first papers on the relation between performance and CEO turnover. The major hypothesis they examine is that the probability of a top management change is inversely related to stock price performance. Although top managers’ contribution to firm value is not directly observable, stock returns are a potential source of information on the ability of the CEO. The authors use stock market performance as indication of the ability of the CEO. Furthermore, the authors distinguish between three internal mechanisms to support their hypothesis. The first one is monitoring by the board of directors. The second is mutual monitoring among firm’s managers and the third one is monitoring by holders of large
10
share blocks. The authors predict that if these mechanisms are effective there will be a negative relation between the probability of a top management change and share performance. They find support for their hypothesis.
Coughan and Schmidt (1985), also examine the relationship between CEO turnover and performance. A board has the power to effect a change in management. If boards discipline managers for actions or results that harm shareholders, stock price performance will be a predictor of changes in management. The empirical confirmation of such a relation, however, is complicated because there exist many other possible reasons for a change in top management: attraction of a successful executive to a better-paying position at another firm; normal retirement; or death. The hypothesis, the frequency of CEO turnover is related to past stock price performance is supported.
Weisbach (1988), examines the relation between performance and CEO turnover using different performance measures as indication of CEO ability. The first measure they use, consistent with prior literature, is stock return. As second measure of corporate performance is accounting earnings. Despite the many problems with using earnings data as measure of profitability, earnings data have one large advantage over stock price data for the purposes of measuring the performance of the CEO: earnings data measure short-term profits. The stock price reflects the present discounted value of the expected future cash flows of the company. The results show that there is a relation between both stock returns and changes in earnings and the probability that a CEO will be replaced.
Furthermore, Farrell and Whidbee (2003), examine CEO turnover and replacement decisions from a different perspective by examining the role of performance expectations. They argue that 1-year analyst forecast errors (the deviation of realized earnings from expected earnings) provide additional information regarding a CEO’s performance beyond simple earnings. To the extent that earnings forecast proxy for the board of director’s earnings expectations, forecast errors may capture the component of firm performance that the board attributes, in large part, to CEO performance. The results suggest that CEOs are not simply held accountable for the overall level of firm performance, but that boards of directors also use firm performance expectations as part of their criteria for evaluating CEO performance. Further, when industry-adjusted 1-year analyst
11
forecast dispersion is low or the number of analyst following a firm is high, CEOs with negative forecast errors face a greater likelihood of turnover.
2.3 Moderators in the relationship between performance and CEO turnover
Section 2.3.1 discusses the current moderators in the relationship between performance and turnover. Section 2.3.2 will subsequently look at tenure as a moderator and afterwards the hypothesis is stated.
2.3.1 Current moderators in the relationship between performance and CEO turnover
Some research has explored which variables influence the degree in which poor performance leads to the decision to oust the incumbent CEO.
Engel et al. (2003), examine if the relationship between performance and CEO turnover is affected by properties of the firm’s accounting system. They expect that when accounting is more informative about managerial performance, the board of directors should rely more heavily on accounting returns in making decisions about continuation of CEO employment. Hence turnover probability should increase with decreases in accounting returns in firms where accounting information is a better measure of managerial performance. They basically find that when information is of higher quality, the board of directors will use it more rigorously when deciding about whether to retain or dismiss the CEO.
Weisbach (1988), extends the research of Warner et al. (1988) and Coughan and Schmidt (1985). They are motivated to do so as prior research does not explore the differences in monitoring between managers who serve as directors who are full-time employees of the company (inside directors) and directors who are not full-time employees of the company. The hypothesis tested is: ‘inside and outside directors behave differently in their decisions to remove top management’. Boards of directors are widely believed to play an important role in corporate governance, particularly in monitoring top management. Directors are supposed to supervise the actions of management, provide advice and veto poor decisions. The board is the shareholders first line of defence against incompetent management. In extreme cases it will replace an errant CEO. The findings suggest that firms with outsider-dominated boards are significantly more likely than
12
firms with insider-dominated boards to remove the CEO on the basis of performance, as measured by such publicly available measures as earnings or stock returns.
2.3.2 CEO tenure as moderator in the relationship between performance and CEO turnover
As described in the previous paragraph, prior research focusing on the relationship between performance and CEO turnover examined the different performance measures used and different moderators influencing the relationship. However little research look at tenure as a moderator influencing this relationship. CEO’s are endowed with a given level of talent. The CEO and the firm have common knowledge about the distribution over CEO talent, but neither party knows the actual level of CEO talent. So, when a company hires a CEO they don’t know if the talent of the CEO match the requirements of the firm and they have to learn and update this over time (Gibbons and Murphy, 1992; Holmstrom, 1999; and Hermalin and Weisbach, 2008).
Likewise, Milbourne (2002), states that a CEO’s true underlying ability is unobservable. And thus market participants form beliefs over CEO ability and update them according to Bayes rule as new information is observed. The firm generates a single, noisy terminal cash flow that depends on who is the CEO in place at the end of the game. CEO tenure is positively related to the market’s perception of the CEO’s ability, it implies that the firm’s board has historically been inclined to retain this executive.
Further, Bushman et al. (2010) look at the role played by performance risk in impacting a board’s ability to learn about a CEO’s unknown talent. The fundamental insight of the paper is that the impact of performance risk on the ability of boards to learn about CEO talent from firm performance depends crucially on the underlying sources of the risk. The idea is that if volatility in performance outcomes is driven primarily by unobservable CEO talent, firm performance is diagnostic about such talent, allowing boards to accurately assess CEO talent and to replace low talent incumbents. However, if volatility in performance out- comes is driven by factors unrelated to CEO talent (e.g., noise, economy-wide effects, etc.), then a board’s ability to infer CEO talent from performance is more limited, making it difficult to cleanly distinguish an incumbent’s talent level from the assessed talent of potential replacement CEO’s. From this we
13
can conclude that if volatility in performance outcomes is driven by unobservable CEO talent a board can learn and assess the CEO’s talent and replace him. As tenure increase, the firm have had more time to learn about the CEO’s talent.
In addition Jiang and Lee (2011), examine the association between CEO reputation and corporate capital investments. They find that the stock market’s responses to announces of capital investments are more favourable for firms with more reputable CEO’s. Because CEO’s build up reputational capital over their career via their repeated dealing with capital market participants. In the context of capital investments, more reputable CEO’s are more likely to select positive net present value projects because the successful corporate capital investments enhance CEO’s expected compensation in the managerial labour market. In other words, CEO reputation helps to mitigate problems arising from information asymmetry between the firm and the investors. Moreover, a more reputable CEO may have higher ability to obtain more precise information on investment opportunities, make better investment choices and achieve successful project outcomes with greater likelihood. And CEO reputation rises over his or her tenure.
So in the end, when a company hires a CEO they don’t know exactly what his or her capabilities are. The company update the beliefs about the ability of the CEO over time. In addition the reputation of a CEO is build over time with the accomplishment of achievements. These two arguments imply that the firm and the market learn about the capabilities over his tenure. And CEO’s with a longer tenure have more opportunities to show the firm that his or her capabilities match with what is required. Therefore I expect that tenure has a negative moderating effect on the relationship between performance and CEO turnover. The hypothesis I would like to test is as follows.
H1: CEO tenure negatively moderates the negative relationship between firm performance and CEO turnover.
14
3. Research Methodology
In this chapter, section 3.1 describes the sample, afterwards, in section 3.2 the variable measurement is described. In section 3.3 the empirical model used is given. And finally in section 3.4 some descriptive statistics are given.
3.1 Sample
The sample in this research consists of companies that appear in the Execucomp database which is part of the Compustat database. This database covers roughly S&P1500 companies. The reason for using Execucomp is that it allows me to gather the data needed for calculating the tenure of CEO’s. In Execucomp there is data available on the starting and ending dates of CEO’s. This data is needed to calculate tenure, on which the focus is in this thesis. Because Execucomp data starts from 1992, the sample will cover a period between 1992 and 2012. Accounting data on firm performance is obtained from the Compustat database. For the stock market performance, stock price data is obtained from the Center for Research in Security Prices (CRSP).
The sample covers 35,607 firm years, due to key missing data in the observations, 1,327 observations are dropped.
Furthermore after merging the Execucomp data with the Compustat data on firm performance I lose another 1,193 observations. Second, after merging the data with stockprice information from the CRSP database I lose another 18,117 observations. Finally 156 duplicate observations are deleted, which leaves a sample of 14,814 observations.
Finally to distinguish between routine and non-routine CEO changes, I remove observations that are routine. Execucomp provides the reason why a CEO left. There are four options: unknown, resigned, retired or deceased. I classify unknown and resigned as non-routine changes, and classify retired and deceased as routine changes which I remove from the sample.1 After removing the 2,013 observations that are routine changes the sample consists of 12,747
1 I acknowledge that classifying turnover as routine versus non-rountine is somewhat ambiguous. That is, not in all
cases this reason is given and therefore the difficulty remains to determine which turnovers are routine or non-routine.
15 observations.
Because turnover is defined as a CEO change in the year following the year for each observation I don’t have turnover data for 2012. Which leaves me with a total sample of 11,190 observations over the period from 1992 to and including 2011. Of all CEO’s 8.84% experienced a change, which equals to 989 turnovers.
3.2 Variable measurement
3.2.1 Measures of firm performance
In prior research different measures of firm performance are used. Some researchers use market measures and some use accounting measures. Other researchers used both market and accounting measures. In this thesis three performance measures are used. Following Allgood and Farrell (2000), return on assets (ROA) and earnings per share (EPS) are used as accounting measures for firm performance. Where return on assets is calculated by earnings before interest and taxes divided by total assets. Further stock performance is used as a market measure of firm performance. To calculate stock performance I use year-end stockprices and calculate the percentual change in stockprice each year.
3.2.2 CEO turnover
In this study the dependent variable is CEO turnover. This variable takes a value of 1 if there was a CEO change and a 0 if there was no change. When a variable has only two outcomes it is named a dummy or dichotomous variable. As mentioned before, CEO turnover is coded only for CEO changes that are regarded as non-routine changes (see section 3.1). As consistent with prior literature, I expect that CEO turnover is higher for companies where performance is poor.
3.2.3 CEO Tenure
In this study I will test the effect of tenure on the relationship between performance and CEO turnover. I expect that tenure has a negative effect on this relationship. To determine the tenure of a CEO, the start date as a CEO is deducted from the end date as a CEO. If the CEO doesn’t change, the ultimate datum of the dataset represents the end date. I will measure tenure with a
16
dummy variable which equals to 1 if tenure is greater than the median and which equals 0 if tenure is equal or lower than the median. So the value 1 represents a long tenure and the value 0 represents a short tenure.
3.2.4 Control variables
In prior research, the most commonly reported reason for departure is retirement. Despite that I tried to address for this by excluding routine turnovers from my sample, I also try to control for this by making use of the control variable Age which represents the age of the departing CEO. Furthermore a dummy variable for gender is created to control for a distinction between male and female CEO’s.
3.3 Empirical model
The hypothesis is tested by means of the following empirical model. Consistent with most prior work studying the relationship between performance and CEO turnover a logit regression model approach will be used. The model is as follows:
CEO_Turnover = β0 + β1ROA + β2ROA*Tenure + β3EPS + β4EPS*Tenure +
β5StockPerformance + β6Stockperformance*Tenure + β7Tenure + Controls + ε
Where,
Turnover = An indicator variable equal to one if there is a CEO change in the year following the Year for each observation
ROA = ROA, calculated as the ratio of accounting earnings before interest and
taxes to total assets
EPS = Earnings per share at Year-end
Stockperformance = Percentual change in stockprice for that Year
Tenure = Dummy variable which takes the value of one if a CEO’s tenure is greater than the median. And the value of zero if a CEO’s tenure is equal or lower
than the median.
17
Gender= Dummy variable which equals to 1 if the CEO is a male and 0 if the CEO
is a female.
The coefficients β1, β3, and β5 give the relationship between respectively ROA, EPS, and stock
market performance on the one hand and the likelihood of CEO turnover on the other hand for CEO’s with a short tenure (i.e., shorter than the median). The sum of coefficients (β1 + β2), (β3 +
β4), and (β5 + β6) give the relationship between respectively ROA, EPS, and stock market
performance on the one hand and the likelihood of CEO turnover on the other hand for CEO’s with a long tenure. So, the coefficients β2, β4, and β6 represent the difference in the relation
between the three performance indicators in use and the likelihood of CEO turnover between short tenure CEO’s and long tenure CEO’s. On the basis of my hypothesis, I expect that β2<0,
18
3.4 Descriptive statistics
Tabel 1 shows the CEO observations per year.
Table 1: CEO observations
Of the 11,190 CEO observations over the period from 1992 to and including 2011, 8.84% CEOs experienced a change, which equals to 989 turnovers. Generally, CEO turnover percentages are relatively constant over the years, with percentages that typically vary between 7% and 12%.
19
Table 2 reports the descriptive statistics for the variables that will be used in the regression analyses.
Table 2: Descriptive statistics
Variable | Obs Mean Std. Dev. 10% Median(50%) 90% ---+--- CEO_Turnover | 11190 .0883825 .283863 0 0 0 Age | 11190 54.92019 7.486138 46 55 64 ROA | 11190 .0873643 .1165318 .0027637 .085396 .201079 EPS | 11190 1.326625 2.987729 -.67 1.21 3.74 Stockperf~ce | 11190 .1330597 1.146424 -.443618 .0338783 .608487 Tenure | 11190 12.61704 8.94751 3.654794 10.50959 25.0164
The average value of CEO_Turnover is 0.0884, which represents the average percentage of 8.84% of the CEO observations that experienced a turnover. Furthermore, the average CEO had an age of 55 years. The average value of ROA and EPS is 0.0873643 and 1.326625 which means that the average firm has a ROA of 8.73% and EPS of 1.33. Respectively. The average market performance is 0.01330597 in my sample. For Tenure the average is 12.61 years and the median is 10.50959 so CEO observations that were larger than 10.50959 were classified as having long tenure (Tenure=1) while CEO observations where the tenure was equal or smaller than 10.50959 were classified as having short tenure (Tenure=0).
4. Empirical results
In this chapter I will present the logit regression results for the likelihood of turnover events. With two accounting measures of performance (ROA and EPS) and one market measure of performance (Stockperformance) I test if CEO’s with a shorter tenure are more likely to be replaced relative to CEO’s with a long tenure.
20
First I present the logit regression results for the performance measure ROA. Table 3 reports the coefficients and significance level of the coefficients.
Table 3: Logit regression with ROA as performance measure
The findings seem to support the hypothesis. The coefficient on ROA is negative and significant (p<0.05), which suggests that for CEO’s with a short tenure when ROA performance decreases the likelihood of being replaced increases. However, the coefficient on ROA*Tenure, which gives the difference in the relationship between performance and CEO turnover for short vs. long tenure CEO’s is positive and significant (p<0.05). This means that for CEO’s with a long tenure, the negative relationship between performance and the likelihood of CEO turnover is weakened. So, the findings indeed, suggest that tenure negatively moderates the negative relationship between performance and CEO turnover. With respect to the controls, it seems that the
likelihood of CEO turnover is greater when CEO’s are older. This may represent early retirement or older CEO’s having trouble meeting the imposed performance expectations.
r h o 8 . 7 9 e - 0 7 . 0 0 0 0 1 4 2 . 3 5 e - 2 0 1 sigma_u .0017003 .0135572 2.78e-10 10408.88 /lnsig2u -12.75387 15.94657 -44.00857 18.50083 _ c o n s - 6 . 7 5 7 6 5 9 . 3 7 7 0 5 0 7 - 1 7 . 9 2 0 . 0 0 0 - 7 . 4 9 6 6 6 4 - 6 . 0 1 8 6 5 3 Gender_dummy .0503354 .2645112 0.19 0.849 -.4680971 .5687678 A g e . 0 8 3 7 7 4 7 . 0 0 4 9 5 5 4 1 6 . 9 1 0 . 0 0 0 . 0 7 4 0 6 2 4 . 0 9 3 4 8 7 1 T e n u r e _ d u m m y - 1 . 2 3 1 5 . 0 9 8 7 2 5 - 1 2 . 4 7 0 . 0 0 0 - 1 . 4 2 4 9 9 8 - 1 . 0 3 8 0 0 3 ROAxTenure 1.257134 .632102 1.99 0.047 .0182372 2.496031 ROA -.7969299 .3915706 -2.04 0.042 -1.564394 -.0294657 CEO_Turnover Coef. Std. Err. z P>|z| [95% Conf. Interval]
21
Second, I present the logit regression results for the performance measure EPS. Table 4 reports the coefficients and significance level of the coefficients.
Table 4: Logit regression with EPS as performance measure
The findings seem to support the hypothesis. The coefficient on EPS is negative and significant (p<0.01), which suggests that for CEO’s with a short tenure when EPS performance decreases the likelihood of being replaced increases. However, the coefficient on EPS*Tenure, which gives the difference in the relationship between performance and CEO turnover for short vs. long tenure CEO’s is positive and significant (p<0.02). This means that for CEOs with a long tenure, the negative relationship between performance and the likelihood of CEO turnover is weakened. So, the findings, indeed, suggest that tenure negatively moderates the negative relationship between performance and CEO turnover. With respect to the controls, it seems that the likelihood of CEO turnover is greater when CEOs are older. This may represent early retirement or older CEOs having trouble meeting the imposed performance expectations.
.
rho 1.09e-06 .0000152 1.57e-18 .9999987 sigma_u .0018978 .0132021 2.27e-09 1583.519 /lnsig2u -12.53411 13.91297 -39.80302 14.73481 _cons -6.756231 .3766032 -17.94 0.000 -7.49436 -6.018102 Gender_dummy .063588 .2649975 0.24 0.810 -.4557977 .5829736 Age .0841586 .0049407 17.03 0.000 .0744751 .0938421 Tenure_dummy -1.202353 .0849259 -14.16 0.000 -1.368805 -1.035902 EPSxTenure .0614426 .0242635 2.53 0.011 .013887 .1089981 EPS -.0806557 .0163829 -4.92 0.000 -.1127656 -.0485457 CEO_Turnover Coef. Std. Err. z P>|z| [95% Conf. Interval]
22
Third I present the logit regression results for the performance measure Stockperformance. Table 5 reports the coefficients and significance level of the coefficients.
Table 5: Logit regression with Stockperformance as performance measure
The findings don’t seem to support the hypothesis. The coefficient on Stockperformance is negative and significant (p<0.01). Which suggest that for CEO’s with a short tenure when stockperformance decreases the likelihood of being replaced increases. However, the coefficient on Stockperformance*Tenure, which gives the difference in the relationship between performance and CEO turnover for short vs. long tenure CEO’s is positive but not significant (p>0.05). This means that for CEOs with a long tenure, the negative relationship between performance and the likelihood of CEO turnover is not weakened. So, the findings, does not, suggest that tenure negatively moderates the negative relationship between performance and CEO turnover. With respect to the controls, it seems that the likelihood of CEO turnover is greater when CEOs are older. This may represent early retirement or older CEOs having trouble meeting the imposed performance expectations.
rho 2.09e-06 .0000247 1.82e-16 .9999584 sigma_u .0026237 .0155045 2.45e-08 281.1749 /lnsig2u -11.88633 11.81873 -35.05061 11.27795 _cons -6.759455 .3875272 -17.44 0.000 -7.518994 -5.999915 Gender_dummy -.0273209 .2659386 -0.10 0.918 -.548551 .4939093 Age .0843337 .0052161 16.17 0.000 .0741102 .0945571 Tenure_dummy -1.07233 .0841522 -12.74 0.000 -1.237265 -.9073946 StockperformancexTenure .0130559 .1756037 0.07 0.941 -.3311211 .3572329 Stockperformance -.3494058 .1072173 -3.26 0.001 -.5595478 -.1392637 CEO_Turnover Coef. Std. Err. z P>|z| [95% Conf. Interval]
23
Finally, I present the logit regression results for the all the performance measures together. Table 6 reports the coefficients and significance level of the coefficients.
Table 6: Logit regression with ROA, EPS and Stockperformance as performance measures
The findings don’t seem to support the hypothesis. The coefficient on EPS and Stockperformance are negative and significant (p<0.01), which suggests that for CEOs with a short tenure when EPS performance or Stockperformance decreases the likelihood of being replaced increases. The coefficient on ROA is positive and not significant (p>0.05), so here I can’t suggest that for CEOs with a short tenure when ROA decreases the likelihood of being replaced increases. Looking at the coefficients ROA*Tenure, EPS*Tenure and Stockperformance*Tenure, which gives the difference in the relationship between performance and CEO turnover for short vs. long tenure CEO’s not one of the coefficients is significant (p>0.05). So, the findings, does not, suggest that tenure negatively moderates the negative relationship between performance and CEO turnover.
To summarize, the results support the hypothesis for the accounting measures of performance (ROA and EPS), but not for the market measure of performance (Stockperformance). Furthermore if we combine all the performance measure into one model, the hypothesis is also not supported.
rho 2.10e-06 .0000244 2.46e-16 .9999439 sigma_u .0026257 .0153149 2.85e-08 242.1117 /lnsig2u -11.88481 11.66532 -34.74841 10.9788 _cons -6.736976 .3886179 -17.34 0.000 -7.498653 -5.975299 Gender_dummy -.029325 .2664014 -0.11 0.912 -.5514622 .4928121 Age .0851654 .0052125 16.34 0.000 .0749492 .0953817 Tenure_dummy -1.192309 .1048322 -11.37 0.000 -1.397776 -.9868417 StockperformancexTenure -.0402062 .1734659 -0.23 0.817 -.3801932 .2997808 Stockperformance -.2920827 .1015905 -2.88 0.004 -.4911964 -.092969 EPSxTenure .0492413 .0278467 1.77 0.077 -.0053371 .1038198 EPS -.0714749 .0188354 -3.79 0.000 -.1083916 -.0345581 ROAxTenure .5845755 .7437194 0.79 0.432 -.8730878 2.042239 ROA .2253636 .4777484 0.47 0.637 -.711006 1.161733 CEO_Turnover Coef. Std. Err. z P>|z| [95% Conf. Interval]
24
5. Summary and conclusion
This thesis examines the effect of tenure on the relationship between firm performance and CEO turnover. Prior research demonstrates the negative relationship between firm performance and CEO turnover. The goal of this thesis is to empirically investigate if tenure has a negative influence on this relationship. There are two reasons why I expected a negative effect of tenure on this relationship. First when a company hires a CEO the company doesn’t know what the abilities of the CEO are and if this match with what the company expects and demands. Furthermore, a CEO builds a reputation over time and therefore CEO’s with a longer tenure have had time to build a reputation compared with CEO’s with a short tenure.
The hypothesis examined in this paper is: ‘CEO tenure negatively moderates the negative relationship between firm performance and CEO turnover’. Using a sample of 11.190 firm years of American companies that appear in the Execucomp database between 1992-2012, I examined the relationship looking at three different performance measures. The performance measures return on assets and earnings per share are used as accounting-based measures. Further, stockperformance is used as a market-based measure.
The results of the accounting measures of performance (ROA and EPS) support the hypothesis. And shows, indeed, that tenure negatively moderates the negative relationship between
performance and CEO turnover. However, the results on the market measure of performance (stockperformance), does not support the hypothesis.
Finally, I acknowledge that classifying turnover as routine versus non-routine is somewhat ambiguous. That is, not in all cases this reason is given and therefore the difficulty remains to determine which turnovers are routine or non-routine. As recommendation for future research, the turnover events could be investigated more thoroughly for a better distinction between routine and non-routine turnovers.
25
6. References
Alia, A., T. Chen and S. Radhakrishnan (2007). Corporate disclosures by family firms. Journal
of Accounting and Economics. 44. 238–286.
Allgood, S. And K.A. Farrel (2000). The effect of CEO Tenure on the Relation between firm performance and turnover. Journal of Financial Accounting Research. 23. (3). 373-390. Brickley, J. A. (2003). Empirical research on CEO turnover and firm-performance: A discussion. Journal of Accounting and Economics. 36. (1–3). 227–233.
Bushman, R., Z. Dai, X. Wang (2010). Risk and CEO turnover. Journal of Financial
Economics. 96. 381-398.
Coughlan, A.T. and R.M. Schmidt (1985). Executive compensation, management turnover, and firm performance. Journal of Accounting and Economics. 7. 43-66.
Darrough, M.N. and N.M. Stoughton (1986). Moral hazard and Adverse selection: The question of financial structure. The journal of finance. 41. (2). 501-513.
Donaldson, L. and J.H. Davis (1991). Stewardship Theory or Agency Theory: CEO Governance and Shareholder Returns. Australian Journal of Management. 16. (1). 49-64.
Eisenhardt, K.M. (1989). Agency theory: An assessment and review. The academy of
management review. 14. (1). 57-74.
Engel, E., R. M. Hayes, X. Wang (2003). CEO turnover and properties of accounting information. Journal of Accounting and Economics. 36. 197-226.
Farrell, K.A., D.A. Whidbee (2003). Impact of firm performance expectations on CEO turnover and replacement decisions. Journal of Accounting and Economics. 36. 165–196.
Hermalin, B. E. and M. S. Weisbach (1998). Endogenously chosen boards of directors and their monitoring of the CEO. American Economic Review. 88. (1). 96–118.
Hölmstrom, B. (1979). Moral hazard and observability. The Bell Journal of Economics. 10. (1). 74-91.
Huson, M.R., P. Malatesta, R. Parrino (2004). Managerial succession and firm performance. Journal of Financial Economics. 74. (2). 237-274.
Jensen, M. and W. Meckling (1976). Theory of the firm: Managerial behavior, agency costs, and ownership structure. Journal of Financial Economics. 3. 305-360.
26
Corporate Finance. 17. 929–946.
Kang, J. and A. Shivdasani (1995). Firm performance, corporate governance, and top executive turnover in Japan. Journal of Financial Economics. 38. 29-58.
Kaplan, S.N. (1994). Top Executives, Turnover, and Firm Performance in Germany. Journal of
Law, Economics and Organization. 10. (1). 142-159.
Lausten M. (2002). CEO turnover, firm performance and corporate governance: empirical evidence on Danish firms. International Journal of Industrial Organization. 20. 391-414. Milbourn, T.T. (2002). CEO reputation and stock-based compensation. Journal of Financial
Economics. 68. (2). 233-262.
Mirrlees, J.A. (1976). The optimal structure of incentives and authority within an organization. The Bell Journal of Economics. 7. 105-131.
Murphy, K. J. and J. L. Zimmerman (1993). Financial performance surrounding CEO turnover.
Journal of Accounting & Economics. 16. (1–3). 273–315.
Ross, S.A. (1973). The economic theory of agency: The principal’s problem. The American
Economic Review. 63. (2). 134-139.
Shavell, S. (1979). Risk sharing and incentives in the principal and agent relationship. The
Bell Journal of Economics. 10. 55-73.
Warner J.B., R.L. Watts, K. H. Wruck (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. 20. 431-460.
