−
Faculty of Economics and Business
Accounting and Control
−
The moderating effect of reputation
on the relationship between
performance and CEO turnover
VERSION 2.2 – Final DraftAuthor: Willie Huang Student ID: 10419616 Thesis supervisor: P. Kroos Specialization: Control Date of completion: 30.07.2014
Abstract:
This study examines whether CEO reputation has a negative moderating effect on the relationship between poor performance and turnover by conducting archival-based research. Using a sample of 970 observations for the time period 2007 to 2013, I found no significant results that could support the hypothesis on the 5 percent confidence level. However, the empirical results show mixed results for the hypothesis on 10 percent confidence level, when industry- and year effects were taken into account. In which, I found a negative moderating effect of reputation on market-based measures and turnover. This negative relationship could not be found for accounting-based measures. Instead, I found a positive moderating effect of reputation on market-based measures and turnover.
Keywords:
Table of Contents
1. INTRODUCTION ... 5
1.1 BACKGROUND ... 5
1.2 RESEARCH QUESTION ... 6
1.3 MOTIVATION AND CONTRIBUTION ... 6
1.4 STRUCTURE ... 7
2. LITERATURE REVIEW AND HYPOTHESES ... 8
2.1 SEPARATION OF OWNERSHIP AND CONTROL AND INFORMATION ASYMMETRY ... 8
2.2 PERFORMANCE AND CEO DISMISSAL ... 9
2.2.1 Managerial ability and scapegoat ... 9
2.2.2 Overview of empirical studies ... 10
2.3 MODERATORS IN PERFORMANCE-CEO TURNOVER RELATIONSHIP ... 10
2.4 CEO REPUTATION AS MODERATOR ... 11
3. METHODOLOGY ... 13 3.1 SAMPLE SELECTION ... 13 3.2 EMPIRICAL MODEL ... 14 3.3 VARIABLE MEASUREMENT ... 14 3.3.1 Dependent variable ... 14 3.3.2 Independent variables ... 14 3.3.3 Control variables ... 15 4. EMPIRICAL RESULTS ... 16 4.1 DESCRIPTIVE RESULTS ... 16 4.2 MAIN FINDINGS ... 21 4.3 SUPPLEMENTAL ANALYSIS ... 24 5. CONCLUSION ... 25 BIBLIOGRAPHY ... 26
APPENDIX I: LIST OF VARIABLES ... 28
1. Introduction
1.1 Background
Chief Executive Officers (CEOs) are considered as the most important executive of a firm. They represent and bear the image of the firm (Ranft et al., 2006; Perryman et al., 2010). Replacing a CEO and appointing a new CEO can raise questions about who should succeed this position. The board of directors has to make decisions whether to retain the CEO or find a replacement (Weisbach, 1988). The issues that follow from this decision-making process are strongly related to the Agency Theory. This theory assumes information asymmetry related issues between the principal and agent, which can be divided into adverse selection and moral hazard (Eisenhart, 1989). Adverse selection occurs when the principal has no indication whether the claimed abilities and skills of the agent match the requirements of the firm. Through repeated actions and achievements by the CEO the board can learn about the true abilities of the CEO. Moral hazard happens after contractual signing of the agent. It is described by Eisenhart (1989, p. 61) as the lack of effort on the apart of the agent; the agent might not deliver the agreed upon effort.
To address the concerns that follow from adverse selection and moral hazard, the board sets up requirements and performance measures to evaluate the abilities of the CEO. Good performance leads to an extended tenure and bad performance often to a dismissal. Furthermore, performance measures give CEOs specific incentives to perform certain actions. Over time these actions will translate into achievements that build up CEO reputation. One could say that reputation is a collection of past demonstrations of managerial ability. Reputation is an important factor that thrives the performance of a CEO. This is explained within the Efficient-contracting Theory (Jian & Lee, 2011).
Efficient-contracting Theory assumes that CEO reputation is build over time with achievements. By presenting good results the CEO sends a message to the market for decision agents that they are adequate. Thus, increasing their value and strengthen their position in the labor market. Also, CEOs with strong reputational capital are more likely to select projects with a positive net present value that will benefit the firm (Jian & Lee, 2011). Furthermore, Good performing CEOs tend to have relative long tenures and as the length of a CEO’s tenure progresses the CEO is gaining more power in the firm (Zajac & Westphal, 1996). Powerful CEOs are able to control the board and influence director appointments. This may increase the danger of managerial entrenchment and it may increase the difficulty of removing the CEO since the fate of the insiders are often linked to the CEO’s. Weisbach (1988) explains that insider-dominated boards are less likely to dismiss poor performing CEOs.
In most cases the board has beforehand no indication about the abilities and behaviors of the newly appointed CEO (Bushman et al., 2010). The board tries to observe the managerial ability to assess whether to retain or dismiss a CEO. Moreover, turnover may occur due to a wide variety of reasons. CEO turnover can
be driven by the board’s assessment that the CEO’s managerial ability is low or does not match the requirements of the firm (Engel et al., 2003). The board uses performance measures to examine the managerial ability of a CEO. Huson et al. (2004) provides two theoretical views regarding reasons for a dismissal. First, Managerial Ability Theory assumes that CEO-talent varies amongst CEOs. Therefore, replacing a poor performing CEO will increase the outcome of firm performance. Second, Scapegoat Theory assumes that all CEOs have same level of talent and therefore no managerial quality differences. Consequently, replacing a CEO does not improve firm performance (Huson et al., 2004). In this case, the board fires an incument CEO to maintain a credible threat of dismissal to the remaining employees. Henceforth, the CEO becomes an scapegoat for poor performance.
I argue that reputation is an important factor, which can insulate the CEO from dismissal following poor performance. High ability CEOs are more likely to have longer tenures in firms, rather than low ability CEOs. This is based on two assumptions. First, reputation is a long-term resource that is build over time. Second, a favorable reputation is a trait that can be used to generate value (Ranft et al., 2006).
1.2 Research question
The research question is based on two assumptions; (1) Reputation is a long-term resource that is build over time and (2) a favorable reputation is a trait that can be used to generate value (Ranft et al., 2006). The latter suggests that CEOs can use their reputation to influence the board’s retain- or dismissal decisions. Based on this assumption this paper will investigate the following research question:
“Does reputation insulate the CEO from dismissal following poor performance?” 1.3 Motivation and contribution
Prior literature looked into various matters regarding the relationship between performance and CEO turnover, but not whether building a reputation insulates the CEO from dismissal following poor performance. Several researchers have examined the relationship between performance measures and CEO turnover (Warner et al., 1988; Farrell et al., 2003; Engel et al., 2003). Warner et al. (1988) argue that stock price performance reflects the CEO’s ability to run the firm. They found small significant support that CEO turnover is inversely related to stock price performance. Other researchers suggest that analyst-forecast errors provide information regarding a CEO’s performance (Farrell et al., 2003). Any discrepancy between analyst-expected performance and realized performance is due to CEO’s lack of skill. While, Engel et al. (2003) uses various performance measures based on accounting- and market information to examine CEO turnover decisions. Weisbach (1988) used market-based measures as well to examine the relationship, but also
looked if corporate governance structure moderates the relationship between performance and CEO turnover.
Prior literature mainly focuses on the relationship performance measures and CEO turnover. There is little research done on whether reputation insulates the CEO from dismissal following poor performance. Cianci & Kaplan (2010) examined the relationship reputation and performance. The researchers conducted an experiment in which MBA students had to make judgments about CEO reputation following poor performance. The researchers concluded that reputation was not reduced if the CEO had a pre-existing reputation. Thus, I assume that the board might be tolerant against CEOs with a reputation.
This study contributes to the existing literature by examining whether reputation insulates the CEO from dismissal following poor performance. Therefore, filling the gap in literature with evidence by conducting archival-based research.
1.4 Structure
The remainder of this paper proceeds as follows: In section two discusses the prior literature and develops the hypotheses. Section three describes the data and sample selection procedures. Section four provides the analysis and the results. Summary and conclusion is presented in section five.
2. Literature review and hypotheses
2.1 Separation of ownership and control and Information asymmetry
Firms with numerous equity holders are characterized by the separation of ownership and control. As it is costly and not practical for all residual claimants, also known as equity holders to participate in decision control1 or decision
management2 (Fama & Jensen, 1983). Likewise, the ownership is broadly
distributed amongst equity holders that individual equity holders cannot exercise real power to oversee managerial performance (Demzetz, 1983). Instead, decision rights are delegated from the equity holders (principal) to the CEO (agent). In essence, the CEO receives the decision authority to act on behalf of the principal and is expected to act in the best interest of the firm. In organizational context the CEO performs decision management and the board of directors executes decision control. The segregation between decision management and the economic owners of a firm is named in literature the separation of ownership and control.
The establishment of separation of ownership and control within a firm creates agency problems. Prior literature, explains this concept with the Agency Theory (Eisenhart, 1989; Fama & Jensen, 1983). The theory states that agency problems revolve around (1) the misalignment of interests and (2) issues regarding verifying and monitoring the agent (Eisenhart, 1989). Agency issues can be subdivided into adverse selection and moral hazard.
The board carries the responsibility to appoint a capable CEO and assess their performance over time. In most cases the board is unfamiliar with the true abilities of a CEO. Therefore, the board is exposed to a certain level of information asymmetry. Agents have more or hidden information about their claimed abilities, while principals does not possess all agent-information, and mainly perceives information through observed performance. Therefore, the principal has beforehand no indication whether the claimed abilities and skills of the agent match the requirements of the firm. This is known as adverse selection. Furthermore, it can occur that CEOs do not deliver the agreed upon effort after contractual signing due to a lack of effort on the part of the agent (Eisenhart, 1989). Thus, interests of the CEO might be misaligned with the wishes of the equity holders. This assumption is based on the belief that equity holders and CEOs both participate in utility-maximization, whereas CEOs might be engaged in acts that benefit themselves, thereby pursuing own interests instead of the demands of residual claimants. This is also known as moral hazard.
Hence, boards are vulnerable to misrepresented reputations due to information asymmetry between the board and CEO-candidate (Dahlstrom & Ingram, 2003). The lack of CEO-candidate information causes the likelihood that the board makes a mismatch between CEO abilities and requirements of the firm (Zhang, 2008). Through repeated actions and achievements displayed by the CEO,
1Decision control: the body that performs ratification and monitoring of decisions.
2Decision management: the body that performs the initiation and implementation of decisions.
the board can update their beliefs and knowledge about the CEO. Subsequently, with the updated information the board can assess whether to retain or replace the incumbent CEO. However, firm performance and CEO ability are not seamlessly correlated, whereas the perceived CEO ability and firm performance can fluctuate over time (Harris & Holmstrom, 1982).
2.2 Performance and CEO dismissal
2.2.1 Managerial ability and scapegoat
It is reasonable to assume that CEOs possess a certain level of talent, which is displayed as managerial ability. The board tries to observe the managerial ability to assess whether to retain or dismiss a CEO. Moreover, turnover may occur due to a wide variety of reasons. CEO turnover can be driven by the board’s assessment that the CEO’s managerial ability is low or does not match the requirements of the firm (Engel et al., 2003). The board uses performance measures to examine the managerial ability of a CEO. Besides, Huson et al. (2004) provides two theoretical views regarding reasons for a dismissal, (1) Managerial Ability Theory and (2) Scapegoat Theory.
First, Managerial Ability Theory assumes that CEO-talent varies amongst CEOs. Therefore, the given level of CEO talent moderates the outcome of firm performance. The board tries to assess the CEO talent since it has no indication of the actual level beforehand. If uncertainty about the CEO ability rises or negative deviations between observed performance and expectations occur, then the likelihood of a departure rises. Huson et al. (2004) explains that CEOs that display poor performance are likely to have low ability or bad luck. Since, CEO quality varies amongst CEOs, replacing a low ability CEO will increase the expected firm performance. As for CEOs who are subject to bad luck are as well dismissed. The belief is that the dismissal of an unlucky CEO will bring a reversal of bad luck. Therefore, the Managerial Ability Theory addresses adverse selection problems.
Second, Scapegoat Theory suggests that all CEOs have same level of talent and therefore no managerial quality differences. Consequently, replacing a CEO does not improve managerial quality or firm performance (Huson et al., 2004). The assumption is that dismissal following poor performance is due to bad luck, rather than the low ability of the CEO. The board fires the unlucky CEO to maintain a credible threat of dismissal for the remaining employees. Subsequently, keeping the remaining employees motivated to extent their effort in performing their tasks. Henceforth, the CEO becomes a scapegoat for poor firm performance. Thus, the Scapegoat Theory addresses moral hazard problems.
2.2.2 Overview of empirical studies
Prior literature has extensively examined the relationship between performance measures and CEO turnover. A substantial amount of research has tried to identify various determinants, which influence CEO turnover (Warner et al., 1988; Engel et al., 2003; Farrell & Whidbee, 2003; Jenter & Kanaan, 2006; Kaplan & Minton, 2006). Research has found significant evidence that firm performance is an important determinant of CEO turnover.
Warner et al. (1988) examined the relation between a firm’s stock returns and top management changes. They assume that stock price performance reflects the managerial ability, since the proposition is that performance measures used are related to stock returns. Thus, they expect a negative relation between the probability of CEO turnover and firm’s share price. Furthermore, the researchers found significant support for the hypothesis that earnings are a significant predictor of turnover (Warner et al., 1988).
Other researchers, such as Farrel & Whidbee (2003) argue that analyst forecast errors – deviations from realized and expected firm performance – provide additional information about CEO’s managerial quality. The researchers assume that analyst forecasts represent a fair portion of the board’s expectations (Farrell & Whidbee, 2003). Therefore, they argue that any deviation from analyst forecasts and board expectations are caused by the low managerial ability of the CEO. The researchers found an inverse relationship between industry-adjusted earnings-per-share (EPS) growth rate forecasts and the likelihood that an outsider is appointed CEO. The researchers state the CEO replacement decision can be explained by EPS, but only then there is uncertainty amongst analysts regarding firm’s future performance.
More recent studies, such as Jenter & Kanaan (2006) used relative peer performance measures to examine the relationship. They focused on determinants, such as industry stock returns and market returns. Their results indicate that the likelihood of a forced turnover is determined by performance of CEO-peers. Thus, CEOs who underperform their peer group experience a higher likelihood of dismissal. The researchers also found that boards partially filter industry performance from their assessment of CEO quality, but are according to the researchers unsuccessful in filtering exogenous shocks.
Kaplan & Minton (2006) extend the research on relative performance. They provided evidence that boards not only base their retention decisions on poor performance relative to the industry, but also to poor industry and to poor market performance (Kaplan & Minton, 2006).
2.3 Moderators in performance-CEO turnover relationship
Numerous studies have examined various moderators in the performance-CEO turnover relationship (Weisbach, 1988; Parrino et al., 2003; Williams & Livingstone, 1994). A moderator variable weakens or strenghtens the influence of an predictor
variable on a dependent variable (Holmbeck, 1997). In the section below an overview of moderators in performance-turnover studies are given.
Weisbach (1988) examined whether corporate governance structure has an influence on CEO turnover. The general belief is that outsider directors are more likely to carry the duty to evaluate performance and make retention decisions, rather than insider directors. Since, insiders are usually appointed by the CEO, therefore less willing to remove the incumbent CEO. Moreover, outsider directors have incentives to develop reputations as experts in decision control (Weisbach, 1988). Furthermore, Weisbach (1988) provided evidence that management turnover varies with the composition of outsiders and insiders in the board. Thus, good corporate governance strengthens the negative relationship between performance and the likelihood of CEO turnover.
Engel et al., (2003) extends the research by examining the quality of performance measures and their impact on the relationship between performance and turnover. The researchers assume that accounting based measures are more informative about the managerial performance of a CEO. Thereby, accounting-based measures receive greater weight in turnover decisions. The justification for this suggestion is based on the level of accuracy and sensitivity that accounting information has and if market returns are more variable. However, the researchers found mixed support for the hypothesis that firms rely more heavily on market-based measures when accounting information is less timely or market returns are less variable (Engel et al., 2003).
Others (Parrino et al., 2003) looked whether changes in the shareholder composition has an impact on turnover decisions by the board. Institutional investors are often the majority equity holders of the firm. They implicitly show their approval or disapproval for management decisions and performance by selling or buying of firm shares. Parrino et al. (2003) state that the likelihood of turnover increases with greater institutional selling of firm’s stock. Hence, shareholder composition moderates the turnover decisions made by the board.
2.4 CEO reputation as moderator
This paper uses Milbourn’s (2003) definition of reputation, which is defined as the managerial ability demonstrated and accumulated by the CEO. Managerial ability consist of (1) generic-, (2) industry-specific-, (3) firm-specific knowledge, skills, and experience that the CEO accumulates over the years (Holcomb et al., 2009; Zhang, 2008). However, reputation that is developed by a CEO is difficult to transfer across other firms and industries (Zhang, 2008). Therefore, this paper only examines the reputation built during a CEO’s tenure within their current firm.
CEOs have incentives to develop and manage their reputations, because maintaining a certain image can be rewarding or useful (Ranft et al., 2006). According to the Efficient-contracting Theory reputation is build over time with achievements. By presenting good results the CEO sends a message to the market for decision agents that they are adequate. Thus, increasing their value and strengthen their position in the labor market. CEOs with strong reputational capital
are more likely to select projects with a positive net present value that will benefit the firm (Jian & Lee, 2011).
Good performing CEOs tend to have relative long tenures. This can be explained by the assumption that CEOs with well-established reputations are expected to have performed better, and have generated higher firm performance in the past, and posses a higher level of managerial ability. Therefore, these CEOs have survived past retention decisions (Ranft et al., 2006). Thus, long tenure is an indicator of repetitive good performance. Moreover, according to Cianci & Kaplan (2010) reputation is not damaged easily due to poor performance. Their results from experimental research show that the plausibility of management’s explanation following poor performance had no influence on the perceived management’s reputation. However, this is only valid when the management has a pre-existing favorable reputation.
Based on the proposition that long tenure reflects repetitive good performance and favorable reputations are not easily damaged due to poor results, I assume that boards might be tolerant in dismissal decisions against CEOs who have a proven track record prior to poor performance. Boards might anticipate that the CEO will perform better in forthcoming years since the CEO has performed well in the past. Therefore, boards might be tolerant or delay the dismissal for CEO with a favorable reputation. The following hypothesis is formulated to examine the assumption:
H1: CEO reputation has a negative moderating effect on the relationship between poor performance and turnover.
3. Methodology
3.1 Sample selection
The CEO data for the time period 2007 to 2013 is obtained from COMPUSTAT’s ExecuComp. The ExecuComp database covers most of the firms from S&P 1500 Composite index. Subsequently, the ExecuComp database provides a filter whether the executive was CEO during the fiscal year (CEOANN). This filter is used to restrict the number of observations in the sample.
The initial sample consists of 3.262 observations. Stock market- and financial statement data are collected from the merged CRSP/COMPUSTAT database. After merging the data, the sample consists of 2.268 observations. Consequently, the sample is supplemented with CEO duality information derived from RiskMetrics. This process yields a sample of 1.455 observations. Furthermore, financial institutions and public utilities (SIC codes between 4000-4999 and 6000-6999) are eliminated from the sample, because these firms maybe regulated and different from other unregulated companies. This reduces sample to 1.078 observations. Lastly, observations with missing data for any variable for the regression analysis are eliminated. The final sample consists of 970 observations for the time period 2007 to 2013. Table 3.1 specifies the sample selection process.
TABLE 3.1 SAMPLE SELECTION # of observations Initial sample 3.262 Less:
Loss due to merging CRSP and COMPUSTAT 994
Loss due to merging RiskMetrics 813
Filter Utilities and Financial industries 377
Missing variables 108
Total excluded observations -2.292
Final sample 970
3.2 Empirical model
In order to examine the moderating effect of reputation on the relationship between performance and turnover, this paper uses a logistic regression analysis with CEO turnover as a binary dependent variable. The regression equation employed is stated as:
CEO_TURNOVER = α + β1 AccPerf + β2 (AccPerf × CEO_REPU) + β3 STOCK_PF + β4 (MarketPerf × CEO_REPU)
+ β5 CEO_REPU + CONTROLS + ε (1)
The relationship between accounting performance and the likelihood of CEO dismissal for CEOs with low reputation is represented by the coefficient β1, while the sum of coefficients β1 + β2 give the relationship between accounting performance and the likelihood of CEO dismissal for CEOs with a s high reputation. β3 captures the relationship between market performance and the likelihood of CEO dismissal for CEOs with low reputation. The relationship between market performance and the likelihood of CEO dismissal for CEOs with a high reputation is captured in the sum of coefficients β3 + β4.
Based on the hypothesis, I expect the coefficients for both accounting- and market performance to be positive (β2>0 and β4>0). A positive value indicates that reputation weakens the negative relation between performance and dismissal.
3.3 Variable measurement
3.3.1 Dependent variable
CEO turnover is defined by (Barron et al., 2011) as “the departure from the official position of CEO and not necessarily the firm”. This means that the departing CEO can take a new function in the company, such as chairman of the board of directors or other honorable position. This paper follows this definition and classifies when a CEO turnover occurs as one, and zero otherwise.
3.3.2 Independent variables
Several researchers use return-on-asset (ROA) as indicator of accounting performance (Parrino et al., 2003; Jalal & Prezas, 2012). ROA is defined as the ratio earnings-before-interest-and-taxes to total assets. The ratio can be interpreted as how profitable a firm is relative to its total assets. It also indicates how efficient the CEO is by generating profits with assets held by the firm.
Furthermore, stock price performance is used to capture the market performance. Stock price performance reflects a manager’s ability to create firm
value (Engel et al., 2003). Stock price performance is computed by stock price end-year minus stock price begin-end-year devided by stock price begin-end-year.
Regarding CEO reputation prior literature states that is difficult to quantify reputation (Jian & Lee, 2011; Engel et al., 2003). This paper uses CEO tenure to measure the reputation of a CEO. A long CEO tenure implies that CEO has survived previous dismissal decisions due to repetitive good performance. CEO_REPU is a dummy variable, which classifies CEOs with tenures longer or equal to the median of the total sample as one, and takes the value of zero otherwise.
3.3.3 Control variables
This section specifies the control variables used in the logistic regression model. Three variables are used to control for other possible factors influencing CEO turnover. These variables are (1) CEO duality, (2) firm size, and (3) CEO age.
First, this paper uses CEO duality as control variable since powerful CEOs can have an impact on the likelihood of CEO dismissal (Zhang, 2008). Powerful CEOs are able to control the board and influence director appointments. This increases the difficulty of removing a CEO. This paper defines CEO duality as a CEO whom is also chairman of the board. An one is noted when a CEO is also chairman of the board and zero otherwise.
Second, this paper controls for firm size effects on CEO turnover. Several studies found a positive relation between the likelihood of CEO turnover and firm size (Huson, 2001; Parrino 1997). Prior literature suggests that larger firms have higher frequency of management turnover since large firms have promotion and retirement policies that stimulate shorter tenures in top management positions (Warner et al., 1988). Firm size is calculated by using the natural logarithm of the book value of total assets at fiscal year-end
Third, CEO age is expected to impact the likelihood of CEO turnover since turnover also can occur due to planned retirement. This paper addresses the misinterpretation of turnover-due-to-regular-retirement by controlling for CEO age. This paper assumes that the retirement age is equal or older than 60 years. A dummy variable value of one is used to register if the CEO’s age is equal or above 60 years, and zero otherwise.
4. Empirical results
4.1 Descriptive results
Table 4.1 presents the descriptive statistics for all variables, which are used in the logistic model. The table contains the means, standard deviations, the values of 25th and 75th percentile, and the median. Table 4.2 examines differences between the turnover (CEO_TURNOVER=1) and non-turnover group (CEO_TURNOVER=0). Table 4.3 and Table 4.4 provide the Pearson and Spearman correlation coefficients.
TABLE 4.1
DESCRIPTIVE STATISTICS
Variable N Mean Std. Dev. 25% Median 75%
CEO_TURNOVER 970 0.5196 0.4999 0 1 1 ROA 970 0.0761 0.6227 0.2409 0.0658 0.1137 STOCK_PF 970 0.1421 0.6412 -0.1735 0.0629 0.2668 CEO_REPU 970 0.5680 0.4956 0 1 1 DUALEMPL 970 0.5144 0.5000 0 1 1 CEO_AGE 970 0.3753 0.4844 0 0 1 SIZE 970 7.3491 2.3535 5.6644 7.5886 8.8932 Variable definitions: CEO_TURNOVER STOCK_PF ROA CEO_REPU DUALEMPL CEO_AGE SIZE = probability of turnover;
= stock performance = stock price end-begin / stock begin fiscal year
= return on assets = EBIT / total assets = CEO reputation
= CEO and chairman = CEO age
= natural logarithm of total assets
As Table 4.1 shows, about 51.96 percent of the all the CEOs in the sample experience a turnover (median = 1). Furthermore, the mean for the independent variable ROA is 0.0761 (median = 0.0658) and STOCK_PF has a mean of 0.1421 (median = 0.0.0629), respectively the standard deviation amounts 0.6227 and 0.6412. Both ROA and stock performance have large standard deviations, suggesting that the performances are variable amongst observations in the sample. Moreover, the results show that 56.80 percent of the CEOs have built a reputation during their tenure at their current firm. Furthermore, the results report that 51.44 percent of the CEOs have a dual function as chairman of the board. Moreover, 37.53 percent of
the CEOs are younger than 60 years. Further, the mean for the variable SIZE amounts 7.3491 (median = 7.5886).
Table 4.2 shows the differences between the turnover and non-turnover sample. The turnover sample consists of CEO who experienced a turnover (N= 504) and CEO who did not experience a turnover (N=466). Additionally, table reports that the mean of ROA in the turnover sample is 0.0745, which is slightly lower than the mean of 0.0778 reported in the non-turnover sample. This confirms the expectation that turnover is more likely to occur in poor performing firms.
Furthermore, the table shows that the average stock price performance for turnover CEOs is 0.0645, which is lower than the 0.2260 reported by the CEOs from the non-turnover sample as expected. Subsequently, the median of the turnover sample is -0.0105, while the non-turnover sample has 0.1316. The p-value is 0.0001 this means that stock performance is a significant indicator of turnover. Furthermore, both turnover- and non-turnover sample have relatively large standard deviations for stock performance (0.6858 and 0.6652). The large standard deviation suggests a relative large fluctuation in the delivered performance by the CEO. In addition, the large standard deviation may suggest that other factors are involved in turnover decisions since good performers also experience dismissal.
Additionally, Table 4.2 shows that 55.75 percent of the CEOs who experience a turnover have built a reputation, while a similar amount can be found in the non-turnover sample respectively 57.04 percent. The results suggest that non-turnover occurs with roughly same probability regardless of CEO reputation.
With regards to the control variables, Table 4.2 reports that 52.78 percent of the turnover CEOs has a dual position as chairman of the board, versus 50 percent in the non-turnover sample. The results suggest that turnover occurs with roughly same probability regardless of CEO power due CEO duality. Additionally, the turnover sample reports that 45.24 percent of the CEOs are older than 60 years, while in the non-turnover sample 29.18 percent are older than 60 years. This suggests that a portion of the reported turnover consists of regular turnovers due to retirement of the CEO. Further, the values reported for the variable firm size shows that turnover occurs more often in smaller companies (7.2190 versus 7.4899).
TABLE 4.2
DESCRIPTIVE STATISTICS (BY LEVEL OF LIKELIHOOD OF TURNOVER)
Section A Section B Section C
Turnover sample (1) Non-turnover sample (0) Total sample
n = 504 n = 466 n = 970
Variable Mean Median Std. Dev. Mean Median Std. Dev. T-value p-value
ROA 0.0745 0.7172 0.3307 0.0778 0.0603 0.8304 0.0835 0.9335 STOCK_PF 0.0645 -0.0105 0.6858 0.2260 0.1316 0.6652 3.9489 0.0001 CEO_REPU 0.5575 1 0.4972 0.5704 1 0.4942 0.6861 0.4928 DUALEMPL 0.5278 1 0.4997 0.5 0.5 0.5005 -0.8643 0.3877 CEO_AGE 0.4524 0 0.4982 0.2918 0 0.4551 -5.2260 0.0000 SIZE 7.2190 7.5119 2.2820 7.4899 9.2270 2.4232 1.7931 0.0733
Both the Pearson- and Spearman correlation coefficients are analyzed in order to examine the strength and direction of the relationship between the variables. The results of the correlation tests are reported in Table 4.3 and Table 4.4 with significant values reported in bold.
The Pearson correlation coefficient measures the strength of the linear relationships between normally distributed variables. As expected, the results report a weak negative correlation of -0.0027 between ROA and turnover. However, this relationship is not significant, therefore the aforementioned assumption cannot be supported. Furthermore, as expected there is a significant negative correlation between stock performance and turnover (-0.1259).
Furthermore, the table shows a significant positive relationship between reputation and stock performance (0.0747). This could suggest that investors / the market expect that reputable CEOs are more able to yield higher firm performance, which is reflected in a higher stock price. Furthermore, as expected CEO age is significantly positively correlated with turnover since some turnovers may be caused by retirement (0.1657). Additionally, a significant positive association is found between CEO age and reputation (0.2589). This suggests that as a CEO matures they are more likely to be regarded as experts.
Moreover, on the significance level of 10 percent a negative value of -0.0575 is found for the correlation between firm size and turnover. This contradicts with the general belief that turnover is more frequent in large companies, rather than small companies. One explanation could be that smaller firms attract low-to-moderate ability CEOs, while larger firms are more likely to employ, attract and retain high ability CEOs through their selection processes and compensation plans.
Table 4.4 provides the Spearman correlation coefficients. This correlation coefficient assesses the relationship between two variables by using a monotonic function. In overall, the findings from Spearman correlation matrix roughly correspond with the results from the Pearson matrix. However, one notable difference is that ROA has a significant positive correlation of 0.1075.
TABLE 4.3
PEARSON CORRELATION
CEO_TURNOVER ROA STOCK_PF CEO_REPU DUALEMPL CEO_AGE SIZE
CEO_TURNOVER 1.0000 ROA -0.0027 1.0000 STOCK_PF a -0.1259 0.0287 1.0000 CEO_REPU -0.0220 0.0358 a 0.0747 1.0000 DUALEMPL 0.0278 0.0050 -0.0640 -0.0310 1.0000 CEO_AGE a 0.1657 a -0.0536 -0.0167 a 0.2589 -0.0139 1.0000 SIZE a -0.0575 -0.0101 a -0.0571 -0.0244 0.0179 -0.0228 1.0000
*bold indicates significance at 5% level A indicates significance at 10% level
TABLE 4.4
SPEARMAN CORRELATION
CEO_TURNOVER ROA STOCK_PF CEO_REPU DUALEMPL CEO_AGE SIZE
CEO_TURNOVER 1.0000 ROA a 0.1075 1.0000 STOCK_PF a -0.1932 a 0.1459 1.0000 CEO_REPU -0.0220 -0.0183 0.0460 1.0000 DUALEMPL 0.0278 -0.0245 -0.0321 -0.0310 1.0000 CEO_AGE a 0.1657 0.0085 -0.0154 a 0.2589 -0.0139 1.0000 SIZE a -0.0535 a 0.0787 0.0402 -0.0356 0.0208 -0.0200 1.0000
*bold indicates significance at 5% level a indicates significance at 10% level
4.2 Main findings
A logistic regression is conducted to test whether reputation insulates the CEO from dismissal following poor performance. Table 4.5 reports the results of the logistic regression model. One noteworthy event is that Stata omits eight observations in model 3.
The output in column 1 shows that the accounting based measure is positive associated with turnover (0.5522), which contradicts with results from prior literature (Engel et al., 2003; Farrell & Whidbee, 2003). Prior literature reports a significant negative association between accounting based performance measures. The positive association intensifies when industry and year dummies are taken into account (0.6796 and 0.7434).. Further, in model 3 ROA has a significant coefficient of 0.7434. Consequently, the table shows that the interaction between accounting-based measures and reputation has a negative value and not significant for all models (-0.6351; 10.7188; -0.8275), which suggests a positive moderating effect of reputation on accounting-based measures and turnover. This contradicts the expectation previous made in this paper (β2>0, see chapter 3.2).
Column 1 reports that market-based performance has a significant negative association with turnover as expected (-0.6074). This means that poor stock performance increases the probability of CEOs losing their jobs. However, the significance wears down as industry and year dummies are added to the logistic regression (p-value = 0.006; 0.342; 0.370). Moreover, the results can be interpreted as that low reputation CEOs are more likely to face turnover due to poor market-performance. Additionally, as expected the interaction between market-based performance and reputation has in all models a positive value (0.2178; 0.4848; 0.5373), which indicates a negative moderating effect of reputation on market-based measures and turnover. This suggests that CEOs with high reputational status are more likely to receive tolerant verdicts in retention decisions based on stock performance.
Moreover, the initial model (1) presents a negative association, but not a significant coefficient between reputation and turnover (-0.2428). While, the coefficients in the models adjusted with year and industry dummies are significant (-0.4820 and -0.4149). This could suggest (1) that in certain industries CEOs with reputation are more able to affect or receive tolerant verdicts in turnover decisions. It also suggests (2) that in certain years, with special conditions i.e. poor overall performance of the whole market, reputable CEOs are less likely to be dismissed.
With regard to CEO duality, there is no significant relationship between CEOs with chairman positions and turnover (p-value = 0.402; 0.250; 0.062). Regarding the control variable CEO age, it has a significant positive relationship with turnover as expected. CEOs who are older than 60 years are more likely to experience a turnover possibly due to retirement. Furthermore, A significant negative relationship between firm size and turnover is found in model 1 (-0.0618). This suggests that CEOs in larger firms face less turnovers than small firms, which contradicts with the general belief that CEOs from larger firms are more frequent dismissed. One explanation could be that smaller firms attract low-to-moderate
ability CEOs, while larger firms are more likely to employ, attract and retain high ability CEOs through their selection process and compensation plans. However, this does not hold for model 2.
The pseudo R2 values of the models are 0.0416, 0.3721, and 0.4036
respectively. The chi2 shows that all three models are statistically significant
TABLE 4.5 LOGISTIC REGRESSION
Model (1) Model (2) Model (3)
Variable Coefficient t-statistic (p-value) Coefficient t-statistic (p-value) Coefficient t-statistic (p-value)
ROA 0.5522 0.133 0.6796 0.054 0.7434 0.044 (ROA*CEO_REPU) -0.6351 0.106 -0.7188 0.089 -0.8275 0.077 STOCK_PF -0.6074 0.006 -0.2459 0.342 -0.2423 0.370 (STOCK_PF*CEO_REPU) 0.2178 0.407 0.4848 0.103 0.5373 0.088 CEO_REPU -0.2428 0.092 -0.4820 0.012 -0.4149 0.041 DUALEMPL 0.1113 0.402 0.2036 0.250 0.3572 0.062 CEO_AGE 0.7800 0.000 0.9639 0.000 0.9973 0.000 SIZE -0.0618 0.031 0.0094 0.805 -0.0025 0.951 *bold indicates
significance at 5% level Year Dummies: No Industry Dummies: No N = 970
Prob > chi2 = 0.0000 Pseudo R2 = 0.0416
Year Dummies: Yes Industry Dummies: No N = 970
Prob > chi2 = 0.0000 Pseudo R2 = 0.3721
Year Dummies: Yes Industry Dummies: Yes N = 962
Prob > chi2 = 0.0000 Pseudo R2 = 0.4036
4.3 Supplemental analysis
In Table 4.6 both firm accounting- and market performance measures are altered into industry-adjusted variables. In which performance measures are adjusted by subtracting the mean value of each corresponding two-digit Standard Industrial Code (SIC) code.
Surprisingly, Table 4.6 shows that both the industry adjusted ROA and stock performance are both significant positive associated with turnover on 10 percent level (0.6132 and 0.1443). This contradicts with prior literature, which report a significant negative relationship between performance and turnover. Consequently, both interactions between firm performance and reputation are negatively associated with turnover. This contradicts with the expectation made in chapter 3.2 (β2>0 and β4>0).
Furthermore, CEO reputation has significantly negative relationship with turnover (-0.3113). As earlier reported in chapter 4.2 CEO duality and CEO age have a positive relationship with turnover. Also, firm size corresponds with results from chapter 4.2.
TABLE 4.6 LOGISTIC REGRESSION
INDUSTRY ADJUSTED PERFORMANCE
Variable Coefficient t-statistic (p-value)
ADJ_ROA 0.6132 0.097 (ADJ_ROA*CEO_REPU) -0.6680 0.090 ADJ_STOCK_PF 0.1443 0.507 (ADJ_STOCK_PF*CEO_REPU) -0.1937 0.465 CEO_REPU -0.3113 0.025 DUALEMPL 0.1357 0.304 CEO_AGE 0.7797 0.000 SIZE -0.0577 0.042
*bold indicates significance at 5% level
N = 970
Prob > chi2 = 0.0000 Pseudo R2 = 0.0304
5. Conclusion
Prior literature mainly focuses on the relationship performance measures and CEO turnover. Research reports a significant negative correlation between performance and turnover. Additionally, prior literature looked into various determinants regarding this relationship, such as corporate governance, shareholder compensation, etc. This study contributes to existing literature by examining whether CEO reputation has a negative moderating effect on the relationship between poor performance and turnover by conducting archival-based research.
Using a sample of 970 observations for the time period 2007 to 2013, I found no significant results that could support the hypothesis on the 5 percent confidence level. However, the empirical results show mixed results for the hypothesis on 10 percent confidence level, when industry- and year effects were taken into account. In which, I found a negative moderating effect of reputation on market-based measures and turnover. This negative relationship could not be found for accounting-based measures. Instead, I found a positive moderating effect of reputation on market-based measures and turnover.
There are some limitations of this paper. One limitation is that the method used to measure the reputation of a CEO is relatively simple as it is difficult to quantify reputation. Future studies may find better methods or refine methods in order to quantify reputation or measure managerial ability. An example for another determinant could be analyst forecasts or Demerjian et al. (2006) method to measure managerial ability.
Bibliography
Allgood, S., & Farell, K. A. (2000). The effect of CEO tenure on the relation between firm performance and turnover. Journal of Financial Research , 23 (3), 373-390. Barron, J. M., Chulkov, D. V., & Waddell, G. R. (2011). Top mangement team turnover, CEO succession type, and strategic change. Journal of Business Research , 64, 904-910.
Bushman, R., Dai, Z., & Wang, X. (2010). Risk and CEO turnover. Journal of Financial Economics , 96, 381-398.
Cianci, A. M., & Kaplan, S. E. (2010). The effect of CEO reputation and explanations for poor performance on investors' judgements about the company's future
performance and management. Accounting, Organizatinos and Society (35), 478-495. Clarke, R. N. (1989). SIC as delineators of economic markets. Journal od Business , 62, 17-31.
Dahlstrom, R., & Ingram, R. (2003). Social networks and the adverse selection problem in agency relationships. Journal of Business Research , 56, 767-775.
Demzetz, H. (1983). The structure of ownership and the theory of the firm. Journal of Law & Economics , 26, 375-390.
Eisenhart, K. M. (1989). Agency Theory: An assessement and review. Academy of Management Review , 14 (1), 57-74.
Engel, E., Hayes, R. M., & Wang, X. (2003). CEO turnover and properties of accounting information. Journal of Accounting and Economics (36), 197-226. Fama, E. F., & Jensen, C. M. (1983). Separation of ownership and control. Journal of Law and Economics , 26 (2), 301-325.
Farrell, K. A., & Whidbee, D. A. (2003). Impact of firm performance expecations on CEO turnover and replacement decisions. Journal of Accounting and Economics (36), 165-196.
Harris, M., & Holmstrom, B. (1982). A theory of wage dynamics. Review of Economic Studies , 49, 315-333.
Holcomb, T. R., Holmes, R. M., & Connelly, B. L. (2009). Making the most of what you have: Managerial ability as source of resource value creation. Strategic Management Journal , 30, 457-485.
Holmbeck, G. N. (1997). Toward terminological, conceptual, and statistical clarity in the study of mediators and moderators: Examples from the child-clinical and
pediatric psychology literatures. Journal of Consulting and Clinical Psychology , 4, 599-610.
Huson, M. R., Malatesta, P. H., & Parrino, R. (2004). Managerial succession and firm performance. Journal of Financial Economics , 74, 237-275.
Huson, M. R., Parrino, R., & Starks, L. T. (2001). Internal monitoring mechanisms and CEO turnover: A long-term perspective. Journal of Finance , 56 (6), 2265-2297. Jalal, A. M., & Prezas, A. P. (2012). Outsider CEO succession and firm performance. Journal of Economics and Business , 64, 399-426.
Jenter, D., & Kanaan, F. (2006). CEO turnover and relative perfomance evaluation. NBER Working Paper Series , 1-49.
Jian, M., & Lee, K. (2011). Does CEO reputation matter for capital investments? Journal of Corporate Finance (17), 929-946.
Kaplan, S. N., & Minton, B. A. (2006). How has CEO turnover changed? increasingly performance sensitive boards and increasingly uneasy CEOs. NBER Working Paper Series , 1-33.
Milbourn, T. (2003). CEO reputation and stock-based compensation. Journal of Financial Economics , 68, 233-262.
Parrino, R. (1997). CEO turnover and outside succession: A cross-sectional analysis. Journal of FInancial Economics , 46, 165-197.
Parrino, R., Sias, R. W., & Starks, L. T. (2003). Voting with their feet: institutiional ownership changes around CEO turnover. Journal of Finance , 68, 3-46.
Perryman, A. A., Butler, F. C., Martin, J. A., & Ferris, G. R. (2010). When the CEO is ill: Keeping quiet or going public? Business Horizons , 53, 21-29.
Ranft, A. L., Ferris, G. R., Zinko, R., & Buckley, M. R. (2006). Marketing the image of management: The cost and benefits of CEO reputation. Organizational Dynamics , 35 (3), 279-290.
Warner, J. B., Watts, R. L., & 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 (20), 431-460.
Williams, C. R., & Livingstone, L. P. (1994). Another look at the relationship between performance and voluntary turnover. The Academy of Management Journal , 37, 269-298.
Zajac, E. J. (1990). CEO selection, succession, compensation and firm performance: A theoretical intergration and empirical analysis. Strategic Management Journal , 11, 217-230.
Zajac, E. J., & Westphal, J. D. (1996). Director reputation, CEO-board power, and the dynamics of board interlocks. Administrative Science Quarterly , 41 (3), 507-529. Zhang, Y. (2008). Information asymmetry and the dismissal of newly appointed CEOs: An empirical investigation. Strategic Management Journal , 29, 859-872.
Appendix I: List of variables
Variable
Description
CEO_TURNOVER A binary variable equal to 1 if a turnover occurs and 0 otherwise; ROA The ratio of earnings-before-interest-and-tax to total assets;
ADJ_ROA The ratio of earnings-before-interest-and-tax to total assets
adjusted by subtracting the mean value of ROA for every two-digit SIC code.;
STOCK_PF Stock price at fiscal year-end minus stock price at beginning of
fiscal year, divided by stock price at beginning of fiscal year;
ADJ_STOCK_PF Stock price at fiscal year-end minus stock price at beginning of
fiscal year, divided by stock price at beginning of fiscal year adjusted by subtracting the mean value of stock performance for every two-digit SIC code;
CEO_REPU A dummy variable equal to 1 if the CEO has a tenure of than the
median (6 years) or more and 0 otherwise;
DUALEMPL A dummy variable equal to 1 if the CEO is also chairman and 0
otherwise;
SIZE The natural log of total assets;
CEO_AGE A dummy variable equal to 1 if the CEO’s age is 60 or above at
Appendix II: DO-File
/*GENERATE CEO DUALITY VARIABLE*/ use "/Users/Will/Desktop/B/RISK.dta" encode cusip, gen(cusip1)
drop cusip
rename cusip1 cusip
encode employment_ceo, gen(employment_ceo1)
encode employment_chairman, gen(employment_chairman1) drop employment_ceo
drop employment_chairman
rename employment_ceo1 employment_ceo
rename employment_chairman1 employment_chairman
gen dualempl=1 if (employment_chairman==1 & employment_ceo==1) replace dualempl=0 if (employment_chairman==. & employment_ceo==1) label var dualempl "DUAL employment, CEO & CHAIR"
drop if dualempl==. sort cusip year
duplicates drop cusip year, force xtset cusip year
sort cusip year
save "/Users/Will/Desktop/B/RISK.dta", replace clear
/*GENERATE COMPUSTAT & CRSP VARIABLE*/ use "/Users/Will/Desktop/B/CRSP.dta"
encode gvkey, gen (gvkey1) drop gvkey
rename gvkey1 gvkey rename fyear year sort gvkey year duplicates report duplicates drop xtset gvkey year
gen stock_pf= (prcc_f - prcc_f[_n-1]) / prcc_f[_n-1] if gvkey==gvkey[_n-1] label var stock_pf "Stock performance"
gen ROA= ebit / at
label var ROA "earnings before tax and interest to total assets"
save "/Users/Will/Desktop/B/CRSP.dta", replace clear
/*GENERATE EXECU VARIABLE*/ use "/Users/Will/Desktop/B/EXECU.dta" gen tenure= (year(leftofc))-(year(becameceo)) label variable tenure "CEO tenure in years" encode gvkey, gen (gvkey1)
drop gvkey
rename gvkey1 gvkey
duplicates drop gvkey year, force xtset gvkey year
generate ceo_turnover = 0 if exec_fullname==exec_fullname[_n+1] & gvkey==gvkey[_n+1]
replace ceo_turnover = 1 if exec_fullname!=exec_fullname[_n+1] & gvkey==gvkey[_n+1]
replace ceo_turnover=0 if tenure==. & ceo_turnover==. replace ceo_turnover=1 if tenure!=. & ceo_turnover==. label var ceo_turnover "Turnovers, 0 no turnover, 1 turnover" sort gvkey year
drop if exec_fullname==exec_fullname[_n+1] & gvkey==gvkey[_n+1] save "/Users/Will/Desktop/B/EXECU.dta", replace
clear /*MERGE*/
use "/Users/Will/Desktop/B/EXECU.dta"
merge gvkey year using /Users/Will/Desktop/B/CRSP.dta keep if _merge==3
drop _merge
encode cusip, gen (cusip1) drop cusip
rename cusip1 cusip xtset cusip year sort cusip year
merge cusip year using /Users/Will/Desktop/B/RISK.dta keep if _merge==3
tostring sic, replace gen sic1d=substr(sic,1,1)
label var sic1d "one-digit SIC code" destring sic1d, replace
drop if sic1d==6 drop _merge drop if stock_pf==. drop if ROA==. drop if tenure<0 drop if becameceo==.
save "/Users/Will/Desktop/B/EXECU.dta", replace clear
/*GENERATE UNKOWN TENURE*/ use "/Users/Will/Desktop/B/EXECU.dta" gen tenure2=tenure
replace tenure2=(2013-(year(becameceo))) if tenure==. sum tenure2, d
label var tenure2 "All observations tenure as per 2013" gen ceo_repu=0 if tenure2 <6
replace ceo_repu=1 if tenure2 >=6
label var ceo_repu "reputation, 0<median, 1>=median" /*GENERATE CONTROL VAR*/
gen ceo_age=1 if age>=60 replace ceo_age=0 if age<60 label var ceo_age "1>60, 0<60" gen size=ln(at)
label var size "firmsize, natural log of total assets" /*GENERATE INTERACTION VARIABLE*/ gen b2=(ceo_repu * ROA)
label var b2 "interaction var high ability CEO ROA" gen b4=(ceo_repu * stock_pf)
label var b4 "interaction var high ability CEO Stock performance" save "/Users/Will/Desktop/B/EXECU.dta", replace
/*GENERATE ADJ PERFORMANCE*/ gen sic2d=substr(sic,1,2)
label var sic2d "two-digit SIC code" destring sic2d, replace
by sic2d year, sort: egen ROA_M = mean(ROA) gen adj_ROA = ROA - ROA_M
label var ROA_M "Mean ROA by sic2d year"
label var adj_ROA "Adj. ROA, ROA-ROA mean by sic2d year"
by sic2d year, sort: egen stock_pf_M = mean(stock_pf) gen adj_stock_pf = stock_pf - stock_pf_M
label var stock_pf_M "Mean stock performance by sic2d year"
label var adj_stock_pf "Adj. stock performance, stock - stock mean by sic2d year"
gen adj_b2=(ceo_repu * adj_ROA)
label var adj_b2 "interaction var high ability CEO adj. ROA" gen adj_b4=(ceo_repu * adj_stock_pf)
label var adj_b4 "interaction var high ability CEO adj. Stock performance" /*GENERATE ADDITIONAL CONTROLS*/
tabulate year, gen(year_eff) tabulate sic2d, gen(indu_eff) /*DESCRIPTIVE STAT*/
tabstat ceo_turnover ROA stock_pf ceo_repu dualempl ceo_age size , stats(N mean sd p25 median p75) columns(statistics)
tabstat ROA stock_pf ceo_repu dualempl ceo_age size if ceo_turnover==1, stats(N mean sd p25 median p75) columns(statistics)
tabstat ROA stock_pf ceo_repu dualempl ceo_age size if ceo_turnover==0, stats(N mean sd p25 median p75) columns(statistics)
pwcorr ceo_turnover ROA stock_pf ceo_repu dualempl ceo_age size, sig spearman ceo_turnover ROA stock_pf ceo_repu dualempl ceo_age size, stats(rho p)
ttest stock_pf, by(ceo_turnover) ttest ROA, by(ceo_turnover) ttest ceo_repu, by(ceo_turnover) ttest dualempl, by(ceo_turnover) ttest ceo_age, by(ceo_turnover) ttest size, by(ceo_turnover)
/*LOGIT REGRESSION*/
logit ceo_turnover ROA stock_pf ceo_repu dualempl ceo_age size logit ceo_turnover ROA b2 stock_pf b4 ceo_repu dualempl ceo_age size /*LOGIT ADDITIONAL CONTROLS*/
logit ceo_turnover ROA b2 stock_pf b4 ceo_repu dualempl ceo_age size year_eff2 year_eff3 year_eff4 year_eff5 year_eff6 year_eff7
logit ceo_turnover ROA b2 stock_pf b4 ceo_repu dualempl ceo_age size year_eff2 year_eff3 year_eff4 year_eff5 year_eff6 year_eff7 indu_eff1
indu_eff2 indu_eff3 indu_eff4 indu_eff5 indu_eff6 indu_eff7 indu_eff8 indu_eff9 indu_eff10 indu_eff11 indu_eff12 indu_eff13 indu_eff14 indu_eff15 indu_eff16 indu_eff17 indu_eff18 indu_eff19 indu_eff20 indu_eff21
indu_eff22 indu_eff23 indu_eff24 indu_eff25 indu_eff26 indu_eff27 indu_eff28 indu_eff29 indu_eff30 indu_eff31 indu_eff32 indu_eff33 indu_eff34 indu_eff35 indu_eff36 indu_eff37 indu_eff38 indu_eff39 indu_eff40 indu_eff41 indu_eff42 indu_eff43 indu_eff44 indu_eff45 indu_eff46 indu_eff47 indu_eff48
/*LOGIT SUPPLEMENT ANALYSIS*/
logit ceo_turnover adj_ROA adj_b2 adj_stock_pf adj_b4 ceo_repu dualempl ceo_age size
