Firm performance and CEO turnover Does CEO attributes

Firm performance and CEO turnover: Does CEO attributes matter? Dr Marian Chijoke-Mgbame Prof. Agyenim Boateng Dr. Oscar Mgbame Prof. Kemi Yekini

Outline v. Introduction v. Motivation of the study v. Objective of the study v. Brief review of literature v. Hypotheses v. Data and sample v. Variables definition v. Model specification and Method v. Results v. Findings v. Conclusion

Introduction • The decision to hire and fire a CEO is one of the major decisions made solely by the board of directors of a company. • However, the decision to dismiss a CEO is a complex one as there are consequences to the firm. • Prior studies (Conyon and Florou, 2002; Fiordelisi et al. , 2014; Chen et al. , 2019) show that the decision to fire a CEO is based on the firm performance. • Pitcher et al. , (2000) note that performance accounts for only a small proportion of the CEO turnover decision.

Introduction • There is a consensus in the literature that CEOs are dismissed for poor performance • However, studies have shown that performance is not the only determinant of the turnover decision. • For example, Brunello et al. , (2003); Goyal and Park (2002); Fiordelisi and Ricci (2014) show that the ownership structure, leadership structure and the culture of the firm determines the sensitivity of CEO turnover to firm performance. • We depart from the above studies that have focused on firm specific attributes as they affect CEO turnover and emphasise that the CEO attributes may affect the turnover -performance relationship

Motivation of the study • Our study derives motivation from the upper echelon theory, (Hambrick and Mason, 1984) which suggest that the strategies and outcomes of organisations are often shaped by the demographic characteristics (such as age, tenure, education and experience) of the executives. • Given the above theoretical stance, it therefore means that the board may consider the demographic characteristics of the CEO in deciding whether to fire or retain the CEO. • Our proposition is that the sensitivity of turnover to performance maybe dependent on some of the CEO personal attributes.

Objective of the study • The objective of the study then, is; To examine the effect of specific CEO demographic characteristics on the relationship between CEO turnover and firm performance We also examine whether board independence affects the above relationship.

Brief review of literature • To be a credible disciplinary measure, conventional wisdom suggests that poorly performing CEOs be replaced. • Theoretical and empirical studies confirm a negative relationship between firm performance and CEO turnover. • For example, Coughlan and Schmid (1985), Warner et al. , (1998) provide evidence of a negative relationship between CEO change and stock price performance in the US. • Similarly, Puffer and Weintrop (1991) find that for a sample of 408 CEOs under the age of retirement, turnover occurs when there is a fall in the expected annual earnings per share. • In the UK, Dahaya et al (1998, 2002) provide evidence of a negative relationship between previous year stock performance and CEO turnover.

Brief review of literature • Although prior studies provide evidence of a negative relationship between CEO turnover and performance, the decision to replace a CEO is a complex one which goes beyond firm performance. • The complexities involved in firing a CEO partly stems from; the uncertainties about the abilities of a new CEO especially one that is hired from outside the firm. the learning phase for both the board and the new CEO which may not be appropriate in the case of poor firm performance. negative publicity which may further impact firm performance • Given the above, it is therefore possible that the decision to fire the CEO may be influenced by factors other than performance. • We explain the rationale for our idea using a multi-theoretical framework

Brief review of literature Theories • Upper Echelon Theory (Hambrick and Mason, 1984): suggest that the strategies and outcomes of organisations are often shaped by the demographic characteristics of the executives. The board may also consider such characteristics when a turnover decision is to be made.

Brief review of literature Theories • Human and Social Capital Theory (Becker, 1975; Nahapiet and Ghoshal, 1998; Adler and Kwon, 2002 ): the market for CEOs is highly competitive consisting usually of individuals with varying experience and attributes. While the strategy or the performance of the firm may affect the turnover decision, the human and social capital that the CEO possesses may also play a significant role in the decision. Thus, the human and social capital can be used for the benefit of the individual and the firm.

Brief review of literature Theories • The Resource Dependency Theory (Pfeffer and Salancik, 1978) argues that a firm’s survival and competitive advantage depends on the resources available to the firm. These resources can be derived internally and externally. The experience and networks that the CEO has is an important resource to the firm.

Brief review of literature • Using the aforementioned theories, we specifically examine three key CEO attributes; CEO internal experience, CEO network size and CEO age. • These three attributes are chosen for several reasons. Firstly, they are easily identifiable and objectively measured. Secondly, these attributes are not conferred on the CEO, but have been acquired and developed over the years. Thirdly, they are not transferable rather, they are unique to the CEO.

Hypotheses • H 1 a: There is a negative relationship between the CEO’s internal experience and CEO turnover. H 1 b: The CEO’s internal experience negatively moderates the performance turnover relationship. • H 2 a: CEOs with larger network size are less likely to be dismissed H 2 b: CEO network size reduces the probability that a CEO will be dismissed for poor performance • H 3 a: There is a positive relationship between CEO age and CEO turnover. H 3 b: CEO age increases the probability that a CEO will be dismissed for poor performance.

Data and sample • Data source: We obtain data for CEO attributes and corporate governance for the period 1999 through 2018 from Board. Ex. Firm financial data were obtained from Thomson Reuters Eikon. • The data covers all companies from the FTSE 350 index with the exception of financial companies. • The final sample comprises 2575 firm-year observations after deleting financial firms and firms with insufficient data. •

Variables Definition • Dependent variable: the dependent variable for the study is CEO • Independent Variables: turnover. We follow prior studies such (Fiordelisi and Ricci et al. , 2014, Peasnell et al. , 2005) to measure CEO turnover (CEOT) as a dummy variable taking the value of 1 if there is a change in CEO with respect to the previous year and 0 otherwise. Firm performance: to measure firm performance, we use the return on asset (ROA). CEO attributes: CEO internal experience is measured as the number of years a CEO has spent in the company (Brockman et al. , 2019). CEO network size is measured as the number of boards the CEO has served on and is currently serving on, Yang et al. , (2011). CEO age is measured in years • Control Variables: board size, board independence, firm size, leverage, capital expenditure, firm cash, CEO tenure, CEO gender, CEO cash compensation and CEO equity compensation

Model specification and Method • To examine the impact of CEO attributes on the turnover decision, we estimate the following probit regression model: • Pr (CEOT)���� = �� 01(firm + �� performance) + �� 2(CEO INTEXP )+ �� 3(CEO Network) + �� 4(CEO age)+ �� 5(Control variables ) + ɛ���� …(1) • Pr (CEO turnover)��, �� = �� 0 + �� 1(firm performance ) + INTEXP )+ �� 3(CEO Network) + �� 4(CEO age)+ �� 5(CEO INTEXP*ROA ) + �� 6(CEO Network*ROA )+ �� 7(CEO Age*ROA )+ �� 8(Control variables ) + ɛ����…(2)

Results Table 1 The effect of performance and CEO attributes on CEO turnover Panel A W/O CEO attributes Panel B W/O Interaction Panel C With Interaction Model 1 Marginal Effect -0. 0267** (0. 0111) Model 2 Marginal Effect (0. 0108) -0. 1169*** (0. 0089) -0. 0098** (0. 0050) 0. 2266*** (0. 0702) -0. 0434* (0. 0250) -0. 0570*** (0. 0095) 0. 0201* (0. 0114) 0. 1741** 0. 0741 -0. 4939*** (0. 0618) 0. 0252** 0. 0107 0. 0153*** (0. 0050) Board Size 0. 0683 * 0. 0300 0. 0060 (0. 0365) (0. 0282) (0. 0314) Board. Ind 0. 0103 0. 0525** 0. 0428* (0. 0288) (0. 0244) (0. 0253) Firm Size 0. 0020 0. 0328*** 0. 0202** (0. 0082) (0. 0085) (0. 0094) Leverage -0. 0063 -0. 0059 -0. 0051 (0. 0077) 0. 0061 (0. 0070) Cap. Ex 0. 0011 0. 0008 0. 0013 0. 0047 (0. 0011) Firm Cash -0. 0021 -0. 0015 -0. 0036 (0. 0058) (0. 0051) CEO Tenure 0. 0422 *** 0. 0931*** 0. 0275*** 0. 0076 (0. 0128) (0. 0085) CEO Gender 0. 0722 0. 0124 0. 0032 (0. 0487) (0. 0207) (0. 0240) CEO Cash. Comp -0. 0120 -0. 0434*** -0. 0283* (0. 0162) (0. 0163) (0. 0149) CEO Equity. Comp -0. 0196 * 0. 0165 -0. 0176* 0. 0111 (0. 0112) (0. 0105) Year effect YES YES Industry Effect YES YES Observations 2172 1861 Pseudo R-Square 0. 0731 0. 4211 0. 3159 ROA CEO INTEXP CEO Network CEO Age CEO INTEXP x ROA CEO Network x ROA CEO Age x ROA -0. 0242** Model 3 Marginal Effect This table reports the results for the probit regression of the impact of firm performance in Panel A, CEO attributes in Panel B and CEO attributes interaction with firm performance in Panel C on CEO turnover. All variables are defined

Summary of Results • There is a negative association between firm performance CEO. This suggest that the likelihood of CEO turnover increases as performance declines. • There is a significant negative effect of CEO internal experience on turnover. The results indicate that a one standard deviation increase in CEO internal experience implies a 12% decrease in turnover probability. The significant negative result supports hypothesis 1 a and indicates that CEOs with more internal experience are less likely to be dismissed. • There is a negative effect of CEO network size on the probability of CEO turnover which supports hypothesis 2 a. A one standard deviation marginal increase in CEO network size imply a 0. 10% decrease in turnover probability. # This suggests that CEOs with larger network size are less likely to be dismissed. • There is a significant positive effect of CEO age on turnover probability. The results suggest that the likelihood of turnover increases with age. The result supports hypothesis 3 a.

Summary of Results • Panel C of Table 1, explores the sensitivity of the turnoverperformance relationship to CEO attributes. • The interaction between CEOINTEXP x ROA is negative and statistically significant. This suggests that the likelihood of CEO turnover as a result of poor performance is lower for CEOs with greater internal experience. The result provides support for hypothesis 1 b. • The interaction term CEONetwork x ROA is positive and statistically significant. This means that the sensitivity of turnover to performance is higher for CEOs with larger networks. The results fail to support hypothesis 2 b. One interpretation of this result is that CEOs with larger network size are likely to leave poorly performing firm because of the connections they have built over the years. • The interaction between CEOAge x ROA, reveals statistically positive results. The result indicates that the sensitivity of turnover to performance is higher for older CEOs. We, therefore, find support for hypothesis 3 b.

Results Table 2 Effect of CEO attributes on CEO turnover for the split sample ROA CEOINTEXP CEO Network CEO Age CEOINTEXP x ROA CEO Network x ROA CEO Age x ROA Board Size Firm Size Leverage Cap Ex Firm Cash CEO Tenure CEO Gender CEO Cash Comp CEO Equity. Comp Year effect Industry Effect Observations Pseudo R-Square Independent Boards Marginal Effect -0. 0377* (0. 0217) -0. 0531*** (0. 0127) 0. 0033 (0. 0124) 0. 1179* (0. 0714) -0. 5960*** (0. 0710) 0. 0081 (0. 0114) 0. 0170*** (0. 0036) 0. 0010 (0. 0439) 0. 0183* (0. 0110) 0. 0147 (0. 0112) 0. 0010 (0. 0016) -0. 0045 (0. 0076) 0. 0414*** (0. 0134) 0. 0252 (0. 0283) -0. 0239 (0. 0206) -0. 0082 (0. 0226) YES 1319 0. 4038 Dependent Boards Marginal Effect -0. 3591*** (0. 0991) -0. 0586* (0. 0229) -0. 0528* 0. 0270 0. 2504* 0. 1420 -0. 5130 *** (0. 1145) 0. 0774*** (0. 0279) 0. 0581*** (0. 0119) 0. 0954 (0. 0969) 0. 0003 (0. 0228) 0. 0265* (0. 0151) 0. 0093*** (0. 0030) -0. 0014 0. 0114 0. 0200 (0. 0158) 0. 0206 (0. 0200) -0. 0958** 0. 0418 -0. 0370* 0. 0214 YES 542 0. 4286

Summary of Results • Next, we examine if the results obtained earlier remain the same when we consider the independence of the board. • We split our sample by the median point. Boards with less than 50% independent directors are classified as dependent boards and those at 50% and above are classified as independent boards. • We observe that CEO network and the interaction variable (ROA x CEO network) are not significant for independent boards. The results suggest that for independent boards, the size of the CEO network does not matter when it decides to fire a CEO. We can attribute the results to the effectiveness of independent directors on the board. • On the contrary but similar to earlier results in table 1, the results obtained for dependent boards are positive and statistically significant. This may point to the lack of board effectiveness for dependent boards. • The results for CEO internal experience and CEO age remain qualitatively the same as those obtained in table 1 for both dependent and independent boards. We, therefore, conclude that CEO internal experience and age matter when deciding to replace a CEO.

Conclusion • CEO demographic attributes affects the turnover decision • Specifically, CEO internal experience and age reduces the likelihood of dismissal in the case of poor firm performance. • The results remain qualitatively the same when we consider the independence of the board.