CEO Succession in Large US Listed Banks: Does Prior Leadership Matter?

Transcription

1 CEO Succession in Large US Listed Banks: Does Prior Leadership Matter? Wanwan Zhu, University of Leeds May 2018 Abstract The selection of a new CEO is an important decision for organizations. Based on a unique hand-built dataset of CEO succession events in large US commercial banks, the study finds that CEO s prior leadership experience improves long-term bank performance. We distinguish the leadership experience according to where the leadership is obtained: leadership gained inside the banking group and leadership gained outside the group. The results suggest that the performance effect is mainly driven by the leadership experience that is gained from outside the bank. In addition, the positive relation between prior leadership experience and bank performance is mitigated by the diversification level of banks. The higher level of diversification, the less degree of profitability improvement. Our analysis proves that the improvement of bank profitability is not due to the new CEO s risk-taking behavior. Instead, we find a decrease in bank risk after CEO appointment. JEL classification: G21, G34, J24 Keywords: CEO Succession, Prior Leadership Experience, Bank Performance, Bank Risk, Diversification 1

2 1. Introduction Selecting a CEO is one of the most important hiring events in organizations. Over the past decade, there has been an important new trend in CEO succession with companies increasingly hiring executives with experience as former CEOs to the CEO position (Murphy and Zabojnik 2007b; Karlsson and Neilson 2009). This trend may be driven by organizations being unwilling to take the risk of appointing individuals with no previous leadership-specific experience (Charan 2005). Banks are more complex institutions and require employees with special skills (Philippon and Reshef 2009), thus selecting the right CEO could give banks a significant competitive edge. This study focus on newly appointed CEOs of large US banks. The purpose of this study is to understand whether different forms of prior leadership experience of the new CEO are associated with the change in bank performance and risk-taking behavior in the postsuccession period. The study also examines whether the influence of CEO s prior leadership experience on profitability and risk depends on the complexity of banks business model, measured by the degree of income diversification. We document these effects using a unique hand-built dataset that captures the information of 144 CEO succession events in large listed US banks from 1993 to Critical to our analysis is the measurement of different forms of leadership experience that might characterize the professional profile of the newly appointed CEO. To this purpose, we measure the new CEO Prior Leadership by the logarithm of total number of years he/she worked as the top CEO of a company, subsidiary or market division before the appointment. While previous studies suggest that the context in which prior CEO experience was gained affects the performance consequence (Bragaw and Misangyi 2017), we argue that the place where the leadership experience is obtained matters. Accordingly, we further 2

3 distinguish the leadership experience into two types based on which organization the leadership experience is obtained: leadership gained inside the banking group where appointment occurs and leadership gained outside this group. The empirical analysis starts by examining the relation between the new CEO prior leadership experience and the (industry-adjusted) performance change surrounding the succession. The results indicate that CEO s prior leadership experience is positively related to long-term bank performance change in the post-succession period longer years of prior leadership is associated with larger performance improvement. This result contrasts with the findings in previous studies based on influence on performance played by the previous experience of the newly appointed CEO as a CEO in different firms and on samples of nonfinancial firms (Elsaid, Wang and III 2011; Hamori and Koyuncu 2015). The study then proceeds by investigating whether the place where the leadership experience is obtained matters. Essentially, the initial empirical tests are replicated by looking at whether there is any different role played by the leadership gained inside or outside the bank. Previous studies have documented that the beneficial effect of CEO s prior experience on firm performance is contingent on the context in which such skills are developed (Bragaw and Misangyi 2017), and there is evidence that insider CEO and outsider CEO have different impact on firm performance (Huson, Malatesta and Parrino 2004; Zajac 1990). Thus we have reason to test whether CEO s prior leadership experience obtained from inside and outside the bank affects bank performance in different ways. Our results show that there is a significant positive relation between prior leadership experience gained outside the banking group and bank performance change. However, no effect is found for leadership gained inside the group. The analysis indicates that the positive performance effect is mainly driven by the leadership gained outside the bank, that CEOs 3

4 with prior leadership experience in outside organizations bring better skill sets and enhance bank profitability. Studies in leadership suggest that the complexity of problems require leaders with different skill sets, and is related to leadership performance (Mumford et al. 2000). On the other hand, a leader s knowledge and skills determine his or her ability to solve complex problems (Connelly et al. 2017). The banking literature indicates that the level of diversification affects the performance and risk of US bank holding companies (Stiroh 2004; Stiroh and Rumble 2006). To analyse the impact of business model complexity, we further test whether the positive performance effect of prior leadership experience and the leadership gained outside the group is affected by the level of Diversification, which is an indicator of business model complexity in banks. The results suggest that diversification moderates the relation between leadership experience and bank performance, as well as the relation between leadership gained outside the group and bank performance. It shows that diversification mitigates the positive performance effect. The higher level of bank diversification, the less improvement in bank profitability. What s more, our analysis also finds that the moderating effect of bank diversification is significant only in less diversified banks. As the diversification level increases, the moderating effect becomes weaker and weaker until disappearing. The results so far show that prior leadership experience of the new CEO is positively associated with banks accounting performance in the post-succession period. In order to make our analysis more convincing, we compute market-based performance change with Tobin s Q and conduct a robustness test. The analysis follows the same structure with the main test. We find the positive performance effect still holds. Given the evidence that CEO s prior leadership experience improves bank performance, 4

5 we then investigate what causes the positive performance effect. The results of the positive performance consequence do not rule out the possibility that the performance improvement that we observe is motivated by bad choices of the newly appointed CEO. For instance, to boost profitability, the new CEO might engage in aggressive risk-taking. The second section of our study is to examine whether the prior leadership experience of the new CEO affects banks risk. To test this effect, we compute a risk variable: Change in Earnings Volatility, which is the difference of earnings volatility before and after CEO appointment. We use the same research models as analyzing the performance effect in the prior section. We start the analysis by testing how CEO s prior leadership experience affects bank risk in general. The results show a negative relation between the leadership experience and the change in earnings volatility, indicating that banks appointing a CEO with prior leadership experience helps reducing risk. We then look at whether the effect is affected by the place where the leadership is obtained. Consistent with the results in the previous section, we find that the risk-reducing effect is only significant when the leadership experience is obtained from outside the group. There is no relation found between leadership gained inside the group and the change in earnings volatility. In the next step, we further test whether the risk consequence is the affected by the level of bank diversification. The results suggest that the relation between leadership experience and change in earnings volatility is moderated by the level of bank diversification. The riskreducing effect of leadership experience is stronger in banks with less diversified business models. As the bank diversification level increases, the decrease in bank risk becomes weaker and weaker. There is a same result for the moderating effect on the relation between leadership gained outside the group and the change in bank risk. In summary, the results in this section are perfectly consistent with the first section, showing that the prior leadership 5

6 experience of the new CEO create value for bank performance, and reduces bank risk at the same time. Both the performance and risk effect is driven by the leadership gained outside the bank. Besides, the effects are moderated by the level of bank diversification. Our study provides several contributions to the existing literature. First, the study extends the limited number of studies on CEO succession in banks. Although CEO succession has been studied for decades, existing studies are mostly limited to non-financial firms and there is a lack of attention given to the banking sector. The existing banking literature mainly focuses on the question as to what drives CEO turnover (Webb 2008; Palvia 2011; Schaeck et al. 2011). However, few studies look at the effect of CEO characteristics of newly appointed CEOs. Till now we find only one study on executive turnover and executive characteristics in the banking industry by Nguyen, Hagendorff and Eshraghi (2015). The study examines the impact of executive director characteristics on the short-term market performance in US banks. Our study is quite different from theirs. Nguyen, Hagendorff and Eshraghi (2015) study executive directors and our study focuses on CEOs. The study is an event study on the stock market reactions to appointment announcements. It indicates the market expectation towards the appointment but is not related to the new executives business policies because the new executives are not even in position. Instead, our study investigates the changes in accounting performance in a long post-succession period. Previous studies have suggested that realized long-run outcomes of firms public events need not be consistent with short-run market reactions, and the initial reaction of a semistrong efficient market may be an inefficient long-run predictor of firm value (Delong and Deyoung 2007). To the best of our knowledge, our study is the first to examine the long-term performance effect of CEO succession in the banking sector. Second, our analysis contributes to the stream of research on the importance of CEO characteristics for firm performance (Adams, Almeida and Ferreira 2005; Bennedsen, Perez- 6

7 Gonzalez and Wolfenzon 2006; Kaplan, Klebanov and Sorensen 2012; Custódio and Metzger 2013), and contributes to the studies on CEO experience (Elsaid, Wang and III 2011; Hamori and Koyuncu 2015; Bragaw and Misangyi 2017). We extend the definition of prior CEO experience to prior leadership experience either as a top CEO or a subsidiary/division leader, and explore different forms of the leadership experience. We document an opposite result with related studies. While previous studies in CEO experience found a negative relation between prior CEO experience and firm s accounting performance (Elsaid, Wang and III 2011; Hamori and Koyuncu 2015), our analysis proves that the CEO s prior leadership experience improves bank profitability. Furthermore, our study enriches the analysis by considering the context of CEO succession. Given the evidence that the level of business model complexity affect bank performance and risk (Stiroh 2004; Stiroh and Rumble 2006), and that the complexity of problems require leaders with different skills sets (Mumford et al. 2000), we examine whether the performance and risk consequence of leadership experience is affected by the diversification level of banks business model. The results indicate that the level of bank diversification mitigates the performance effect as well as the risk effect. The rest of the paper is organized as follows. Section 2 summarizes the existing literature and develops hypotheses. Section 3 describes data and variables. Section 4 discusses methodology and analysis results. Section 5 gives conclusions. 7

8 2. Literature Review and Hypotheses Development 2.1. CEO s Prior Leadership experience, Source of Leadership experience and Firm Performance As CEO is the top position of a company and requires specific skills, it is reasonable that CEO s previous experience matters because of the acquired human capital and the enhanced understanding. And in particular, if the CEO worked in a similar position before taking the helm. Several recent studies started to look at whether a CEO s experience as a prior CEO adds value to firm performance. In general, they document a negative relation between prior CEO experience and firm performance. Elsaid, Wang and III (2011) distinguish outside CEO successors between exceos and non-exceos according to whether they have previous CEO experience or not. They find that the stock market reacts more positively to the appointment of an exceo. However, firms with exceo appointment have worse financial performance after succession. Hamori and Koyuncu (2015) also document a negative relation between prior CEO experience and the accounting performance. Bragaw and Misangyi (2017) find a negative relation between prior CEO experience and the market-based performance. The results seem contradictory to the theories of managerial human capital which suggest that prior CEO experience should be beneficial to the new firm as the newly appointed CEO brings honed general management skills (Harris and Helfat 1997; Bailey and Helfat 2003; Murphy and Zabojnik 2007b). Bragaw and Misangyi (2017) explain the negative relation and argue that the job-specific experience may not help, but rather, interfere with the new job due to the change of context. CEOs tend to rely on fewer, more familiar information sources and become overly reliant on a small number of strategic actions that have previously proven to be successful for the firm. However, this will cause problems if the environment changes and 8

9 the CEO sticks to actions that are no longer suitable for the external environment, which will ultimately drag down firm performance. Given the evidence that prior CEO experience affects firm performance, we conjecture that prior working experience as a leader matters. The leadership experience can be obtained not only from a top CEO position but also from leading positions in a subsidiary company or market division. All these experience can be considered as CEO s prior leadership experience. The banking sector is quite different from other industries they are complex institutions and require employees with special skills (Philippon and Reshef 2009). In addition, banks are highly homogeneous, thus we conjecture that the skills and experience obtained from a similar position are easier to be transferred to the new bank. Based on the above reasoning it is very possible that CEO s prior leadership experience has a positive effect on bank performance. The first hypothesis is proposed: Hypothesis 1: After controlling for other factors, CEO s prior leadership experience is positively associated with bank performance change in the post-succession period. The context in which the leadership experience is gained is quite different. For example, Bragaw and Misangyi (2017) contend that the beneficial effect of prior CEO experience on market performance is contingent on the context in which such skills are developed. They document that when CEOs gain their experience in a dynamic industry, it will ameliorate the negative effect such experience have on subsequent market-based performance. An important perspective is whether the leadership experience is obtained from inside or outside the company. As large commercial banks are normally operated as bank groups, the prior leadership of a bank CEO can be distinguished as: leadership gained inside the group, and leadership gained outside the group. 9

10 Although no study has analyzed the source of leadership experience in terms of inside or outside the group, there have been extensive studies on CEO s origin. Existing studies in the succession area have grouped CEO turnover as internal and external succession based on the origin of the new CEO. The CEO is an insider if he or she is appointed from the same company, and an outsider if the CEO has been employed at the firm for one year or less at the time of the succession (Parrino 1997; Huson, Parrino and Starks 2001). Previous studies have done a lot of work regarding how CEO origin affects firm s accounting performance. However, they haven t reached any consensus. For example, Huson, Malatesta and Parrino (2004) find a positive relation between outside appointment and firms operating performance. On the contrary, Zajac (1990) document that firms appointing insider CEOs have better profitability after succession events. There are also studies showing a mixed consequence of outside succession (Davidson, Worrell and Cheng 1990; Davidson III et al. 2002; Bailey and Helfat 2003). There is a trend that companies appear to have a growing appetite for hiring outside CEOs, particularly those who have prior experience as a CEO (Murphy and Zabojnik 2007a; Elsaid, Wang and III 2011). These trends reflect a shift in the relative importance of general managerial ability (managerial skills critical in leading a complex modern corporation but not specific to any organization) and firm-specific managerial ability (skills, knowledge, contacts, and experience valuable only within the organization) (Murphy and Zabojnik 2007a). It is assumed that when firms hire outsider CEOs, they choose candidates with high general skills. Similarly, insider CEOs are more likely to have high firm-specific skills(palomino and Peyrache 2013). Existing studies in generalists and specialists find evidence that generalist CEOs get higher payment than their counterparts (Custódio, Ferreira and Matos 2013) and are associated with higher expected return (Mishra 2014). 10

11 Our study aims to investigate whether the source of prior leadership experience whether the leadership is obtained from inside or outside the group could be a factor that influences bank performance. To be noticed, the source of leadership experience and the CEO origin are two different but related concepts. CEO origin describes all the CEOs and distinguishes between those hired from inside the company and outside the company. By contrast, the source of leadership experience defines only CEOs who have prior leadership experience, and examine where the leadership is gained. However, there are some links between the two concepts. By conjecture, a CEO with leadership experience that is obtained from inside the group is highly possible to be appointed from inside the group. Likewise, there is a high possibility that a CEO with leadership gained outside the group is appointed from outside the group. Our sample (as explained in the next section) confirmed the overlap and inconsistency of the two concepts. As the results regarding the performance effect of CEO origins remain blurring, and there is lack of evidence found in leadership experience studies, it is still an open question how the source of CEO s prior leadership experience affects bank performance. Thus we propose that both types of leadership experience bring a positive effect to bank performance in the post-succession period. The following hypotheses are suggested: Hypothesis 2a: After controlling for other factors, CEO s prior leadership experience gained inside the group is positively associated with bank performance change in the postsuccession period. Hypothesis 2b: After controlling for other factors, CEO s prior leadership experience gained outside the group is positively associated with bank performance change in the postsuccession period. 11

12 2.2. Diversification of Business Model and the Performance Effect Researchers have taken a contingent view of the effects of CEO characteristics. Instead of studying Does CEO characteristics matter? they begin to look at when does CEO characteristics matter? For example, Bragaw and Misangyi (2017) document that particular firm and industry contextual factors ameliorate the negative relation between prior CEO experience and market-based performance. If the CEO experience is gained from a publicly traded firm, or in a highly dynamic environment, the detrimental effect becomes weaker. Studies on CEO origin also give evidence that whether and how CEO origin affects postsuccession performance depends on both the internal and external context (Karaevli 2007; Chung and Luo 2013). Studies in leadership suggest that the complexity of problems require leaders with different skill sets, and is related to leadership performance(mumford et al. 2000). In turn, a leader s knowledge and skills determine his or her ability to solve complex problems(connelly et al. 2017). There is a large variation in the level of business model complexity among firms. As for the banking industry, the past decades have seen a trend that US financial holding companies are offering a growing range of financial services because of the potential diversification benefits. Studies indicate that the level of diversification affects the performance and risk of US bank holding companies (Stiroh 2004; Stiroh and Rumble 2006). The observed shift toward activities that generate fees, trading revenue, and other noninterest income has improved the performance of US financial holding companies (FHCs) from 1997 to Stiroh and Rumble (2006) find evidence that diversification benefits exist between FHCs, but these gains are offset by the increased exposure to non-interest activities, which are much more volatile but not necessarily more profitable than interest-generating activities. A more diversified bank provides various types of financial services, which increases the complexity and difficulty of managing the bank. Thus, we conjecture that the 12

13 benefit from a CEO s prior leadership experience is possibly mitigated by the increasing complexity of banks business model. The following hypotheses are suggested: Hypothesis 3: The level of diversification moderates the positive relation between CEO s prior leadership experience and bank performance such that the positive relationship is weaker when the bank has a more diversified business model. Hypothesis 4a: The level of diversification moderates the positive relation between leadership gained inside the group and bank performance such that the positive relationship is weaker when the bank has a more diversified business model. Hypothesis 4b: The level of diversification moderates the positive relation between leadership gained outside the group and bank performance such that the positive relationship is weaker when the bank has a more diversified business model. 3. Data and Variables 3.1. Sample and Data This study examines new CEO appointments in large, publicly traded US banks from January 1993 to December We use ExecuComp as the starting point to form the sample. Following Fahlenbrach and Stulz (2011) s method, we download firm-year observations for firms with Standard Industry Classification (SIC) codes between 6000 to 6300 from year 1993 to 2013, and exclude firms with SIC code 6282 (Investment Advice), because these are not in the lending business (eg., Eaton Vance Corp). In addition, we manually go through the list of firms with SIC code 6211 (Security Brokers and Dealers), because SIC code 6211 covers not only investment banks but also pure brokerage houses such as A.G. Awards. Although my sample includes investment banks, pure brokerage houses are excluded. This leaves 267 unique banks. To increase transparency, the excluded firms are listed in Appendix 13

14 A. We only keep records for annual CEOs, and identify a new CEO appointment to take place when the name of annual CEO changes from the previous year within one bank. From this initial list, we manually verify the appointment information according to companies annual report and proxy statements. Only the records with right information are kept. And we only retain CEOs for which detailed background information can be collected. We then collect information for newly appointed CEOs. CEO age is obtained from ExecuComp. Information about CEO s leadership experience is hand-collected from a variety of data sources including: companies annual reports (10-K report in SEC filings), proxy statements (DEF 14A report in SEC filings), S&P Capital IQ, Bloomberg and web sources. By doing this we construct a unique dataset for CEO s demographic and background information. The accounting data for the banks is obtained from Compustat. The market data is collected from the Centre for Research in Securities Prices (CRSP) database. In addition, we retrieve data from Bloomberg and S&P Capital IQ to fill some missing values. Our final sample consists of 144 CEO successions from year 1993 to The first CEO turnover event we record occurs in November 1993 and the last turnover occurs in December As will be explained in the next section, our dependent variables employ bank performance data for three years before and three years after CEO turnover. Therefore, our sample period is from 1990 to Measures Dependent Variables This study investigates how prior leadership experience of the newly appointed CEO affects long-term bank performance and risk. We first examine the Change in ROA, which indicates banks profitability. Change in ROA is measured by the difference of ROA before succession and after succession. In choosing the event window, we use a similar approach 14

15 with Huson, Malatesta and Parrino (2004). Profitability before succession is measured by ROA in year-1, and profitability after succession is calculated as the average over year+1 and year+2, the post two years after CEO succession event. ROA is calculated using net income divided by book value of assets. To control for industry effects, we follow King, Srivastav and Williams (2016) s method and use an industry-adjusted ROA, which is defined as a bank s ROA minus the mean ROA of all other banks for the same year. By using an industryadjusted performance measure, we eliminate any effect that is driven by the outside environment and is beyond CEO s control (Holmstrom 1982; Gibbons and Murphy 1990; Jenter and Kanaan 2015). Then we look at the change of bank risk. We measure it as the Change in Earnings volatility before and after CEO succession. Earnings volatility before the succession is calculated as the standard deviation of industry-adjusted ROA through year-1 to year-3, while earnings volatility after the succession is the standard deviation of industryadjusted ROA through year+1 to year+3. In order to indicate the percentage of changes in ROA and earnings volatility, we multiply the changes by 100. So the results in our tables indicate how many percentages of change happens surrounding the CEO succession events Independent Variables The study looks into how bank performance is affected by CEO s prior leadership experience and where the leadership is obtained. To measure prior leadership, we followed the framework of work experience measurements developed by QuiŃOnes, Ford and Teachout (1995), which suggest that the appropriate measurement for job-specific experience is the time spent on the job. Enlighted by this, we capture CEO Prior Leadership Experience as the logarithm of total number of years the successor worked as the top CEO of a company, CEO of a subsidiary or CEO of a market division before the appointment. As a large portion of banks are conglomerates with a number of subsidiaries and various divisions, we define 15

16 prior leadership if the successor has prior experience in the following situations: 1) The successor worked as the top CEO of a company before the appointment. For example, Walter V. Shipley was appointed as CEO of JPMorgan Chase & Co in 1994 and he previously worked as the CEO of the Chemical Banking Corp from 1983 to ) The successor worked as the CEO of a subsidiary. For example, Henry L. Meyer was hired as the new CEO of KeyCorp in 2001 and previously he was the CEO of KeyBank National Association, a subsidiary of KeyCorp. 3) The successor worked as the CEO of a market division. Michael L. Corbat, the new CEO of Citigroup Inc in 2012, was the CEO of the group s Global Wealth Management division between 2008 and 2009, and the CEO of the group s Europe, Middle East & Africa market division between 2011 and According to where the leadership experience is gained, the leadership is distinguished into two types: leadership gained inside the group, and leadership gained outside the group. Leadership Gained inside the Group is defined as the logarithm of total number of years the successor worked as the top CEO of a company, CEO of a subsidiary or CEO of a market division within the bank group before the appointment. Leadership Gained outside the Group is the logarithm of total number of years the successor worked as the top CEO of a company, CEO of a subsidiary or CEO of a market division outside the bank group before the appointment Moderating Variable An important dimension of our study is to examine whether the relation between CEO prior leadership and bank performance is moderated by the level diversification of banks. To measure Diversification, we follow the basic Herfindahl-type approach used in Morgan and Samolyk (2003); Stiroh (2004a) Thomas (2002) and Stiroh and Rumble (2006). Diversification accounts for variation in the breakdown of net operating revenue into two 16

17 broad categories: net interest income, NET, and non-interest income, NON, which includes fiduciary income, fees and service charges, trading revenue, and other sources of non-interest income. A higher value of Diversification indicates a more diversified mix: 0.0 means that all revenue comes from a single source (complete concentration), while 0.5 is an even split between net interest income and non-interest income (complete diversification). 2 2 Diversification=1 - (SH NET +SH NON ) (1) SH NET = NET NET+NON SH NON = NON NET+NON (2) Control Variables To control for other possible explanations of bank performance, we control for factors for both CEO level and firm level. At the CEO level, our control variables include CEO age, outsider, and financial industry experience. CEO Age is the logarithm of the age of the successor when he/she is appointed. Outsider is a dummy variable that equals to 1 if the successor is an outsider and 0 if the successor is an insider. The study follows Parrino (1997) and Huson, Parrino and Starks (2001) s definition that an outsider is a new CEO who has been employed at the firm for one year or less at the time of the succession. And a CEO who has been working in the company for more than one years is classified as an insider. A CEO who previously worked in the subsidiary company is defined as an insider. If the appointment is after a merging event and the new CEO comes from the company that was taken over, the CEO is classified as an outsider. Financial Industry Experience is defined as the logarithm of total number of years the successor has been working in financial firms such as banks, insurance company and accounting firms. 17

18 To account for bank-specific factors that could influence the dependent variables, we control for firm-level conditions for the year prior to the CEO turnover event, including bank size, bank age, equity capital, charter value, and deposits. Previous studies on executive succession have consistently identified the role of firm size and firm age on organization performance (Tushman and Rosenkopf 1996; Karaevli 2007). Therefore, this study puts Bank Size and Bank Age as controls for performance. Bank size is measured as the logarithm of total assets. Bank age is measured as the logarithm of total number of years the bank has been founded. There is evidence that the level of capitalisation and investment opportunities also influence bank performance (Berger and Bouwman 2013; Demirguc Kunt, Detragiache and Merrouche 2013), so this study controls for these traits through Equity Capital (fraction of equity book value to total assets, which is also called the equity ratio) and Charter Value (logarithm of market to book value of equity). The study controls for Deposits (fraction of customer deposits to total assets) since banks with a larger amount of deposits are less likely to face funding fragility thus influencing performance (Demirguc Kunt, Detragiache and Merrouche 2013). In order to control bank characteristics before CEO appointment, all the firm-level controls (except for bank size) are taken one year lag, thus the values in year-1 are applied. And all the variables are winsorized by 99%. [Insert Table 1 here] Table 1 presents a distribution of CEO succession over the sample years. Panel A lists the total number of CEOs in each year, the number of new CEO appointments for the year, and within the newly appointed CEOs how many are insiders. There are 144 CEO appointments between the year 1993 and 2013, among the 144 new CEOs 111 are insiders and 33 are outsiders. Panel B shows the number of CEOs with prior leadership. Among the 144 CEO successors 63 of them have prior leadership experience before the appointment. In terms of the source of prior leadership, 22 CEOs obtained prior leadership from inside the group while 18

19 45 obtained leadership from outside the group. We also find that almost all the CEOs with prior leadership obtained from inside the group are insiders: 21 CEOs are insiders among 22 CEOs with prior leadership from inside the group. Meanwhile, for CEOs with prior leadership obtained from outside the group, the proportion of insiders and outsiders are nearly half and half. We find 23 outsiders and 22 insiders among the 45 CEOs who obtained the prior leadership from outside the group. The results further confirmed the necessity to differentiate the two concepts. Table 2 gives definitions of all the variables, and Table 3 presents summary statistics in the analysis. The CEOs have an average of log years of prior leadership experience, log years of prior leadership gained from inside the group, and log years of prior leadership gained from outside the group. The average age of new CEOs is in natural log, and they have log years of financial industry experience on average. The average size of the banks in our sample is 9.849, with an average age of log years. The average of Equity capital, charter value and deposits are 0.093, and respectively. The level of bank diversification ranges from to 0.500, with a value of on average. [Insert Table 2 here] [Insert Table 3 here] 4. Methodology and Results 4.1. Does CEO s Prior Leadership Experience Affect Bank Profitability? The study firstly addresses the question: Does the prior leadership experience of a CEO affect bank performance after the CEO appointment? In this section, we investigate the relation between CEO prior leadership and the Change in ROA, which indicates a firm s 19

20 profitability. We further look at how this relation is affected by the level of bank diversification. The analysis is conducted with the following regression models: Change in ROA = α + β1 * Prior Leadership Experience+ γ * Controls + ε (3) Change in ROA = α + β1 * Prior Leadership Experience + β2 * Diversification + β3 * Prior Leadership Experience * Diversification + γ * Controls + ε (4) Our study follows Huson, Malatesta and Parrino (2004) s method in studying the changes in performance surrounding CEO succession. Performance before CEO succession is measured as the industry-adjusted ROA in event year-1. Performance after succession is measured as the average industry-adjusted ROA over event year+1 to +2. The industryadjusted ROA is ROA of the year minus the mean value of the industry. Prior leadership experience is defined as the logarithm of total number of years the successor worked as the top CEO of a company, CEO of a subsidiary or CEO of a market division before the appointment. We take Stiroh and Rumble (2006) s definition of Bank diversification, DIV, which accounts for variation in the breakdown of net operating revenue into two broad categories: net interest income, NET, and non-interest income, NON. A higher value of DIV indicates a more diversified mix: 0 means that all revenue comes from a single source (complete concentration), while 0.5 is an even split between net interest income and noninterest income (complete diversification). [Insert Table 4 here] Table 4 shows the results of our tests. Column 1 and 3 report results when we control bank size, bank age, equity capital, charter value, deposits and CEO age. Column 2 and 4 report results after adding outsider and financial industry experience as extra controls. The first two columns test the relation between prior leadership experience and change in ROA. It shows that prior leadership is significantly and positively associated with ROA change, and 20

21 this effect is not changed after we add outsider and financial industry experience as extra controls. The results suggest that banks that appoint successors with longer years of prior leadership have more significant improvement in profitability. The increase of one year s prior leadership results in the improvement of profitability by percentage (column 1) and percentage (column 2). Hence hypothesis 1 is supported. Our result is quite different from previous studies in prior CEO experience. While Elsaid, Wang and III (2011) and Hamori and Koyuncu (2015) find there is a negative relation between prior CEO experience and firms accounting performance, our study documents an opposite result, that prior leadership creates value to firms accounting performance after CEO succession. One reason for the inconsistency might be we test the relation in a new industry, the banking industry. Previous studies explain the negative relation with the argument that prior CEO experience is firm-specific and difficult to transfer. CEOs who come into their jobs with prior CEO experience tend to have a hardened worldview and set of actions, which makes it slower for them to adapt and learn in a new environment. Given by this explanation, we argue that the portability of prior CEO experience might be different in the banking industry. The banking sector is unique they are complex institutions and require employees with special skills (Philippon and Reshef 2009). The prior leadership experience in a similar position would be a great value compared with those without this kind of experience. In addition, as the banking industry is highly homogeneous, the skills and experience obtained from a similar position would be easier to be transferred to the new company, thus bringing value to the profitability. The empirical results support our hypothesis. In the next step, we would like to test whether the relation between CEO s prior leadership experience and bank performance is affected by the level of bank diversification, an indicator of the business model complexity of banks. Column 3 and 4 in Table 4 gives the 21

22 results of this test. An interaction item is added to the regression model. We find the results still report a positive and significant relation between prior leadership and ROA change. Diversification alone doesn t affect performance change. However, it will mitigate the positive relation between prior leadership and performance change. The higher level of diversification, the less degree of profitability improvement by prior leadership experience. As banks level diversification ranges from to 0.500, we pick several figures within this range to test the effect. We find that the interaction effect is only significant when the banks have low diversification level, but is not significant when the banks have high diversification level. For example, the extreme cases are: when diversification equals the minimum value of 0.032, the joint coefficient of prior leadership experience is and is significant (0.028 in p-value). When diversification takes the maximum 0.500, the joint coefficient is and is not significant (0.890 in p-value), which indicates the benefits of leadership experience is offset in banks with a high level of diversification. Overall, the results are consistent with our conjecture that the benefit from CEO s prior leadership experience is mitigated by the level of bank diversification. Hypothesis 3 is supported. Moving on to the analysis of the control variables, we find that equity capital and charter value are significantly and negatively related to change in ROA. The results are inconsistent with previous studies. While they document that better-capitalised banks and banks with higher charter values have greater profitability (Adams and Mehran 2012; Berger and Bouwman 2013; Demirguc Kunt, Detragiache and Merrouche 2013; Bennett, Güntay and Unal 2015; King, Srivastav and Williams 2016), our analysis shows that this is not true during a CEO turnover event. Next, we examine whether the relation between CEO s prior leadership experience and change in profitability is affected by where the leadership is obtained. The leadership 22

23 experience is distinguished between the one gained from inside the group and that obtained from outside group. The regression models below are applied to test this effect: Change in ROA = α + β1* Leadership Gained inside the Group + β2 * Leadership gained outside the Group + γ * Controls + ε (5) Change in ROA = α + β1* Leadership Gained inside the Group + β2 * Leadership Gained outside the Group + β3 * Diversification + β4 * Leadership Gained inside the Group * Diversification + β5 * Leadership Gained outside the Group * DIV + γ * Controls + ε (6) [Insert Table 5 here] Table 5 reports the results of this analysis. It follows the same structure with table 4 that column 1 and 3 report results when we control bank size, bank age, equity capital, charter value, deposits and CEO age, while column 2 and 4 report results after adding outsider and financial industry experience as extra controls. The first two columns test the relation between the source of leadership experience and change in ROA, while column 3 and 4 analyze how this relation is moderated by the diversification level of banks. The results in column 1 and 2 prove that leadership gained outside the group is positively and significantly associated with the dependent variable-change in ROA. It suggests that banks have more significant improvement in profitability if they hire successors with longer years of prior leadership experience that is obtained from outside the group. The increase of one year s leadership gained outside the group will result in the improvement of ROA by percentage (column 1) and percentage (column 2). On the contrast, no significant relation is found between leadership gained inside the group and ROA change, which suggest that this type of prior leadership adds no value to bank profitability. To sum up, the place where the CEO gained the leadership does matter. The positive relation is mainly driven by 23

24 the leadership that is obtained from outside the group. Thus hypothesis 2a is rejected and hypothesis 2b is supported. Moving onto column 3 and 4 which test the moderating effect of bank diversification, the results suggest that the level of diversification alone doesn t affect performance change. However, it will mitigate the positive relation between leadership gained outside group and ROA change. The results suggest that in banks with a higher level of diversification, the value of leadership gained outside group decreases. Similar to the previous analysis, we find the interaction effect is only significant when the banks have low diversification level, but is not significant when the banks have high diversification level. Regarding prior leadership that is obtained from inside the group, it still proves to have no influence on profitability change after the level of diversification is considered. Hypothesis 4a is rejected and hypothesis 4b is supported. To summarize, while the new CEO s prior leadership experience is positively associated with profitability change in general, the place where the CEO obtained the leadership matters. The positive effect on bank profitability is mostly driven by the experience that is gained from outside the group, which indicates that the knowledge and skills obtained from a different bank create value to the current bank. The positive relation between CEO s prior leadership experience and change in profitability is mitigated by the level of bank diversification. The higher level of diversification, the less improvement in bank profitability. However, the moderating effect of diversification is only significant in banks with a low level of diversification Robustness Check We have documented that CEO prior leadership experience improves banks long-term accounting performance. In this section, we measure bank performance with market-based 24

25 measurement as a robustness check. We use Tobin s Q to measure the market-based performance. The Change in Tobin s Q is measured as the difference of Tobin s Q before and after CEO succession. Tobin s Q before the succession is measured by Tobin s Q in year-1, and Tobin s Q after the succession is measured by Tobin s Q in year+2. All Tobin s Q are industry-adjusted. [Insert Table 6 here] Table 6 shows the results of our tests. While the result doesn t show a significant relation between prior leadership experience and Tobin s Q, it suggests that prior leadership gained outside the group improves Tobin s Q. The increase of one year s leadership experience results in the improvement of stability by percentage (column 3). The results also suggest that diversification level of banks moderates the relation between leadership experience and the change in Tobin s Q, as well as the relation between leadership gained outside the group and Tobin s Q. The higher level of bank diversification, the less increase of Tobin s Q. We can conclude that CEO s prior leadership experience improves not only banks accounting performance but also market-based performance Does CEO s Prior Leadership Experience Affect Bank Risk? In this section, we evaluate whether CEO s prior leadership experience is associated with the change in bank risk measured by earnings volatility. We construct the Change in Earnings Volatility as the volatility of industry-adjusted ROA before after CEO succession. The volatility of ROA before succession is calculated as the standard deviation of industryadjusted ROA in three years before the CEO appointment, while the volatility of ROA after succession is the standard deviation of industry-adjusted ROA in post three years after the CEO appointment. We then examine whether the relation between prior leadership experience and change in earnings volatility is moderated by the diversification level of 25

26 banks. We use the same regression models as in the performance test, but change the dependent variable into Change in Earnings Volatility. Table 7 and Table 8 present the results of our test. [Insert Table 7 here] We report the results in Table 7 with a similar structure in the previous section. Column 1 and 3 reports results when we control bank size, bank age, equity capital, charter value, deposits and CEO age. Column 2 and 4 report results after adding outsider and financial industry experience as extra controls. The first two columns test the relation between prior leadership experience and change in earnings volatility, while column 3 and 4 show how this relation is affected by the level of bank diversification. While the first two columns do not show any significant relation between leadership experience and the dependent variable, when we add bank diversification as a moderating variable, it shows a negative relation between leadership experience and change in earnings volatility. And this effect is robust after we add outsider and financial industry experience as extra controls. The results indicate that banks appointing successors with longer years of prior leadership experience are associated with a more significant level of decrease in earnings volatility. Since earnings volatility is a reflection of firm risk, we can draw a conclusion that new CEO s prior leadership experience helps ameliorate bank risk. More specifically, the increase of one year s prior leadership experience will result in a decrease of earnings volatility by percentage (column 3) and percentage (column 4). Although existing studies have tested the market reaction towards prior CEO experience (Elsaid, Wang and III 2011), the relation between prior CEO experience and firms accounting performance (Elsaid, Wang and III 2011; Hamori and Koyuncu 2015) as well as market-based performance (Bragaw and Misangyi 2017), there is no study looking at how CEO s prior leadership experience affect firm risk. To the best of our knowledge, our study is the first to examine this question. 26