This paper aims to investigate the roles of non-CEO inside directors (NCIDs) in the new CEO-firm matching process using the context of unplanned CEO departures when immediate CEO succession planning becomes a sole board responsibility. Although critics argue that inside directors decrease the monitoring effectiveness of a board, inside directors arguably possess superior firm-specific experience and knowledge that can be beneficial during the leadership transition.
The authors use a comprehensive, manually collected data set of unplanned CEO departures from 1993 to 2012.
The authors find that NCIDs play an important role in the CEO transitioning process. They help firms identify qualified inside replacements and provide stability as the new permanent or interim CEO. In addition, NCIDs facilitate the transfer of information and help the new external CEOs succeed. They show that the longer the NCID stays with the company, the longer the tenure of the new CEO. They also document that the presence of NCIDs improves operating and stock performance; especially when the new CEO is hired from outside of the firm.
The impact of NCIDs is particularly important when the firm hires an outsider as the new CEO. These results suggest that board composition affects frictions in the CEO labor market.
The literature has predominantly focused on the downside of having inside directors. Too many inside directors on a firm’s board is often associated with ineffective boards and entrenchment. To the contrary, the authors focus on a potential benefit of having inside directors.
Krigman, L. and Rivolta, M.L. (2019), "Can non-CEO inside directors add value? Evidence from unplanned CEO turnovers", Review of Accounting and Finance, Vol. 18 No. 3, pp. 456-482. https://doi.org/10.1108/RAF-06-2018-0111
Emerald Publishing Limited
Copyright © 2019, Emerald Publishing Limited
Can the composition of the board of directors help mitigate the friction costs associated with CEO turnovers? The CEO-firm match theory posits that the CEO labor market is efficient and competitive and that the matching between CEOs and firms is optimal (Jenter et al., 2016). However, both anecdotal and empirical evidence show that CEO departures, particularly unplanned CEO departures, can create friction (i.e. costs associated with appointing a non-optimal CEO replacement), be disruptive, and negatively affect shareholder value. Consider, for example, the unanticipated resignation of Hewlett-Packard’s (HP) former CEO Mark Hurd on August 6, 2010. HP’s market value fell by $10bn and it took over six months for the stock price to recover. In addition, HP’s return on assets (ROA) fell from 7.50 per cent to −11.87 per cent over the three years following Hurd’s departure, nearly a twenty per cent decline. Beyond anecdotal evidence, a recent study on corporate governance and CEO succession by PWC finds that frictions can cost up to $112bn when firms are unprepared for a CEO transition. Despite the large magnitude of these costs, there is little empirical evidence on what factors can help mitigate them. We explore whether board structure, specifically the presence of non-CEO inside directors (NCIDs) can help firms reduce friction costs during unplanned CEO transitions.
We argue that NCIDs possess superior firm specific knowledge relative to other board members. As such, they can be a source of insightful guidance in both the selection of the new CEO, and the operation of the firm following the leadership transition. Zorn et al. (2017) relate the absence of NCIDs to excessive CEO pay, increased financial misconduct and worse firm performance. They argue that inside directors strengthen the monitoring capabilities of independent directors by providing access to critical information. Masulis and Mobbs (2011) and Mobbs (2013) suggest that inside directors can be high quality internal candidates for the CEO position. However, if an NCID is passed-over for the CEO position, it is doubtful he would work to help the new CEO during the transition. If the NCID is not a candidate for the permanent CEO position, he could provide continuity and stability serving as interim while the firm performs a comprehensive search for a new permanent CEO.
Following the selection of the new CEO, we argue that NCID guidance and firm-specific knowledge should be most helpful if the board decides to hire an outsider as the new CEO rather than selecting an internal candidate. Outsiders presumably have experience in bearing CEO-level responsibilities, however, they may take time to assimilate and transition to the existing corporate culture. NCIDs could serve as a guide though this process. The literature has predominantly focused on the downside of having inside directors. Too many inside directors on a firm’s board is often associated with ineffective boards and entrenchment. To the contrary, we focus on a potential benefit of having inside directors.
Unplanned CEO departures can arise due to death and illness, unexpected forced departures due to lawsuits and criminal investigations, and poor performance or disagreements with the board. It is under these circumstances that immediate CEO succession becomes a primary board responsibility and the departing CEO is not available for consultation. We examine whether the presence of NCIDs on the board improves a firm’s ability to weather the loss in executive leadership.
We find that the likelihood of firms identifying an internal CEO replacement is significantly greater when firms have at least one NCID on the board. We demonstrate that inside directors indeed often become the new CEO after the sudden loss of the CEO. However, even when an NCID does not assume the role of either permanent or interim CEO, we show that NCIDs are still beneficial when there is CEO turnover. Prior literature suggests that CEO turnover is costly and that replacing a CEO can cost shareholders nearly $200 m. We demonstrate that newly appointed CEOs have longer tenure when there is an NCID on the board. We also show that the longer the NCID stays on the board following the transition, the longer the tenure of the new CEO. This is consistent with the NCID providing continued assistance and guidance to help assure the success of the new CEO following the transition.
Extant literature has focused on the valuation effect of outside CEO replacements in isolation and ignored the potential role inside directors can play in smoothing the leadership transition. We fill this gap by examining the valuation and performance implications of inside versus outside CEO replacements interacted with board composition. In isolation, outside replacement CEO firms have worse operating performance following a leadership change. We also find that the presence of an NCID makes no difference in the operating performance of firms following an unplanned CEO transition. However, when firms hire new CEOs from outside the firm and they have at least one NCID on the board, they perform significantly better. Industry-and-performance adjusted operating performance from one year before to three years after the CEO departure is eight per cent higher for NCID firms that hire outside new CEOs. We also show that the stock performance of firms with an
NCID that hire a new CEO from outside the firm significantly outperform comparable firms with no NCID for up to three years following the appointment of the new CEO.
Our paper helps to shed light on the important role directors, particularly NCIDs, can play in mitigating friction costs when firms face the critical shock of unexpectedly losing a CEO. Given the importance of NCIDs, the findings in this paper help inform the debate on uniform mandates for boards. Despite the importance of CEO succession planning, empirical evidence on its necessity and the planning process has been scarce, possibly because firms are hesitant to disclose detailed succession planning information. For instance, when Ford Motor Co. announced that Mark Field would succeed Allan Mulally as new CEO, Ford’s spokeswoman mentioned that the “company takes succession planning very seriously and has succession plans in place for each of the key leadership positions. However, for competitive reasons, Ford does not discuss succession plans externally.” We overcome the lack of publicly available detail in succession planning by using a comprehensive and hand-collected dataset with unplanned CEO departures and offer a glimpse of the role board members can play in facilitating successful succession.
We organize the remainder of the paper as follows; Section 2 presents the hypothesis development, Section 3 describes the data and provides summary statistics, Section 4 presents our empirical results, Section 5 provides robustness checks and Section 6 concludes.
2. Hypothesis development
Existing literature has predominantly focused on inside directors as providers of information to the board and as internal monitors of the firm (Mobbs (2013)). We extend the literature and focus on the value NCIDs can provide to the firm under the extraordinary circumstances of non-planned CEO departures. These departures create an exogenous shock that tests a firm’s ability to weather the sudden shift of leadership. When the departing CEO is no longer available for consultation in the new CEO selection process, the remaining directors bear the responsibility to plan for succession and run the company during the transition.
In addition to monitoring and advising, inside directors play a third important role, which is potentially to become the new CEO. Masulis and Mobbs (2011) define inside directors holding other public board seats as certified inside directors (CIDs), and argue that the knowledge and skills possessed by these CIDs provide incentives for the current CEOs to improve performance, or the CIDs may replace them. Mobbs (2013) shows that these highly qualified inside directors are an important source of competition for CEOs and are likely replacements for poorly performing CEOs. Given that inside directors possess firm specific knowledge, and that they are already familiar with other board members and senior managers, we hypothesize that inside directors should be able to assume the role of CEO in either a permanent role for a quick recovery or in a temporary role to provide stability during the search for a new permanent CEO. As firms with more NCIDs have a larger pool of internal candidates, it may take less time to appoint a candidate from inside than outside the company. Our first set of hypotheses relate to the choice of an inside versus outside CEO replacement.
Boards with a greater number of NCIDs are more likely to appoint an insider as the new CEO after an unplanned CEO departure.
Unplanned CEO departures generate uncertainty and disruption costs during the transition period. Following prior literature, we proxy these costs using new CEO turnover, CEO departure and replacement announcement returns, changes in operating performance and abnormal stock returns following the CEO turnover.
Boards of directors necessarily have incomplete information about CEO candidates. Many firms hire the wrong person and subsequently need to replace the new CEO, leading to increased CEO turnover. Repeated CEO turnover generates substantial cost to companies given the average severance pay that firms are providing their executives (Huang, 2011). In fact, existing literature has found that new CEOs, especially those hired from the outside, typically have 18 months to prove their competency to shareholders (Zhang and Rajagopalan, 2008, 2010). Having NCIDs can enhance the ability of the firm to identify internal CEO candidates through two mechanisms. First, a NCID could become the new CEO. Second, NCIDs’ familiarity with the workings of the company and the personnel could enhance the ability of the board to identify an internal non-NCID CEO candidate. We argue that boards with strong insider presence should have superior firm-specific information to use in selecting a new CEO.
Prior studies suggest that firms with inside directors have better operating performance (Masulis and Mobbs (2011)) and higher valuations (Coles et al. (2008)). We focus on NCIDs and test two opposing views regarding the roles NCIDs may play with regard to performance surrounding unplanned CEO departures. Agency theory suggests that CEOs select inside directors to facilitate entrenchment (Bebchuk and Fried (2003). NCIDs seek job and compensation security, and thus are unlikely to take positions against the new CEO, which leads to weakened monitoring and advising roles, and poorer firm operating performance. To the contrary, the optimal contracting literature suggests that firms choose board composition to maximize shareholder value by improving board knowledge and expertise. This allows superior monitoring and advice management Fama and Jensen (1983), Raheja (2005), Harris and Raviv (2008)). If firms optimally choose their board composition and continuously re-adjust it, it is unlikely that we would observe a significant relationship between NCID and the change in firm performance around unplanned CEO departures. However, structures such as classified and staggered boards, as well as external frictions (i.e. government regulations such as SOX) may force firms to deviate from optimal board structures (Yermack (2004), Masulis and Mobbs (2011)). If that is the case, we expect to find that NCIDs enhance firm operating performance and value.
Studies as early as Fama and Jensen (1983) argue that outside directors lack information on firm projects and that inside directors possess superior firm specific information relative to outside directors. Adams and Ferreira (2007) advocate “friendly board” structures with insiders, as they argue that insiders possess more firm-specific information, and that a friendly board facilitates the transfer of information from insider to outside board members. Confirming this, Mace (1986) reports that based on a survey of corporate directors, outside directors frequently use insiders as a source of information. Therefore, the presence of an NCID should help facilitate the flow of information to the board and make for management that is more effective throughout the CEO transition period. The hypothesis we test with respect to transition and disruption costs is:
Firms with NCIDs should experience lower transition and disruption costs.
We argue that if this hypothesis is true, then firms should have longer new CEO tenure and stronger operating and stock performance. These effects should be the strongest when the new CEO is an outsider.
3. Data and sample selection
3.1 Sample construction
We identify 2,874 CEO departures during the period 1993-2012 from ExecuComp, which covers current and former S&P 1500 firms. By ending the sample in 2012, we are able to examine financial performance at the firm for up to three years following the turnover event. For these 2,874 CEO departures, we use Factiva, Lexis Nexis, and proxy statements to gather the following information about the incumbent, successor, and interim CEOs: experience and background, dates of departure and takeover, and the cause of the turnover event. We classify the causes of departure into natural retirement, forced resignation, and sudden departures (i.e. sudden death or illness). We hand collect the earliest announcement date of incumbent CEO departure and new replacement CEO appointment as well as the actual incumbent CEO departure and new CEO takeover dates. For the subsample of unplanned CEO departures, we investigate whether the departure is due to sudden death, illness, or is due to the incumbent CEO having been hired away, either by a better company or through a government job. We manually fill in any observations where CEO appointment date is missing in ExecuComp to identify CEO tenure. Control variables, including stock returns, accounting information, and CEO characteristics come from CRSP, Compustat, and ExecuComp. Board experience and corporate governance data are sourced from the RiskMetrics board databases. For the purposes of our analysis, we focus on unplanned CEO departures and exclude planned retirements. Our final sample consists of 582 unplanned CEO departures from 1993 to 2012. As can be seen in Table I, the percentage of boards containing NCIDs has declined over time. Prior to the enactment of Sarbanes Oxley in 2002, 49 per cent of firms had at least one NCID on the board. Since 2003, the average has dropped to only 24 per cent of firms having an NCID on the board.
We also find that the per cent of firms hiring an outsider as the new CEO is highly variable over time and averages 45 per cent for the full sample.
We categorize the departures by the stated reason the CEO left. Eleven per cent of the sample is due to death or illness, 10 per cent is due to CEOs leaving for better opportunities, and 66 per cent of the sample is forced departure. Table II
3.2 Non-CEO inside director characteristics
Using public filings, we identify 248 individual NCIDs from the 208 firms that report having an NCID. Table III contains summary information on these directors. The most common titles for the NCIDs are President, COO, Chairman and CFO. We track these directors for three years following the change in CEO to determine if they remain with the firm and on the board. We find that in 52 per cent of the firms with NCIDs, an NCID remains on the board and with the firm for at least one year following the CEO transition. Forty per cent of firms have an NCID remain for two years and 33 per cent of firms still have an NCID with the company and on the board three years following the CEO departure.
3.3 Measuring firm operating and stock performance
Following Barber and Lyon (1996), we measure operating performance as the change in industry and performance-adjusted ROA from before to after the arrival of the new CEO. This approach controls for potential mean reversion in accounting returns for poorly performing firms. We match each sample firm with an unplanned CEO departure with a control firm with no CEO departure. Industry-and-performance adjusted ROA is then defined as each sample firm’s ROA less the ROA of the control firm, matched on primary two-digit SIC industry and with the ROA within 10 per cent in the previous year. If no firm in the same two-digit industry has a one-year ROA within 10 per cent, we select a firm in the same one-digit SIC industry, and then match based on one-year ROA within 10 per cent. We calculate the change in industry and performance adjusted ROA for one, two and three years following the change in leadership. We also calculate one-, two- and three-year excess stock returns using the same control firms. We define excess return as each sample firm’s cumulative market adjusted stock return less the stock return of a control firm, matched on industry and prior year firm performance.
4. Empirical results
4.1 Univariate analysis
We begin our analysis by comparing firm and corporate governance characteristics for firms with and without NCIDs. Table IV presents univariate comparisons across these two samples. We provide the definition of all variables in the Appendix. Panel A of Table IV contains firm characteristics. We find that firms with NCIDs have higher Tobin’s Q, and prior year industry and performance adjusted performance. These firms are also slightly younger, and have lower institutional ownership and return volatility. In our estimations, we will control for these differences.
In Panel B of Table IV, when we turn to corporate governance characteristics, by construction, more differences that are significant emerge. Firms with at least one NCID have greater insider presence (by definition). Typically, these firms have two inside directors, one of which is the CEO, whereas firms with no NCIDs have one inside director, namely the CEO. Firms with NCIDs have larger boards and lower board independence. These firms are also more likely to be clustered prior to 2001 (Table I), before the board independence requirement mandated by the Sarbanes Oxley Act (SOX) took effect.
4.2 Inside succession and the appointment of inside director as replacement CEO
We next examine new CEO characteristics and report the results in Table V. Panel A contains univariate statistics and tests of differences between NCID and non-NCID firms’ permanent replacement CEO characteristics for the full sample. We find that the new CEO is significantly more likely to be an insider if the firm has at least one NCID. In firms that have at least one NCID, an internal permanent CEO replacement is found 65 per cent of the time, whereas when there is no NCID, internal candidates are identified only 36 per cent of the time. For firms with at least one NCID, the permanent CEO replacement is an employee board member 46 per cent of the time, whereas in firms with no NCIDs, the new CEO is an internal board member only 18 per cent of the time. Thus, consistent with our first hypothesis, we find that firms with greater inside director representation are significantly more likely to draw the new CEO from inside the firm. We believe two channels facilitate this process. First, inside directors become the new CEO 46 per cent of the time and second, inside directors are likely better connected with the employees of the firm, and therefore are better situated to identify qualified internal candidates.
In Panel B, we present the same data for the sub-sample of CEO departures we classify as unexpected. These departures include unexpected death and illness, and resignation due to lawsuits and criminal investigations. In these unplanned and unexpected circumstances, the board would necessarily have less time for succession planning if they do not already have a plan in place. We find that when the departure is truly unexpected, firms with NCIDs are significantly more likely to appoint the new CEO from within the firm; 84 per cent of new CEOs are current employees and 59 per cent are current board members. When there is not an NCID present, internal candidates are the replacement 48 per cent of the time and in only 19 per cent of the instances, a current board member takes over as CEO.
To examine the determinants of an inside succession and the appointment of an inside director as the new permanent CEO, controlling for firm characteristics, we use Probit models. The two independent variables of interest are whether firms have at least one NCID (NCID= 1), and the total number of inside directors on board (Number of NCIDs). We also include Unexpected CEO departure to determine whether the choice of replacement CEO is related specifically to the sudden death, illness or resignation of the incumbent CEO.
Following Coles et al. (2008), we include a number of other variables in the regressions to identify whether other firm and corporate governance characteristics have an impact on the choice of replacement CEOs. These variables include R&D intensity to capture firm complexity, firm size (log market Cap), firm age (log firm age), and board size (log board size). We include Free Cash Flow to measure whether financial considerations motivate the choice of an inside successor. We also include an indicator to capture whether the incumbent CEO was also a firm founder (Founder CEO). We include an indicator equal to one if the firm has a Classified Board and an indicator equal to one if the CEO departure occurred following the enactment of Sarbanes Oxley (Post Sarbanes Oxley). We also control for the level of Institutional Ownership.
Naveen (2006) argues that firms with greater information asymmetry are more likely to use inside succession. When information asymmetry and the cost of becoming informed are high, we expect firms to be more likely to appoint an inside CEO replacement. Therefore, we control for R&D Intensity to capture information asymmetry in our estimations. Following Duchin et al. (2010), we use the residual from an OLS regression of the number of analysts covering the firm regressed against firm size, as the size-adjusted number of analysts (Analyst Coverage), as our measure of information asymmetry. Denis and Denis (1995) argue that firms are more likely to appoint an outside successor following prior poor performance. Therefore, we include both prior year cumulative market adjusted stock return (Prior year stock return) as well as Stock Return Volatility to control for stock performance and firm risk, respectively.
Table VI contains the results of the Probit estimations and marginal probabilities examining inside succession as it relates to NCIDs. In Panel A, the dependent variable is an indicator equal to one if the new CEO came from within the firm. In panel B, we restrict the dependent variable to be equal to one if the new CEO was an NCID within the firm. Consistent with the univariate results, we provide strong evidence that board composition matters in the choice of inside versus outside CEO succession.
As can be seen in Panel A, having at least one NCID on the board increases the probability of inside succession by 26.5 per cent (based on marginal probabilities). Additionally, each incremental NCID increases the likelihood of inside succession by 16 per cent. Firms are significantly more likely to use inside succession when the have larger boards that are less independent and following the enactment of Sarbanes Oxley. Firms with higher free cash flow and Tobin’s Q are also more likely to use inside succession.
Existing literature has found that complex firms and those with greater information asymmetry tend to use inside successions. After controlling for NCIDs and firm and governance characteristics, we find no significant relationship between inside succession and R&D intensity. We believe that the difference between our paper and prior studies is due to sample construction. Other studies examine CEO turnovers in general whereas we focus on unplanned departures and exclude planned retirements. Even in the case of forced departures (a sizeable portion of our sample), in more complex and growth oriented firms, boards may select a new CEO based more on firm need and/or on an effort to turnaround poor performance and less on the information environment.
The likelihood of boards appointing an inside director as the new CEO demonstrates a similar pattern as the likelihood of inside succession (Panel B of Table VI). We find that controlling for firm and industry effects, the likelihood of an NCID becoming the new CEO is 29.7 per cent. Each additional NCID increases the likelihood of an NCID assuming the role of new CEO by 16 per cent. Most of the fundamental firm-level control variables do not show up as significant determinants of inside succession. This suggests that the selection of new CEO replacement is related more to board composition than other firm characteristics. Overall, the results in Table VI are supportive of our first hypothesis. Controlling for firm and governance characteristics, the marginal probability of a NCIDs becoming the new CEO is nearly 30 per cent. Additionally, NCID presence significantly increases the likelihood of inside succession even when they do not assume the role of CEO.
4.3 Board composition and transitional costs of unplanned CEO departure
Having documented that board composition is associated with the selection of the new CEO; we next consider whether it can help mitigate the costs of unplanned CEO departure. Table VII provides summary statistics on the disruption and transition costs of CEO departure and replacement.
Measures include the CAR surrounding the CEO departure and replacement announcements, new CEO tenure and changes in operating and stock performance for one through three years following the appointment of the new CEO.
Panel A provides details on the full sample. All firms experience significant negative performance surrounding the announcement of the CEO departure. On average, we find very few differences in the transition costs for firms with and without NCIDs. A few notable exceptions are the tenure for the new CEO and two and three-year stock performance. New CEOs stay on average 6 months longer at firms with an NCID on the board. Interestingly, NCIDs on the board are associated with significantly worse stock performance following the CEO departure. The average two-year excess stock return for NCID firms is −13.4 per cent compared to a gain of 5.2 per cent for firms with no NCIDs. Thus on first pass, it appears that NCID presence is harmful to firm performance. We find no statistically significant difference in changes in operating performance, the number of days it takes to replace the CEO (not tabled) and the announcement CARs for both the CEO departure and replacement, surrounding sudden CEO leadership transitions related to the presence of an NCID on the board.
Our contention is that NCIDs should make the most difference for the firm when the new CEO is an outsider. In these circumstances, the NCID should be a source of information for the new CEO and provide continuity. Therefore, we partition the data by whether the firm chooses an inside or outside replacement for the CEO. Panel B of Table VII contains details for the sample of firms that chose an internal CEO replacement. We find no difference in the tenure of the new CEO, the number of days without leadership (not tabled) and the announcement CARS. In terms of operating performance changes, we find that the average change in industry and performance adjusted ROA is higher for firms with no NCIDs. Consistent with the results on the full sample, when the CEO replacement is an internal candidate, we find that firms have stronger stock and operating performance if they do not have an NCID on the board. Masulis and Mobbs (2011) argue that certain inside directors (e.g., better connected inside directors) can be a replacement for the current CEO. Consistent with this conjecture, Mobbs (2013) finds that these inside directors are significantly more likely to become CEOs relative to other inside directors. Thus, it is possible that the weaker stock and firm performance observed in firms with NCIDs when the CEO replacement is an inside candidate is associated with internal competition for the new CEO position after unplanned CEO departure.
Panel C of Table VII provides details on the sample of firms that chose an outside CEO replacement. Again, univariate statistics reveal few differences in transition costs between NCID and no NCID firms. We find that based on both mean and median; new outside CEOs have longer tenure when there are NCIDs on the board. We also find that these firms have better operating performance over one and three year horizons following the arrival of the new outside CEO.
Finally, we test for differences across inside or outside CEO replacement by NCID presence. Panel D of Table contains the p-values from t-tests and Kruskal–Wallis median tests. Significant differences emerge. Given that a firm has an NCID on the board, we find that both the future operating and stock performance is significantly stronger when the new CEO is an outsider. For example, the change in operating performance (industry and performance controlled) from one year prior to the new CEO arrival to one year following averages an increase of 6.26 per cent for NCID firms compared to a decline of 1.07 per cent for non-NCID firms. Over the same horizon, the excess 12-month stock return for NCID firms is +9.8 per cent compared to −9.3 per cent for non-NCID firms. When the firm replaces the CEO with an insider, we find future operating performance is stronger when there is no NCID present. We detect no statistically significant difference in stock performance when there is no NCID present. Thus, consistent with our hypothesis, NCID presence appears to have the most positive influence when the new CEO is from outside the company. We illustrate these comparisons graphically in Figure 1. It is important to keep in mind that these univariate tests do not control for firm and corporate governance differences we reported in Table IV. Therefore, we turn to a multivariate analysis that controls for both firm differences and the endogenous nature of board structure.
4.4 Multivariate analysis on new CEO tenure
Our contention is that the presence of an NCID on the board should help facilitate the transition of leadership. One way the firm could benefit is if the new CEO has a longer tenure with the firm. We see two paths through which an NCID could affect the new CEO tenure. First, the presence of an NCID could aid the board by providing time and guidance to select a candidate that has a higher chance of success. This effect should be present regardless of how long the NCID stays with the firm following the transition. The second pathway is that the NCID could be helpful to the new CEO and the board after the transition and thus the effect should be dependent upon the length of time the NCID stays on the board.
A potential area of concern is the possibility that new CEO tenure is jointly determined with NCID presence on the board, thus causing endogeneity. Following Goldstein and Nelling (1999), we address this by estimating a system of simultaneous equations to examine the relations among the variables. The dependent variable in the first equation is the new CEO tenure (in years). The independent variable of interest is NCID, an indicator equal to one if the firm had an NCID on the board at the time of the CEO transition. We also create three new indicator variables to capture the length of time the NCID remains with the firm (NCID1, NCID2 and NCID3 are indicator variables equal to one if the NCID remains on the board for 1, 2 or 3 years respectively following the arrival of the new CEO). Our conjecture is that the presence of an NCID should be more important when the new CEO is from outside of the firm. Therefore we include an indicator variable equal to one if the firm hires an outside new CEO (Outside CEO Replacement), and an interaction term (NCID (1, 2 or 3) * Outside CEO Replacement) to capture the role NCIDs play when an outside new CEO is hired.
The dependent variables in the second equation are NCID, NCID1, NCID2 and NCID3 as defined above. For identification purposes, in the second equation, we also include an instrumental variable – the geographic location of the company (the number of CEOs within 50 km radius of a company’s headquarters (Firm Count w/in 50 Km)) to predict the presence of NCIDs. Recent studies on the geographic location of the firm find that firms’ headquarter location is an important determinant of board structure (Masulis, Wang, and Xie (2012); Knyazeva, Knyazeva, and Masulis (2013); and Alam, Chen, Ciccotello, and Ryan (2014)). Specifically, these studies show that proximity to larger pools of local director talent leads to boards that are more independent. Thus while the geographic location of the company may influence how firms structure the board (i.e. the use of NCIDs), it is unlikely to directly influence how long the new CEOs stay.
We rely on related prior studies, including Yermack (1996), Naveen (2006), Coles et al. (2008), and Coles et al. (2012), for guidance in selecting independent variables. We include an indicator variable (Unexpected Departure) equal to one if the CEO departure was caused by sudden death/illness or resignation due to lawsuits and criminal investigations, but not forced dismissal. We include the control variables board size (Log Board Size), board independence and new CEO age (Log CEO Age) which is included to proxy for new CEO experience (Brickley, 2003). To control for information asymmetry, we include the residual from an OLS regression of the number of analysts covering the firm regressed against firm size, as the size-adjusted number of analysts (Analyst Coverage). We include stock return volatility from the year prior to the CEO departure (Stock Return Volatility) to control for risk. We include an indicator equal to one if the firm has a classified board (Classified Board) and an indicator equal to one if the CEO departure occurred following the enactment of Sarbanes Oxley (Post Sarbanes Oxley). We also control for institutional ownership. The estimations contain industry and calendar time fixed effects. Table VIII contains details of the estimations.
Column 1 of Table VIII contains the results where NCID equals 1 if there was an NCID on the board at the time of the CEO departure. In equation (1) of the simultaneous equation estimation, we find that when the new CEO is an outsider, consistent with our hypothesis, new CEO tenure is longer with the presence of an NCID. The coefficient on NCID*Outside Replacement is positive and significant. Having an NCID on the board is not significantly related to the new CEO tenure when the new CEO is from inside of the firm (the coefficient on NCID on Board is not significant). Additionally, we see in equation (2) that new CEO tenure is significantly related to the presence of NCID, thus confirming that NCID and New CEO Tenure are jointly determined.
In columns 2, 3 and 4, we model the relation between New CEO Tenure and NCID (1, 2, and 3) continued presence on the board. As in the estimation in column (1), we find that new CEO Tenure is also jointly determined with the continued presence of an NCID on the board. In general, outside New CEOs have lower tenure, but when the NCID stays and the new CEO is an outsider, new CEO tenure is longer. These results hold when the NCID remains on the board for 1, 2, and 3 years following the new CEO arrival. We also find that firms with higher institutional ownership have lower new CEO tenure whereas firms with classified boards have longer new CEO tenure. Overall, NCID presence both at the time of leadership change, and in the years following the new CEO appointment appear to be important to new CEO tenure when the new CEO is an outsider.
4.5 Multivariate analysis on operating performance
Univariate statistics showed that NCID presence is associated with the post-transition operating performance of unplanned CEO departure firms. We now examine these results in a multivariate setting. To control for the endogenous choice of having an NCID on the board, we estimate an endogenous treatment-regression model. We instrument NCID with the geographic location of the company (the number of firms within 50 km radius of a company’s headquarters). We use the same control variables as we did in the estimations for CEO tenure in Table VIII. In addition, we control for information asymmetry using R&D Intensity, the delay in appointing the new CEO (Log Delay in Replacing CEO), firm size (Log Market Capitalization), firm age (Log Firm Age) and Tobin’s Q.
The dependent variable in the estimations is the change in industry-and-performance-adjusted ROA one, two and three years following the new CEO appointment. The independent variables of interest include a binary variable (NCID) to indicate whether the board has at least one NCID. We include an indicator variable equal to one if the firm hires an outside new CEO (Outside CEO Replacement), and an interaction term (NCID * Outside CEO Replacement) to capture the role NCIDs play when an outside new CEO is hired. Table IX contains the estimations.
Consistent with prior studies, we find that the appointment of an outside CEO is associated with significantly worse operating performance than the appointment of an inside CEO replacement. However, after controlling for the endogeneity of NCID presence, as well as firm characteristics, we find that the change in firm operating performance following CEO turnover is significantly higher when a non-CEO insider serves on the board of directors and the firm hires an outside CEO.
Specifically, the coefficients on the interaction term NCID* Outside CEO Replacement are positive and significant at least for three years after the CEO turnover. Firms with NCIDs at the time of an outside CEO appointment are associated with 9.18 per cent higher one year ROA compared to firms without
NCIDs. The relative performance improvement increases to 15.78 per cent over two years.
4.6 Analysis on stock performance
Having demonstrated that operating performance is stronger for firms that both have an NCID and choose an outside CEO replacement we ask whether this improved operating performance translates into shareholder value. We use Fama-French three-factor calendar time regressions to examine stock performance. We evaluate the abnormal monthly returns of four sub-samples of data related to NCID and inside versus outside CEO replacement. We create portfolios of 12, 24 and 36 months. Details are contained in Table X.
We find significant positive abnormal returns for firms with at least one NCID that hire a new outside CEO for up to three years following the CEO departure. Abnormal performance averages 1.2 per cent per month for 24 months (33.7 per cent compounded abnormal performance over 24 month) following the CEO transition. Over three years, the abnormal performance drops to 0.59 per cent per month. For firms that choose an inside CEO replacement, we find no abnormal performance regardless of the presence or absence of an NCID. For firms with no NCID and an outside CEO replacement, in the 12-month portfolio, we find abnormal performance of 78 basis points per month. We find no abnormal performance for portfolios of stocks with no NCID and for those with an NCID that select an inside replacement for the CEO. Overall, both the operating and stock performance results are consistent with our second hypothesis; NCIDs add value when the firm hires from the outside. In the next section, we provide robustness checks on our results.
5. Robustness tests – subsample analysis
In our analysis of unplanned CEO departures over the years 1993-2012, we find results supportive of our hypotheses – having an NCID on the board reduces CEO departure transition costs and improves both the operating and stock performance in the years after the transition, especially when the new CEO is an outside hire. In this section, we identify sub-sections of the data for further insight and analysis.
The type of CEO departure where the presence of an NCID should have the strongest impact is when the old CEO departs due to sudden death and/or illness. In these instances, there is little to no time for planning the transition and the NCID should offer the greatest service to the board in selecting the replacement CEO. We replicate the estimations from Table IX, which focuses on operating performance following the CEO transition using only the 68 instances where the old CEO either died or took suddenly ill. We report the interaction term NCID * Outside CEO Replacement from the estimations in Panel A of Table XI.
For the death/illness subsample, we find that two and three year industry and performance adjusted operating performance is stronger when the firm has an NCID and the firm hires an outside new CEO. The magnitude of the estimated coefficients is higher than that for the full sample. Thus, in the sub-sample where the results should manifest strongest, that is exactly what we find.
Another concern is that the presence of an heir apparent (i.e. a single manager being groomed for succession) would dampen both the need for, and the impact that an NCID has on the firm following the CEO departure. Therefore, our results could be driven by the presence of an heir apparent rather than the presence of an NCID. To examine this, we exclude all firms where we could identify an heir apparent. We define heir apparent as an executive with the title of President or Chief Operating Officer that is at least 5 years younger than the departing CEO is. After eliminating these NCIDs, our sample reduces to 526 observations. Again, we report the interaction term NCID * Outside CEO Replacement from the estimation in Panel B of Table XI. Similar to the results for the death/illness subsample, the presence of an NCID when the firm hires an outside CEO replacement continues to result in stronger operating performance in three years following the CEO departure when we exclude heir apparent firms. We next replicate the Fama-French three-factor calendar time regressions from Table X on the No Heir subsample. Panel C contains the details. Consistent with the results for the full sample, we find significant positive abnormal performance (alpha) for 12, 24 and 36 months for firms with an NCID that choose an outside CEO replacement. Consistent with the results presented for the full dataset, we find that the presence of an NCID when the firm hires an outside CEO replacement continues to result in stronger operating and stock performance in three years following the CEO departure.
While considerable literature post SOX has focused on the corporate governance in firms with majority independent (i.e. outside) directors, we know relatively little about the corporate governance in firms with NCIDs after 2001. These directors arguably possess both superior firm specific knowledge as well as experience, and thus should better be able to govern companies facing crises. We posit that the presence of NCIDs is likely to lower friction costs if there is an unplanned shift in executive leadership.
Using a manually collected sample of 582 unplanned CEO departures, we examine if NCID presence can improve a firms’ ability to weather a sudden shift in CEO leadership. We find that NCIDs appear to play an important role in the CEO transitioning process. They can help firms identify qualified inside replacements or provide stability as the new permanent CEO. In addition, NCIDs seem to facilitate the transfer of information and help the new external CEOs succeed. Supporting this, we show that the longer the NCID stays with the company, the longer the tenure of the new CEO. We also document that the presence of NCIDs improves operating and stock performance; especially when the new CEO is hired from outside of the firm. These results shed light on one important factor that can help mitigate the friction in the new CEO-firm matching process.
Our results also highlight the extent to which different firms place different needs on their boards; board composition is not and should not be identical across all firms. We offer that the call to uniformly independent boards is apt to impose costs on many firms facing a shift in executive leadership, making it difficult for them to find suitable CEO replacements and/or to facilitate the transition of an externally hired new CEO. Restrictions on board composition could pose negative consequences on firms’ governance effectiveness, especially in times of leadership change.
|Year||No. of sudden CEO departures||% of firms with NCIDs||% of firms hiring outside new CEO|
Table I presents a summary of our sample of 582 unplanned CEO departures across time. We present the number of these departures, the per cent of firms with NCIDs and the percentage of firms that replaced the departing CEO with an outsider to the firm
|Cause of unplanned departure||(%)|
|Forced turnover due to poor performance||66.2|
|Chief Financial Officer||22||13.3|
|Chief Operating Officer||34||20.6|
|Executive Vice President||30||18.2|
|Senior Vice President||16||9.7|
|Firms with NCID Continued Presence on the Board Following New CEO Arrival|
Table III presents details for the 248 NCIDs for the sample of 582 unplanned CEO departures for the years 1993-2012. Panel A contains the position held by the NCID in addition to board membership. Panel B summarizes the tenure of the NCIDs at the firm following the departure of the old CEO; Total is equal to the number of NCIDs: There are 208 firms with NCIDs. Some NCIDs hold multiple titles
Firm and Governance Summary Statistics Partitioned by NCID on Board
(NCID) = 0
(NCID) > 0
|Panel A: Firm Characteristics|
|Firm Age (years)||21.7||17||19.6||17||−0.0785||−0.1855|
|Market Cap (000)||7,624.70||1,127.30||7,602.77||1,508.80||−0.9924||(0.0814)*|
|Market to Book||1.51||0.99||1.48||0.96||−0.8568||−0.7792|
|Panel B: Corporate Governance Characteristics|
|% Inside Directors||14.10%||11.10%||24.90%||22.20%||(0.0001)***||(0.0001)***|
|Number of Inside Directors||1.2||1||2.33||2||(0.0001)***||(0.0001)***|
|Post year 2001||69.50%||100.00%||46.10%||0.00%||(0.0001)***||(0.0001)***|
|Departing CEO Founder||4.30%||0.00%||7.90%||0.00%||−0.1274||−0.1087|
Table IV presents descriptive statistics for the sample of 582 unplanned CEO departures for the years 1993-2012. We partition the data by the presence or absence of a NCID on the board of directors. Panel A contains data on firm characteristics. Panel B contains corporate governance characteristics. We present mean and median values as well as the p-value from t-tests and Kruskal–Wallis tests of means and median differences across subsamples.
***indicate significance at the 10, 5 and 1% levels, respectively
New CEO Summary Statistics Partitioned by NCID on Board
|NCID = 0||NCID > 0||t-test||Kruskal-Wallis|
|Panel A: New CEO Characteristics|
|Currently on Board||18.10%||0.00%||46.20%||0.00%||−0.0001||−0.0001|
|Panel B: Unexpected CEO Departure|
|Currently on Board||19.00%||0.00%||59.40%||100%||−0.0001||−0.0001|
Table V presents descriptive statistics for the sample of 582 unplanned CEO departures for the years 1993-2012. We partition the data by the presence or absence of a NCID on the board of directors. Panel A contains data on the new CEO for the full sample. Panel B contains data for the subset of observations we classify as unexpect CEO departures. We define unexpected CEO departures as those arising due to unexpected death and illness, and lawsuits and criminal investigations. We present mean and median values as well as the p-value from t-tests and Kruskal-Wallis tests of differences in means and medians across subsamples;
***indicate significance at the 10, 5 and 1% levels, respectively
Determinants of Inside Succession
|Panel A: Inside succession||Panel B: Inside director as the new CEO|
|NCID||0.6999*** (0.0001)||26.48||0.8542*** (0.0001)||29.76|
|Number of NCIDs||0.4076*** (0.0001)||16.02||0.4749*** (0.0001)||16.00|
|Unexpected CEO departure||0.9326*** (0.0001)||34.21||0.9407*** (0.0001)||34.38||0.4849*** (0.0001)||18.64||0.5044*** (0.0001)||17.67|
|Classified Board||−0.1373 (0.2400)||−5.41||−0.0935 (0.4198)||−3.68||−0.0130 (0.9140)||−0.43||0.0465 (0.6942)||1.57|
|Institutional Ownership||−0.1288 (0.6510)||−5.07||−0.0728 (0.7970)||−2.86||−0.0922 (0.7561)||−3.08||−0.0167 (0.9554)||−0.56|
|Post Sarbanes Oxley||0.3656*** (0.0074)||14.40||0.3319** (0.0135)||13.06||0.1948 (0.1654)||6.41||0.1602 (0.2511)||5.34|
|Founder CEO||0.1905 (0.4956)||7.36||0.1936 (0.4966)||7.46||0.3642 (0.1836)||13.15||0.377 (0.1695)||13.76|
|R&D Intensity||−1.1922 (0.2169)||−46.95||−1.1720 (0.2187)||−46.08||−0.5598 (0.5684)||−18.70||−0.623 (0.5152)||−20.99|
|Log Market Cap||0.0300 (0.4762)||1.18||0.0326 (0.4318)||1.28||0.0468 (0.2875)||1.56||0.0477 (0.2722)||1.61|
|Log Firm Age||0.1359 (0.1126)||5.35||0.1212 (0.1544)||4.76||0.1235 (0.1718)||4.12||0.0924 (0.2982)||3.11|
|Free Cash Flow||1.2828** (0.0467)||50.52||1.0485* (0.0987)||41.22||0.3778 (0.5542)||12.62||0.0653 (0.9157)||2.19|
|Log Board Size||0.3442 (0.2305)||13.55||0.2312 (0.4200)||9.09||0.0222 (0.9400)||0.74||−0.1346 (0.6484)||−4.53|
|Analyst Coverage||−0.0091 (0.5277)||−0.35||−0.0113 (0.4249)||−0.44||−0.0203 (0.1517)||−0.67||−0.0249* (0.0708)||−0.84|
|Prior Year Stock Return||−0.1581 (0.1105)||−6.22||−0.1576 (0.1101)||−6.19||0.0142 (0.8881)||0.47||0.0143 (0.8863)||0.48|
|Tobins Q||0.0117 (0.7246)||0.46||0.0146 (0.6572)||0.57||0.0345 (0.3103)||1.15||0.0379 (0.2483)||1.27|
|Board Independence||−0.2189 (0.3956)||−8.62||−0.1502 (0.5629)||−5.90||−0.1587 (0.5620)||−5.30||−0.006 (0.9831)||−0.20|
|Stock Return Volatility||3.0096 (0.4036)||118.53||2.5185 (0.4826)||99.03||1.5926 (0.6780)||53.21||0.724 (0.8518)||24.39|
|Constant||−1.8493*** (0.0070)||−1.5893** (0.0206)||−1.8782*** (0.0061)||−1.4954** (0.0312)|
|Wald Chi Square||111.46***||99.86***||12.05||73.63|
Table VI provides Probit regressions and marginal effects examining the choice of the new CEO as either a current employee of the firm (Panel A) or as an inside director of the firm (Panel B). Independent variables include a set of firm and governance characteristics, and an NCID indicator equal to one if the firm has at least one NCID, and the Number of Inside Directors on the board. Unexpected CEO departure is an indicator variable equal to one if the CEO departure is caused by unexpected death, illness or resignation caused by lawsuits or criminal investigations, and zero otherwise. See the Appendix for a list of other variable definitions. Robust standard errors are adjusted for heteroskedasticity and p-values are provided in parentheses;
***indicate statistical significance at the 10, 5 and 1% level respectively
New CEO Tenure and Post-Transition Operating Performance
|Non-CEO Inside Director = 0||Non-CEO Inside Director > 0||t-test||Wilcoxon|
|Panel A: Full Sample|
|New CEO Tenure (Years)||4.1||3.5||4.7||3.6||−0.0514*||−0.1696|
|Departure CAR [−3,+3]||−1.80%||−1.33%||−2.84%||−1.91%||−0.4902||−0.9734|
|Announcement CAR [-3,+3]||1.65%||0.35%||0.52%||0.22%||−0.3525||−0.7679|
|Chg. in Indperf Adj. ROA (t+1)||2.14%||0.78%||1.14%||0.41%||−0.515||−0.2423|
|Chg. in Indperf Adj. ROA (t+2)||5.45%||0.83%||0.65%||0.98%||−0.1375||−0.4322|
|Chg. in Indperf Adj. ROA (t+3)||6.14%||1.07%||3.19%||1.45%||−0.4288||−0.8824|
|Excess Return 12 Months||4.72%||−0.54%||−4.12%||−0.08%||−0.1771||−0.2756|
|Excess Return 24 Months||5.27%||7.81%||−13.45%||−8.90%||−0.0201**||−0.0316**|
|Excess Return 36 Months||6.87%||6.72%||−16.72%||−10.36%||−0.0163**||−0.0325**|
|Panel B: Inside Replacement|
|New CEO Tenure (years)||4||3||4.6||3.1||−0.1545||−0.5255|
|Departure CAR [-3,+3]||−0.09%||−0.86%||−2.91%||−1.80%||−0.1285||−0.4259|
|Announcement CAR [-3,+3]||0.95%||−0.19%||−0.64%||−0.66%||−0.3139||−0.6148|
|Chg. in Indperf Adj. ROA (t+1)||3.12%||0.78%||−1.07%||−0.14%||−0.0246**||−0.0288**|
|Chg. in Indperf Adj. ROA (t+2)||11.22%||1.77%||−0.64%||0.64%||−0.0533*||−0.0019***|
|Chg. in Indperf Adj. ROA (t+3)||11.08%||2.08%||1.03%||1.45%||−0.1299||−0.0982*|
|Excess Return 12 Months||5.16%||−1.92%||−9.33%||−1.49%||−0.0495**||−0.1524|
|Excess Return 24 Months||6.77%||10.08%||−24.94%||−14.87%||−0.0012***||−0.0008***|
|Excess Return 36 Months||10.24%||9.53%||−26.43%||−20.66%||−0.0019***||−0.0018***|
|Panel C: Outside Replacement|
|New CEO Tenure (years)||4.2||3.5||5||4.9||−0.0994*||−0.0899*|
|Departure CAR [-3,+3]||−3.28%||−1.72%||−2.68%||−2.62%||−0.8165||−0.4544|
|Announcement CAR [-3,+3]||2.25%||1.34%||3.05%||1.92%||−0.6806||−0.343|
|Chg. in Indperf Adj. ROA (t+1)||1.29%||0.72%||6.26%||2.10%||−0.0605*||−0.0586*|
|Chg. in Indperf Adj. ROA (t+2)||0.15%||−0.25%||3.72%||2.06%||−0.211||−0.368|
|Chg. in Indperf Adj. ROA (t+3)||1.54%||−0.69%||8.51%||1.43%||−0.0658*||−0.0644*|
|Excess Return 12 Months||4.31%||0.80%||9.83%||9.47%||−0.66||−0.6535|
|Excess Return 24 Months||3.85%||2.00%||18.44%||19.28%||−0.2811||−0.2304|
|Excess Return 36 Months||3.69%||−6.94%||12.99%||4.74%||−0.585||−0.3938|
|Panel D: Tests of differences between Panel B and Panel C (p-values)|
|New CEO Tenure (years)||−0.4468||−0.6418||−0.5067||−0.1163|
|Departure CAR [-3,+3]||−0.111||−0.2162||−0.9194||−0.6385|
|Announcement CAR [-3,+3]||−0.4163||−0.1745||-0.0295||-0.0285|
|Chg. in Indperf Adj. ROA (t+1)||−0.312||−0.4165||-0.0079||−0.1737|
|Chg. in Indperf Adj. ROA (t+2)||-0.0604||-0.006||−0.1288||−0.447|
|Chg. in Indperf Adj. ROA (t+3)||−0.1326||-0.0122||-0.0838||−0.3416|
|Excess Return 12 Months||−0.9194||−0.8853||-0.0924||−0.1217|
|Excess Return 24 Months||−0.779||−0.4976||-0.0015||-0.0024|
|Excess Return 36 Months||−0.6007||−0.2962||-0.0233||-0.0312|
Table VII presents univariate statistics for the sample of 582 unplanned CEO departures for the years 1993-2012. We partition the data by the presence or absence of a NCID on the board of directors. Panel A contains data for the full sample on new CEO tenure and operating performance. Panel B contains data for the sub-sample of firms that chose an inside replacement for the CEO and Panel C contains firms that chose an outside CEO replacement. We present mean and median values as well as the p-value (italic values) from t-tests and Kruskal–Wallis tests of differences in means and medians across subsamples. See the Appendix for a list of other variable definitions.
***indicate significance at the 10, 5 and 1% levels, respectively
Determinants of New CEO Tenure
|NCID on Board||4.0373**||0.4077||0.3783||0.5023|
|NCID Remains on Board 1 Year||3.2149***|
|NCID Remains on Board 2 Years||3.0849***|
|NCID Remains on Board 3 Years||2.9866***|
|Outside CEO Replacement||0.2831||0.2483||0.2456||0.2646|
|NCID Remains on Board (1, 2 or 3 years) * Outside CEO Replacement||−0.13||−0.4265||−0.3135||−0.6088|
|Log Board Size||1.8161**||1.8014*||1.882**||1.8799**|
|Log Firm Age||−0.204||−0.2337||−0.2191||−0.215|
|Log CEO Age||−3.577***||−3.5468***||−3.5105***||−3.547***|
|Log Delay in Replacing CEO||−0.055||−0.0421||−0.049||−0.0489|
|Wald test statistic||73.33***||105.94***||79.77***||136.66***|
Table VIII presents the results from a system of simultaneous equations examining new CEO tenure and NCID continued presence on the board follow the new CEO's arrival. Independent variables include firm and governance characteristics. NCID is an indicator equal to one if the firm has at least one NCID, and NCID1 (2, 3), an indicator equal to one if the identified NCID remains with the firm for at least one (two, three) years following the change in leadership. We control for industry and calendar time fixed effects. See the Appendix for a list of variable definitions. p-values are provided in parentheses.
***indicate statistical significance at the 10, 5 and 1% level respectively
NCIDs, Outside CEO Replacement and Operating Performance Changes
|Dependent Variable:||1 Year ROA||2 Year ROA||3 Year ROA|
|NCID * Outside CEO Replacement||0.0825**||0.1292**||0.1252*|
|Outside CEO Replacement||−0.028||−0.1501*||−0.0954*|
|Log Board Size||−0.1098***||−0.0796||−0.1778|
|Log Firm Age||0.0038||−0.0133||0.0369|
|Log CEO Age||0.0347||0.044||−0.0378|
|Log Delay in Replacing CEO||0.0034||0.0113||−0.0026|
|Wald test statistic||110.7***||79.23***||112.31***|
Table XI presents the results from two-stage Linear Regressions with Endogenous treatment effects examining the changes in industry and performance adjusted ROA in the 3 years following the change in CEO leadership. We include two indicators: (NCID) to capture the presence of an NCID on the board and (Outside CEO Replacement) to capture the decision of the board to replace the CEO with an external candidate. We also include the interaction of these two indicators. We model the presence of an NCID as endogenous. We include firm and governance related control variables which are defined in the Appendix. The p-value of Wald tests of independence are provided. Bootstrapped p-values are provided in parentheses.
***indicate statistical significance at the 10, 5 and 1% level respectively
Monthly Performance Attribution Regressions
|Intercet||Risk-Free Rate||Small-Big Mkt Cap||High-Low M/B||Adjusted
|12 Month Portfolio|
|NCID = 1||0.0124**||1.5493***||0.5848**||0.8167***||35.83|
|NCID = 0||0.0078*||1.3472***||1.0271***||0.2965*||52.52|
|NCID = 1||−0.0002||1.2017***||0.2188*||0.1934||43.62|
|NCID = 0||0.0001||1.3303***||0.6588***||0.7858***||47.77|
|24 Month Portfolio|
|NCID = 1||0.0122**||1.253***||0.531**||0.7957***||47.97|
|NCID = 0||0.0019||1.2667***||0.8025***||0.2303**||57.21|
|NCID = 1||0.0024||1.0436***||0.3513***||0.2287*||52.14|
|NCID = 0||0.0006||1.3131||0.6074||0.692||66.60|
|36 Month Portfolio|
|NCID = 1||0.0059*||1.1492***||0.6002***||0.6845***||56.66|
|NCID = 0||0.0006||1.2324***||0.8296***||0.2445**||57.72|
|NCID = 1||0.0001||1.1227***||0.4297***||0.4414*||63.75|
|NCID = 0||0.0009||1.2914***||0.5328***||0.5001*||71.18|
Table X presents the results from calendar time performance attribution regressions for portfolios of 12, 24 and 36 months following the CEO transition. We partition the estimations by (NCID=1) to capture the presence of an NCID on the board and (Outside versus Inside Replacement) to capture the decision of the board to replace the CEO with an internal (external) candidate. Heteroskedasticity-adjusted t-statistics are provided in parentheses. *, ** and *** indicate statistical significance at the 10, 5 and 1% level respectively
|1 Yr ROA||2 Yr ROA||3 Yr ROA|
|Panel A: Death & Illness Subsample|
|NCID* Outside CEO Replacement||0.081 (0.1367)||0.179** (0.0367)||0.206** (0.0294)|
|(Observations = 68)|
|Panel B: No Heir Apparent Subsample|
|NCID* Outside CEO Replacement||0.075** (0.0474)||0.130** (0.0361)||0.136* (0.0772)|
|(Observations = 526)|
|Panel C: No Heir Apparent Subsample|
|12 Month Portfolio Stock Performance||t-stat||p-value|
|24 Month Portfolio Stock Performance|
|36 Month Portfolio Stock Performance|
Table XI presents robustness results from estimations run on subsets of our dataset. In Panels A and B we use the two-stage Linear Regressions with Endogenous treatment effect models presented in Table 8. We include two indicators: (NCID) to capture the presence of an NCID on the board and Outside CEO Replacement to capture the decision of the board to replace the CEO with an external candidate. We also include the interaction of these two indicators. We model the presence of an NCID as endogenous. In Panel C we presents the results from calendar time performance attribution regressions for portfolios of 12, 24 and 36 months following the CEO transition. We only report the coefficients of interest and supress the details. p-values are provided in parentheses.
***indicate statistical significance at the 10, 5 and 1% level respectively
HP reported in a press release that Hurd resigned due to an investigation surrounding a claim of sexual harassment by a former contractor to the company. Following Hurd’s resignation, the HP board appointed Cathie Lesjak, the CFO of the firm, as interim CEO.
At the time of Hurd’s resignation, there were no non-CEO inside directors on the board. HP hired Leo Apotheker, an outsider, as the new CEO. His tenure as CEO only lasted 10 months.
Using a sample of 2,500 largest companies in the world, a 2015 PwC Strategy& CEO succession study states that the difference in median shareholder returns for companies undergoing planned versus forced CEO successions is approximately $120bln. “The cost of failed CEO succession planning”, PwC’s Strategy&. Available at: www.strategyand.pwc.com/reports/cost-failed-ceo-succession-planning
We refer to directors who are an employee of a firm an inside director. For instance, if the chief financial officer (CFO) is sitting on the board, he/she will be classified as a non-CEO inside director.
Mobbs (2013) focuses on the value added nature of talented non-CEO inside directors in the circumstances of forced CEO turnover due to poor performance.
We manually classify the causes of CEO departures and exclude planned retirements from our sample. We classify a CEO departure as planned if the incumbent CEO’s departure announcement contains key terms indicating that the departure is part of an orderly transition of power (i.e. retirement). We classify a CEO departure as unplanned if the incumbent CEO’s departure is caused by sudden events such as death and sudden illness or by forced dismissal. 8 See Taylor (2010), Worrell et al. (1986), Behn et al. (2006), Salas (2010), and Rivolta (2018).
Adams, Hermalin and Weisbach (2010) discuss the changes in board composition since the enactment of the Sarbanes Oxley Act (SOX) in 2002. Specifically they report that boards have become larger, more independent and have more committees.
Key words indicating an orderly transition of power include, but are not limited to “succession”, “succession planning”, “succession plan”, “natural transition”, “retirement age”, “retirement”, and “orderly transition of power”.
We use the same control firms in both the operating performance and stock return analyses.
We calculate three measures of firm performance: raw ROA (untabulated), Fama French 48 industry-adjusted ROA (untabulated), as well as industry-and-performance-adjusted ROA to control for mean reversion. Results are qualitatively the same across all three measures.
We use Stata etregress, which estimates an average treatment effect (ATE) and the other parameters of a linear regression model augmented with an endogenous binary-treatment variable. Estimation is by full maximum likelihood.
“Despite the promise that an outside CEO brings to many companies, considerable evidence indicates that external CEOs perform worse than internal CEOs”. Page 190, Corporate Governance Matters, 2nd edition, Larcker and Tayan (2015).
Following prior research, we classify internal competition based on whether the new CEO is promoted through an heir apparent (i.e., a single manager being groomed for succession) or a horse race (i.e., at least two internal candidates competing in a tournament) succession process (Shen and Cannella, 2002; Zhang and Rajagopalan, 2008; Behn, Dawley, Riley, and Yang, 2006; Mobbs and Raheja, 2012).
Performing a robustness replication of the calendar-time stock performance analysis on the death/illness sub-sample is not possible due to the low number of observations.
Appendix. Variable definitions
Age is the new CEO age obtained from Execucomp as of the year of CEO departure.
Analyst coverage is defined as the residual from a regression of the number of analysts covering the firm in the year prior to the CEO departure on firm size.
Announcement CAR is the cumulative market adjusted abnormal returns [−3,+3] at the first public announcement of the new CEOs’ appointment.
Board size is the number of board directors in the firm the year before the CEO turnover.
Board independence is the fraction of the board composed of independent directors.
Change in Indperf-adj ROA is the change in industry-and-performance-adjusted ROAs from one year up to three years after the new CEO appointment.
Current employee is an indicator variable equal to one if the replacement CEO has been with the hiring company for at least 2 years prior to the departure, and zero if the they are hired from outside.
Currently on board is an indicator variable equal to one if the replacement CEO has been an insider/employee in the firm for at least 2 years and as a director in the firm for at least 6 months, and zero otherwise.
Excess returns are defined as each sample firm’s cumulative market adjusted stock returns less the stock returns of the control firms, matched on industry and prior year firm performance.
Departing CEO founder is an indicator variable equal to one if the departing CEO was the founder, and zero otherwise.
Departure CAR is the cumulative market adjusted abnormal returns [-3,+3] at the first public announcement of the incumbent CEOs’ departure.
Firm age is the maximum number of years between CRSP listing age and Compustat listing age.
Firm volatilityt-1 is the standard deviation of daily stock price during the prior calendar year.
Industry-adjusted ROAt-1 is measured as a sample firm’s ROA minus the median industry ROA, using the Fama and French (1997) 48-industry classification.
Industry-and-performance-adjusted ROAt-1 is defined as each sample firm’s ROA less the ROA of a nonsample firm, matched on primary two-digit SIC industry and with the ROA within 10% in the previous year. If no firm in the same two-digit industry has a year-1 ROA within 10%, We first select the firm in the same one- digit industry, and then disregard industry and only match on year-1 ROA within 10%.
Institutional ownership is the percentage of the shares outstanding owned by institutional investors.
Interim appointment is an indicator variable equal to one if firms appointed an interim CEO, and zero otherwise.
Market value of equity is calculated using end of the year closing price of equity to multiply common stock shares outstanding.
Market to book is the market to book ratio of equity.
New CEO tenure is the new CEO’s tenure at the current firm.
NCID is an indicator variable equal to one if firms have at least one NCID, and zero otherwise.
Outside CEO replacement is an indicator variable equal to one if the new CEO has not been an employee of the firm for the past two years, and zero otherwise.
Prior year stock return is the cumulative market adjusted stock return over the year immediately preceding the turnover.
percentage of insider directors and Number of inside directors are the percentage and number of inside directors on the board.
Post year 2001 is an indicator variable that equals to 1 if the CEO turnover happens after the year 2001, and 0 otherwise.
R&D intensity is defined as the ratio of research and development expenditure to sales. We calculate R&D intensity by taking the maximum value of 0, or R&D expense from Compustat, whichever is larger, and then divide it by sales.
ROA is the operating earnings before interest and taxes (OIBDP) over total book assets (AT).
Stock return volatility is the standard deviation of daily stock price during the prior calendar year.
Tobin’s Q: (MVE + Assets - BVE - Deferred Taxes)/Assets
Total # of inside directors is the total number of inside directors on the board.
Unplanned CEO departure is an indicator variable equal to one if the CEO departure is caused by sudden death, illness or resignation caused by lawsuits or criminal investigations or forced dismissals, and zero if the CEO departure is caused by planned successions.
Unexpected CEO departure is an indicator variable equal to one if the CEO departure is caused by sudden death, illness or resignation caused by lawsuits or criminal investigations, and zero otherwise.
# of days without leadership is the number of days between the departure announcement of the incumbent CEO and the appointment of either an interim or a permanent CEO replacement.
Adams, R.B. and Ferreira, D. (2007), “A theory of friendly boards”, The Journal of Finance, Vol. 62 No. 1, pp. 217-250.
Adams, R.B., Hermalin, B.E. and Weisbach, M.S. (2010), “The role of boards of directors in corporate governance: a conceptual framework and survey”, Journal of Economic Literature, Vol. 48 No. 1, pp. 58-107.
Alam, Z., Chen, M., Ciccotello, C. and Ryan, H. (2014), “Does the location of directors matter? Information acquisition and board decisions”, Journal of Financial and Quantitative Analysis, Vol. 49 No. 1, pp. 131-164.
Barber, B.M. and Lyon, J.D. (1996), “Detecting abnormal performance: the empirical power and specification of test statistics”, Journal of Financial Economics, Vol. 41 No. 3, pp. 359-399.
Bebchuk, L. and Fried, J. (2003), “Executive compensation as an agency problem”, Journal of Economic Perspectives, Vol. 17 No. 3, pp. 71-92.
Behn, B.K., Dawley, D.D., Riley, R. and Yang, Y. (2006), “Deaths of CEOs: are delays in naming successors and insider/outsider succession associated with subsequent firm performance?”, Journal of Managerial Issues, Vol. 18, pp. 32-46.
Brickley, J. (2003), “Empirical research on CEO turnover and firm-performance: a discussion”, Journal of Accounting and Economics, Vol. 36 Nos 1/3, pp. 227-233.
Coles, J.L., Daniel, N.D. and Naveen, L. (2008), “Boards: does one size fit all?”, Journal of Financial Economics, Vol. 82, pp. 329-356.
Coles, J.L., Naveen, D. and Naveen, L. (2012), “Board advising”, Working paper.
Denis, D. and Denis, D. (1995), “Performance changes following top management dismissals”, Journal of Finance, Vol. 50 No. 4, pp. 1029-1057.
Duchin, R., Matsusaka, J.G. and Ozbas, O. (2010), “When are outside directors effective?”, Journal of Financial Economics, Vol. 96 No. 2, pp. 195-214.
Fama, E.F. and Jensen, M.C. (1983), “Separation of ownership and control”, Journal of Law and Economics, Vol. 26 No. 2, pp. 301-325.
Goldstein, M.A. and Nelling, E.F. (1999), “Market making and trading in nasdaq stocks”, The Financial Review, Vol. 34 No. 1, pp. 27-44.
Harris, M. and Raviv, A. (2008), “A theory of board control and size”, Review of Financial Studies, Vol. 21 No. 4, pp. 1797-1832.
Huang, P. (2011), “Marital pre-nups? A look at CEO severance agreements”, Working paper.
Jenter, D., Matveyev, E. and Roth, L. (2016), “Good and bad CEOs”, Working paper, London School of Economics.
Johnson, B.W., Magee, R., Nagarajan, N.J. and Newman, H.A. (1985), “An analysis of the stock price reaction to sudden executive death: implications for the managerial labor market”, Journal of Accounting and Economics, Vol. 7, pp. 151-174.
Kaplan, S.N. and Minton, B.A. (2011), How Has CEO Turnover Changed?, Working paper The University of Chicago.
Knyazeva, A., Knyazeva, D. and Masulis, R. (2013), “The supply of corporate directors and board independence”, Review of Financial Studies, Vol. 26 No. 6, pp. 1561-1605.
Mace, M.L. (1986), Directors: Myth and Reality, Harvard Business School Press, Boston, MA.
Masulis, R. and Mobbs, S. (2011), “Are all inside directors the same? Evidence from external directorship market”, The Journal of Finance, Vol. 66 No. 3, pp. 823-872.
Masulis, R., Wang, C. and Xie, F. (2012), “Globalizing the board room – the effects of foreign directors on corporate governance and firm performance”, Journal of Accounting and Economics, Vol. 53 No. 3, pp. 527-554.
Mobbs, S. (2013), “CEOs under fire: the effects of competition from inside directors on forced CEO turnover and CEO compensation”, Journal of Financial and Quantitative Analysis, Vol. 48 No. 3, pp. 669-698.
Mobbs, S. and Raheja, C. (2012), “Executive promotions: compensation, CEO influence, and CEO succession”, Journal of Corporate Finance, Vol. 18 No. 5, pp. 1337-1353.
Naveen, L. (2006), “Organizational complexity and CEO succession planning”, Journal of Financial and Quantitative Analysis, Vol. 41 No. 3, pp. 661-684.
Raheja, C. (2005), “Determinants of board size and composition: a theory of corporate boards”, Journal of Financial and Quantitative Analysis, Vol. 40 No. 2, pp. 283-306.
Rivolta, M.L. (2018), “Worth the wait? Delay in CEO succession after unplanned CEO departures”, Journal of Corporate Finance, Vol. 49, pp. 225-251.
Salas, J.M. (2010), “Entrenchment, governance and the stock price reaction to sudden executive deaths”, Journal of Business and Finance, Vol. 34, pp. 656-666.
Shen, W. and Cannella, A. (2002), “Revisiting the performance consequences of CEO succession: the impacts of successor type, post succession senior executive turnover, and departing CEO tenure”, The Academy of Management Journal, Vol. 45 No. 4, pp. 717-733.
Taylor, L.A. (2010), “Why are CEOs rarely fired? Evidence from structural estimation”, The Journal of Finance, Vol. 65 No. 6, pp. 2051-2087.
Worrell, D.L., Davidson, W.N., III, Chandy, P.R. and Garrison, S.L. (1986), “Management turnover through deaths of key executives: effects on investor wealth”, Academy of Management Journal, Vol. 29, pp. 674-694.
Yermack, D. (1996), “Higher market valuation of companies with a small board of directors”, Journal of Financial Economics, Vol. 40 No. 2, pp. 185-211.
Yermack, D. (2004), “Remuneration, retention, and reputation incentives for outside directors”, The Journal of Finance, Vol. 59 No. 5, pp. 2281-2308.
Yonkers, S.E. (2017), “Geography and the market for CEOs”, Management Science, Vol. 63 No. 3, pp. 609-630.
Zhang, Y. and Rajagopalan, N. (2008), “When the known devil is better than an unknown god: an empirical study of the antecedents and consequences of relay CEO successions”, Academy of Management, Vol. 47 No. 4, pp. 483-500.
Zhang, Y. and Rajagopalan, N. (2010), “CEO succession planning: finally at the center stage of the boardroom”, Business Horizons, Vol. 53 No. 5, pp. 455-462.
Zorn, M.L., Shropshire, C., Martin, J.A., Combs, J.G. and Ketchen, D.J. (2017), “Home alone: the effect of lone-insider boards on CEO pay, misconduct, and firm performance”, Strategic Management Journal, Vol. 38 No. 13, pp. 2623-2646.
The authors are grateful for the helpful comments and suggestions from Tracie Woidtke, Mike Ehrhardt, David Cicero, Michael Goldstein, Richard Herron, Shawn Mobbs, Jerome Taillard, Ralph Walkling, Tina Yang, Chenguang Shang, participants at the University of Tennessee Corporate Governance and Finance Department series, the speaker seminar at Babson College, the Midwest Financial Association annual meeting in 2013, the Financial Management Association annual meeting in 2014 and the 2016 European Financial Management Association meetings in Basel. The paper has also benefited from the summer research funding from the Finance Department at the University of Tennessee in 2012. A previous version of this paper circulated under the name “Board composition and organizational resilience: evidence from sudden CEO departures”.