Network Centrality and Firm Performance in Separate Board Leadership Structures.
Although information advantages may aid CEOs in pursuing higher performance for their firms, these same advantages may also result in shareholders' falling prey to managerial opportunism. Governance mechanisms are used to exert some degree of control over managerial decisions in an attempt to ensure that they are aligned with shareholders' interests. One such mechanism is the establishment of a separate board leadership structure. There appears to be a shift away from CEO duality (Krause and Semadeni, 2013) toward separate leadership structures wherein either a firm insider (e.g., executive employee) or an independent director serves as the chairman. Separate leadership structures are garnering increased scholarly attention (e.g., Krause, 2017); and a recent study finds that the influence of the chairman may rival that of the CEO (Withers and Fitza, 2017).
Krause's (2017) examination of board chair orientation reveals that, when leadership is separated, chairmen broadly exhibit either a collaborative or control orientation. The former is underpinned largely by stewardship assumptions whereas the latter is characterized by the assumptions of agency theory. The exposition presented herein does not depend on either set of assumptions for a particular firm, as this is beyond the current scope. Rather, both theories contribute to explain how information advantages enjoyed by the CEO as well as the board chair might influence firm performance.
This study empirically measures the relative informational advantage of executives by calculating the network centrality of all executives in North America. Such measures capture the size, importance, and spatial position of executives relative to all executives in the professional network. A glowing literature addresses the pros and cons of executive network centrality with regard to firm and market outcomes. Centrality may enable executive entrenchment but also decrease information asymmetries (improving the speed and quality of information) around and through more central executives. Network dynamics are therefore associated with both suboptimal governance and executive entrenchment (El-Khatib et al., 2015), as well as improved governance through informational exchange (Fracassi and Tate, 2012). Executive network centrality is found to be a negative and statistically significant determinant of return on assets, supporting previous findings of an entrenchment effect, for both chief executives and board chairs. But when the incentives of the executive and shareholders are strongly aligned, as in the case of executive chairs, centrality is shown to improve firm performance.
The remainder of the paper proceeds as follows. First, network centrality is discussed, as well as a review of findings pertaining to its influence on organizational outcomes. Arguments are presented pertaining to possible influences of executive centrality on firm performance. Next is an elaboration of how, in the context of separate leadership structures and high CEO centrality, the board chair's network centrality may also affect firm performance. Additionally, as a separate board chair's information set may be qualitatively different based on whether he or she is an independent (outside) chair versus an executive (inside) board chair, each of these is considered in turn. The research methodology is then presented, along with results, conclusions, and implications for future research.
Networks are everywhere, from infrastructure (e.g., electric grids, highways) to biodiversity (e.g., food chains, viral cascades) to the Internet (e.g., Facebook, Google Search) to economics (e.g., global trade, supply chains, word-of-mouth customer acquisition). The study of networks has a long history in mathematical t heory (Bonacich, 1972; Freeman, 1977; Proctor and Loomis, 1951; Sabidussi, 1966) and, recently, economic theory (e.g., Jackson, 2010; Newman, 2010; Padgett and Ansell, 1993). Theoretical and empirical work finds network centrality critical to the efficiency of information flows. Network centrality is a set of characteristics that describe the position of a network participant, in this case an executive, in the greater network. Executives with higher centralities are advantageously positioned in the network and may gather and transmit information with greater efficiency. Moreover, others can obtain meaningful information about and from central figures with greater ease and less cost.
Network Centrality and Information Asymmetry
Executives with higher network centrality have informational advantages relative to less central executives. They have access to better, less costly information with which to make strategic decisions. If executives' interests are aligned with those of shareholders, executives may use informational advantage to promote shareholder interests. This context is in line with stewardship assumptions (Davis et al., 1997), wherein executives attempt to maximize shareholder welfare as responsible "stewards" of the assets they control. Greater centrality enhances environmental scanning and focuses executive attention on areas critical to their firms' strategic situation, resulting in decisions that accomplish shareholders' objectives. Centrality affords efficient information flows, and is shown to benefit firms with regard to the cost of equity, cost of debt, and stock liquidity (Egginton and McCumber, 2018; Ferris et al, 2017; Fogel et al., 2017). In this context, information advantages via network centrality are utilized for the good of the firm and shareholders, and may thus be expected to enhance firm performance.
On the other hand, network centrality may enable executives to benefit themselves at the expense of shareholders, highlighting agency concerns (e.g., Fama and Jensen, 1983). If shareholder and executive interests are not aligned, information asymmetry may harm shareholders. In this context, network centrality may entrench managers, allowing greater compensation and perquisite consumption, or the pursuit of other pet projects that serve their own interests. CEO centrality has been shown to decrease merger and IPO performance while decreasing effective governance, specifically, decreasing CEO performance-turnover sensitivity (El-Khatib et al., 2015; Jandik et al, 2017). From this vantage point, network centrality may contribute to lower firm performance.
In a separate leadership structure, the chair can be viewed from a control and a collaboration perspective (Krause, 2017). A control orientation implies that the chair "views monitoring, evaluating, and disciplining the CEO to be fundamental attributes of his or her working relationship with the CEO" (Krause, 2017: 699). Centrality may enable the chair to monitor key events effectively, evaluate CEO decisions utilizing broader and deeper information, and exercise discipline utilizing both hard and soft information (Cornelli et al., 2013). A collaborative orientation, on the other hand, implies that the chair's primary roles are advice and counsel for strategic decisions and assistance in reducing CEO job demands (Hambrick et al, 2005; Krause, 2017).
Superior information may enable the chair to implement strategic controls more effectively and add valuable insight via access to greater expertise from outsiders. Information can be provided to the CEO resulting from the chair's connections with other boards and industry participants, aiding the CEO in scanning efforts and, again, reducing CEO job demands. Thus, from both a control and a collaboration perspective, a chair with high centrality may be instrumental in influencing performance over and above the CEO effect. However, as with CEOs, chair centrality may be used for the benefit of the chair, lowering firm performance. The psychological and career incentives unique to particular types of board chairs will be considered subsequently.
CEO Network Centrality and Firm Performance
More centrally networked CEOs could be more effective for a variety of reasons. If they acquire quality information pertinent to their strategic situation with relative ease, search costs are reduced and they can scan more efficiently (Daft et al., 1988; Gar g et. al, 2003). Ideally, the information capturing their attention is more relevant and precise, narrowing their field of vision to a more manageable subset of phenomena (e.g., Hambrick and Mason, 1984). Accurate interpretations of the environment pertaining to the firm could enable faster decisions with sufficient confidence and enhance performance (e.g., Baum and Wally, 2003). Firms with highly networked CEOs may acquire new customers and develop products faster, increase innovation, and implement mergers and acquisitions of higher quality. The logic underpinning these benefits is supportive of stewardship theory (Davis et al., 1997), based on the assumption that CEOs' interests are aligned with shareholders' interests. In this context, CEOs with high network centrality use their information advantage to maximize shareholders' wealth. In effect, centrality facilitates better information gathering and dissemination, allowing for increased firm performance.
On the other hand, agency assumptions support the argument that CEO centrality may lower firm performance. Notwithstanding potential benefits of fast, reliable information of value to the firm, such CEOs may not necessarily use their superior information to benefit shareholders. The separation of ownership and control virtually guarantees that the interests of principals and agents are imperfectly aligned. Informational advantage may also be employed to insulate executives from disciplinary forces. El-Khatib and colleagues (2015) find CEO network centrality associated with lower acquisition performance, concluding that "the prospects of increased consumption of pecuniary and non-pecuniary perquisites could be an important motivation for the higher frequency of value-losing takeover bids among high-centrality CEOs" (El-Khatib et al., 2015: 372). They offer evidence that such CEOs are shielded from the discipline of the market for corporate control and advantageously positioned in the executive labor market, making consumption of private benefits less costly. These findings reinforce Fama and Jensen's (1983) suggestion that powerful managers may take extra perquisites, affecting the cost structure of a firm where centrality allows the CEO to act contrary to shareholder interests.
Network benefits accruing to CEOs that run counter to shareholder interests might manifest themselves in myriad ways. The previously-mentioned benefits of CEO centrality purported to facilitate information gathering and dissemination may work to insulate CEOs from discipline and provide greater discretion, which could allow them to consume more perquisites. Acquisition performance is not equivalent to a firm's operational profitability, but opportunities for perquisite consumption afforded by centrality present the firm with non-trivial costs that erode profitability. Flickinger et al. (2016) found that CEOs of firms with performance downturns are less likely to be replaced when they are highly networked. Centrality also implies more executive experience in a greater number of organizations. Such social and human capital often results in higher compensation because of CEOs' expertise and informational advantages afforded by many contacts, which may hinder performance if those benefits are not realized. Thus, if centrality insulates CEOs from disciplinary forces, increases compensation, and allows for more perquisite consumption, then according to this view, CEO network centrality may weaken firm performance.
Given these arguments, the effect of CEO centrality on firm performance is an empirical matter. On one hand, centrality may afford the CEO the informational wherewithal to steer the firm into opportunities and use valuable connections to accomplish the firm's aims while benefiting shareholders. Alternatively, centrality may heighten the severity of agency problems such as goal incongruence, divergent risk preferences, and insulation from disciplinary forces. The result may be higher CEO pay and perquisite consumption without the requisite benefits to shareholders centrality potentially provides. These arguments lead to competing hypotheses:
H1a. CEO network centrality is positively associated with firm performance.
H1b. CEO network centrality is negatively associated with firm performance.
The Impact of Board Chair Network Centrality
Withers and Fitza (2017) demonstrate that separate board chairs substantially affect firm performance. The authors encourage research to discover a causal mechanism for chair influence. It is at this juncture that the potential impact of a well-connected board chair charged with monitoring a well-connected CEO is considered. A highly-networked chair may have the ability to influence monitoring, power distribution, strategic decision-making, and ultimately firm performance. Flickinger and colleagues (2016) employ a social status perspective in linking executives' board network embeddedness and the distribution of power between the CEO and the board chair in German companies, a context in which the CEO and chair are legally separated. Although the authors find highly-networked CEOs less likely to be replaced, they also find that CEO dismissal is more likely to result from underperformance when the chair is highly embedded via outside directorships. The authors surmise that these executives are better able to assess the CEO's performance, garner more momentum among board members for a dismissal decision should the need arise, and enjoy greater access to a rich pool of potential CEO successors, further facilitating replacement when justified.
Unlike the German context where the CEO and chair are institutionally separated, separate leadership structures in North America are voluntary. Thus, the analysis of board chair centrality's influence can be extended to a context allowing for a strong test of performance consequences in the case of separate leadership structures. The current study excludes CEO duality arrangements; hence, all of the chairs in the sample are by definition "separate" board chairs. When chairmen possess high centrality, they may promote decisions anticipated to enhance firm performance in order to positively affect their reputation in the director labor market. But similar to CEOs, they may also engage in perquisite consumption and receive higher salaries as their centrality increases. In sum, all executives may at times be opportunistic, and the current exploration concerns the degree to which network centrality influences agency problems among executive types.
It is argued that centrality enables informational advantage. A reasonable conjecture is that if board chair and shareholder goals are aligned, chairs with greater centrality are better able to enhance firm performance resultant of CEO centrality (per H1a), or mitigate the deleterious effect of CEO centrality on firm performance (per H1b). As reflected in the arguments leading to the competing hypotheses regarding the main effect of CEO network centrality, the direction of performance may be partially driven by whether executives operate primarily under agency or stewardship assumptions.
Larcker et al. (2013) found well-connected boards to be associated with higher stock returns and future growth in return on assets. They note that well-connected boards have better access to information on industry trends, regulatory changes, market data, etc., instrumental in making strategic decisions. Also, reductions in information asymmetry result from valuable social relationships and useful business and political contacts. Prominent boards may stimulate new product development as a result of their impact on internal network activities (Mazzola et al, 2016). The aforementioned benefits should accrue more to central board chairs than to those lower in network centrality.
Executive Chairs and Independent Chairs
An important distinction is that between chairs who are firm executives and those from outside the organization. A separate leadership structure implies that the CEO is not the board chair. In a non-duality situation, the chair can either be an outside or an inside director. An outsider filling the role of board chair is generally referred to as an independent chair. An inside director, on the other hand, is usually a member of the top management team (TMT), referenced herein as an executive chair. The domains of each type of chair are distinct, and explanations regarding the impact of board chair network centrality are described based on each of these domains.
An executive chair, or "insider," is commonly recruited from the top management ranks. Information advantages an executive chair accrues via high network centrality might be expected to enhance firm performance for several reasons. An insider is naturally privy to day-to-day hard and "soft" information about the nuances of the firm and its unique value proposition within its industries, and should therefore be able to contribute in-depth firm knowledge to board discussions (Baysinger and Hoskisson, 1990). A shared vision and tacit knowledge are created as a result of interactions among a firm's executives over an extended period of time (e.g., Berman et al., 2002; Fischer and Pollock, 2004). By virtue of this shared vision, a well-networked executive chair can reduce the CEO's job demands (Hambrick et al., 2005) by better managing interactions with other board members and bringing to management's attention specific information most relevant to the firm's unique situation. Moreover, participation in the TMT over an extended time period endows the executive chair with the acumen to discern the dynamics and power distribution among executives within the TMT, as well as understand the relative strengths and expertise of other firm executives. An intimate understanding of power differences enables more informed judgement about each executive's unique contributions (Finkelstein, 1992), and the information acquired through networks can be tailored to firm-specific needs. From a control point of view, deep understanding of the power dynamics and firm-specific issues permits an insider sufficient information to enforce shareholder interests, particularly when top management is entrenched (Finkelstein et al., 2009). Moreover, monetary incentives peculiar to the CEO that sometimes have dysfunctional consequences for shareholders are less likely to affect the executive board chair to the same degree (e.g., Meulbroek, 2001).
An executive chair is often an heir apparent or has aspirations of promotion elsewhere. The executive chairmen in the current sample are by definition not CEOs. A reasonable assumption is that these executives (e.g., the CFO or COO) wish to perform their best as they may have their sights set on being CEO at their current firm or another firm. If they have high network centrality, their unique combination of informational advantages and career incentives may drive the pursuit of goals that benefit their firms. Moreover, the firm-level information available to them is different from the information available to external (independent) chairs. Thus, higher centrality for non-CEO internal chairs may leverage the informational advantages they have about internal, day-to-day firm operations as well as information afforded by a more advantageous network position and power to perform their jobs at a higher level, boosting performance. He or she should therefore have incentives to "do the right thing" for shareholders by maximizing the effectiveness of the board in its dual roles of monitoring and advising.
In contrast to executive chairs, independent chairs are recruited from outside the organization. They are often believed to be vigilant in safeguarding shareholder interests (Fama and Jensen, 1983). Their independence puts them at an advantage when establishing CEO compensation aligned with shareholders' objectives (Ryan and Wiggins, 2004). They frequently have experience as executives of other firms, and may have sat on multiple boards over their careers. They tend to be aware of innovations and organizational methods employed elsewhere and bring expertise from outside to improve the strategic health of their firm (Carpenter and Westphal, 2001). They provide advice and counsel, information on external trends, and connections with important stakeholders (Hillman et al., 2000). Although less able to evaluate the finer points of strategic issues, TMT dynamics, and day-to-day evolution of strategic decisions, the broader information they possess may be richer in terms of its novelty to management. Thus, networked independent chairs offer valuable resources, but the information garnered by centrality is qualitatively different from that of executive directors. The foregoing leads to hypotheses pertaining to each distinct type of chair in the presence of highly-networked CEOs:
H2. When CEO network centrality is high, executive board chair network centrality will be positively associated with firm performance.
H3. When CEO network centrality is high, independent board chair network centrality will be positively associated with firm performance.
To capture the unique position of executives in the vast network of North American executives, four measures of centrality are calculated: degree, eigenvector, betweenness, and closeness centrality. Each measure is calculated annually as networks are dynamic. Degree centrality is the metric most commonly used in studies of social networks because of its ease in calculation and simple intuition. Degree is simply the number of connections one has to others, akin to the size of one's Rolodex or personal network. An obvious extension of degree centrality is eigenvector centrality, a measure of the influence or importance of one's network. Eigenvector centrality iteratively measures the degree centralities of one's connections; holding degree constant, one may be said to be more influential (have higher eigenvector centrality) when his or her connections also have high degree centrality. Because information flows are not linear (e.g., as with a chain of command), the analysis moves beyond direct connections to calculate betweenness and closeness centralities. Information flows are sometimes close and direct and sometimes indirect and distant, and thus spatial representations of network influence are meaningful. Betweenness centrality captures the notion of brokerage; it is the frequency with which an executive is "between" two others. For example, in a network of x-y-z, y lies between x and z and has a betweenness of 1; y sits between 100% of disconnected nodes. It is possible for x and z to garner information from the other, but y has the power to pass along, alter, or block the information. Complementarily, under the assumption that distance erodes accuracy and timeliness, closeness centrality measures the inverse of the average "distance" (in terms of steps from x to z, for example) from a node to all other nodes. All four measures of centrality offer informational advantages to those with higher centralities.
To calculate centrality variables, all executive and board appointment information from the BoardEx database from 1998-2014 was collected. Two executives are connected if they sit or have served on a board of a public, private, government, or nonprofit entity at the same time. Restricting the sample to the network of North American executives, annual networks were created, resulting in a time series of 17 networks. The 2014 network, for example, is comprised of 260,874 executives and almost 280 million current and past professional relationships among them. Betweenness, closeness, degree, and eigenvector centralities are calculated for all executives in the North American network each, and executives are ranked according to their raw scores; ranks are percentiles. For example, in 2008 a degree centrality of 291 direct connections corresponds to the 70th percentile of all executives' degree centralities that year, while a degree centrality of 3,571 corresponds to the 99th percentile. Centrality percentiles herein are representative of the entire network as opposed to only those of sample firms. Restricting the sample to non-duality firms with requisite public financial data available results in a sample of 6,064 unique firms and 25,223 executives. The sample years include 80,657 CEO-firm-year observations. Of these, the CEO is also the chairman of the board in 34,606 observations; duality occurs 42.91% of time.
Firm profitability was examined as a measure of firm performance, which is simply net income divided by firm assets (ROA). Controls included firm attributes that are common determinants of profitability, specifically cash flow (EBITDA) and leverage (total debt to assets). Because concepts of board structures, entrenchment, and agency theory are governance issues, firm governance was controlled with the Bebchuck et al. (2009) Entrenchment Index. (1) Finally, the Beneish (1999) M-score was included to control for firm financiais and reporting quality. (2) The latter controls greatly diminished the number of observations, however, so in all tests these variables are excluded and included in different models. Finally, time and industry fixed effects were included in all model specifications, clustering robust errors by firm. Summary statistics presented in Table 1 include centrality variables as well as "centrality," the first principal component, and profitability (ROA). Correlations appear in Table 2.
The first relationship examined is that between CEO centrality and firm performance. Individual centrality measures are understandably highly correlated; an executive with a high degree centrality is more likely to also have important (as measured by degree centrality) people in his or her network, and will therefore have a high eigenvector centrality. In order to measure the overall impact of centrality measures, the first principal component of the four measures was extracted, and this common element is simply called "centrality" (see Table 3). Complementarity, each variable was orthogonalized to isolate the unique contribution, if any, of each centrality variable. The analysis above for CEO "centrality" and orthogonal components was rerun. Results are reported in Table 4.
CEO network centrality is negatively associated with ROA. Hypothesis lb is therefore supported and consequently Hypothesis la is rejected, in line with agency rather than stewardship assumptions of the effects of CEO connectedness. This result reinforces the findings of El-Khatib et al. (2015) pertaining to acquisition performance and Flickinger et al. (2016) pertaining to CEO dismissal. Findings of decreases in ROA extend the logic of detrimental influences of CEO network centrality on organizational outcomes. Thus, the concerns portrayed in agency theory appear to manifest themselves in this sample of CEOs' use of privileged information to their own advantage, even to the detriment of shareholders.
Given that higher CEO centrality seems to weaken performance, the next analysis concerns the impact of chair centrality in the presence of powerful CEOs. The sample was restricted to firms whose CEOs are in the top quartile of principal component centrality (i.e., the combined measures "centrality") of sample firms. Indicator variables Powerful Executive Chair and Powerful Independent Chair were created, equal to one if the executive or independent chair is also in the top quartile of centrality of all executives in the sample. Controlling for both CEO and chair centralities, executive chair centrality in the presence of a powerful CEO is found to enhance firm profitability positively and significantly (see Table 5). In the case of executive (inside) chairs, performance appears higher when the chair is well-connected.
The effect of chair centrality in the presence of highly central CEOs was measured, holding all covariates at the means, by examining mean firm profitability when chairs are above and below the 75th percentile of all executives. Table 6 reports results. Differences in means tests indicate that firms with executive chairs in the 75th or greater percentile of centrality enjoy a 0.65% improvement in ROA over firms with a less powerful executive chair, a 13.9% improvement in profitability. No significant difference are found regarding independent chairs. The remedial impact of powerful board chairs on performance when facing powerful CEOs is limited to executive chairs, offering support for Hypothesis 2. High centrality among independent chairs was apparently not a sufficient countervailing force to mitigate performance downturns associated with CEO centrality. Hypothesis 3 is therefore not supported.
DISCUSSION AND CONCLUSION
This exploration of the benefits firms can derive from well-connected executives was begun by considering the effect of network centrality on firm performance from two perspectives. The prediction of a positive impact on performance stemming from stewardship theory runs counter to agency-theoretic assumptions of a negative influence. The arrival at competing hypotheses for CEOs' impact provides a test of centrality's influence on profitability in a broad sample of CEOs, and the negative relationship supports agency arguments. CEOs' impact on performance has been substantiated by stewardship as well as agency assumptions (e.g., Boyd, 1995). But accumulating evidence suggests little benefit for shareholders when the CEO is the sole executive with high network centrality.
A compelling possibility is that although CEOs have an incentive to improve firm performance, network centrality facilitates opportunistic behaviors. Highly central CEOs may be at their peak in terms of their management aspirations and salary levels, while also approaching the end of their careers. Their incentives are to retain their high salaries and resist undue risks that would result in their dismissal. (3) Perhaps more specific contexts representing agency versus stewardship conditions would result in different effects of CEO centrality on performance.
The influence of board chair network centrality was explored in the face of highly-networked CEOs. An integration of collaboration and control perspectives of board chair orientation (Krause, 2017) supported arguments and findings for an impact of board chair network centrality--but only for executive chairs was the impact positive, and this only in the case of highly-networked CEOs. This poses interesting questions on a number of fronts. Compared to insiders, independent chairs are often believed to be more vigilant and in a better bargaining position vis-a-vis CEOs (Fama and Jensen, 1983; Ryan and Wiggins, 2004), and yet their impact on profitability when facing highly-central CEOs was not significant. A recent study by Sauerwald et al. (2016) considers the tension between corporate elite norms versus monitoring norms, concluding that outside directors may not always strive to align the focal firms' shareholder interests with those of the CEO.
The added influence of executive (inside) chairs was statistically and economically significant. Notwithstanding concerns that insiders are often beholden to CEOs (Fredrickson et al., 1988), perhaps the contributions of a shared vision (Berman et al,, 2002; Fischer and Pollock, 2004), understanding of TMT nuances (Finkelstein, 1992), and board contributions of in-depth firm knowledge (Baysinger and Hoskisson, 1990) outweigh such concerns. Additionally, as executive chairs are not the CEO at their current firms or other firms, they are not typically paid as highly as CEOs, and may be less likely to be nearing the end of their careers. They may, therefore, have greater incentives to improve firm performance in hopes of becoming CEO at the current firm or elsewhere. Their network centrality along with insider knowledge provides them with better, broader, and deeper information with which to boost their firms' performance. Their positions in the network also increase their likelihood of promotion because they have more contacts and because centrality is the mechanism through which performance is communicated and promotions are made possible.
Flickinger et al. (2016: 1114) suggest that "... it is conceivable that high-status COBs [board chairs] do not wish to tolerate inferior performance for extended periods of time because such tolerance might tarnish their status and reputation." Again, these authors studied independent chairs in Germany where such separation is mandated, and yet in the current sample the centrality of inside chairs appears consequential to performance improvement. Perhaps the information possessed by the independent chairs is less potent for firm outcomes than the stock of information and firm-specific human capital residing in executive chairs. Insiders' sensitivity to reputation effects derived from their networks may extend to their reputation inside the firm as well, especially if they aspire to be CEO and are dependent on a variety of internal as well as external stakeholders. The current study contributes to the literature on board chair effectiveness by highlighting the benefits of executive chair centrality in the North American context.
These results offer implications for boards of directors charged with selecting CEOs and chairs. If executive board chair centrality is beneficial to performance and CEO centrality is detrimental, it appears boards should ensure that their chairs are highly connected and should encourage them to sit on other boards. Perhaps the dynamics surrounding executive centrality and performance are more nuanced, and that, some contexts may favor a balance of centrality whereas others may not. Further studies aimed at uncovering the effects of CEO and chair centrality, as well as that of the board at large, are warranted.
Several avenues for future research are suggested. CEO centrality's negative influence on profitability is at odds with the logic that information advantages should translate into performance advantages. Information advantages may create opportunities for consuming perquisites as a general rule. Such CEOs may be comfortable because they are somewhat insulated from the disciplinary forces mentioned earlier. Also, does centrality allow CEOs to rely on their networks for information to such a great extent that they use the networks as a substitute for doing sufficient research on their own? Are the data received through centrality not necessarily as accurate and timely as would be ideal? Examination of the degree to which various networks are reliable and informative could reveal "best practices" regarding the value captured from network centrality, and may extend McDonald et al.'s (2008) findings regarding governance mechanisms' influence on CEOs' advice-seeking activities.
Research efforts that attempt to reveal the effects of network centrality in specific situations characterized by agency versus stewardship conditions are encouraged. Although executive centrality was found to be negatively related to performance in general, samples based on precise specifications such as firms' life cycle stage, presence of founders, and percentage of insiders or original top managers on the board could tap into the degree to which executives personally identify with the firm and exhibit great efforts to preserve its long-term health. Discovery of such contexts would be consequential to shareholders as well as executives and their employees. If, as Davis et al. (1997) assert, the gains to be made from trust and cooperation are emphasized more than the risk of opportunism, then centrality may allow for performance maximization rather than cost minimization.
Another avenue concerns how chair centrality reduces CEO job demands (Hambrick et al., 2005). A separate board chair can manage interactions with other directors, reducing CEO distractions and increasing the focus on strategic decisions. Chair centrality might garner more relevant and current information suited to the decisions at hand. Up-to-date analysis of external environmental trends and internal operational issues can equip central chairs in safeguarding and advancing shareholders' objectives. Finally, the influence of centrality on strategic decisions and organizational outcomes may vary, in line with Boyd's (1995) finding that CEO duality may enhance or lower firm performance based on environmental contingencies. Differences in performance effects under CEO duality could provide insight into the degree to which CEO network centrality should be encouraged.
Bruce A. Walters
Professor of Management
Louisiana Tech University
Assistant Professor of Finance
Louisiana Tech University
Baum, R. J., and S. Wally. 2003. "Strategic Decision Speed and Firm Performance." Strategic Management Journal 24(11): 1107-1129.
Baysinger, B., and R. E. Hoskisson. 1990. "The Composition of Boards of Directors and Strategic Control: Effects on Corporate Strategy." Academy of Management Review 15(1): 72-87.
Bebchuck, L., A. Cohen, and A. Ferrell. 2009. "What Matters in Corporate Governance?" Review of Financial Studies 22: 783-827.
Beneish, M. 1999. "The Detection of Earnings Manipulation." Financial Analysis Journal 55:5.
Berman, S. L., J. Down, and C. W. Hill. 2002. "Tacit Knowledge as a Source of Competitive Advantage in the National Basketball Association." Academy of Management Journal 45(1): 13-31.
Bonacich, P. 1972. "Factoring and Weighting Approaches to Status Scores and Clique Identification." Journal of Mathematical Sociology 2: 113-120.
Boyd, B. K. 1995. "CEO Duality and Firm Performance: A Contingency Model." Strategic Management Journal 1(4): 301-312.
Carpenter, M. A., W. G. Sanders, and H. B. Gregersen. 2001. "Bundling Human Capital with Organizational Context: The Impact of International Assignment Experience on Multinational Firm Performance and CEO Pay." Academy of Management Journal 44(3): 493-511.
Carpenter, M. A., and J. D. Westphal. 2001. "The Strategic Context of External Network Ties: Examining the Impact of Director Appointments on Board Involvement in Strategic Decision Making." Academy of Management Journal 44(4): 639-660.
Core, J. E., W. R. Guay, and D. F. Larcker. 2003. "Executive Equity Compensation and Incentives: A Survey." Economic Policy Review 9(1): 27-50.
Cornelli, F., Z. Kominek, and A. Ljungqvist. 2013. "Monitoring Managers: Does it Matter?" The Journal of Finance 68(2): 431-481.
Daft, R. L., J. Sormunen, and D. Parks, D. 1988. "Chief Executive Scanning, Environmental Characteristics, and Company Performance: An Empirical Study." Strategic Management Journal 9(2): 123-139.
Davis, J. H., F. D. Schoorman, and L. Donaldson. 1997. "Toward a Stewardship Theory of Management." Academy of Management Review 22(1): 20-47.
Egginton, J., and W. R. McCumber. 2018. "Executive Network Centrality and Stock Liquidity." Louisiana Tech University working paper.
El-Khatib, R., K. Fogel, and T. Jandik. 2015. "CEO Network Centrality and Merger Performance." Journal of Financial Economics 116: 349-382.
Fama, E. F. 1980. "Agency Problems and the Theory of the Firm." Journal of Political Economy 88(2): 288-307.
Fama, E. F., and M. C. Jensen. 1983. "Separation of Ownership and Control." The Journal of Law and Economics 26(2): 301-325.
Ferris, S. P., D. Javakhadze, and T. Rajkovic. 2017. "The International Effect of Managerial Social Capital on the Cost of Equity." Journal of Banking and Finance 74: 69-84.
Finkelstein, S. 1992. "Power in Top Management Teams: Dimensions, Measurement, and Validation." Academy of Management Journal 35(3): 505-538.
Finkelstein, S., D. C. Hambrick, and A. A. Cannella. 2009. Strategic Leadership: Theory and Research on Executives, Top Management Teams, and Boards. Oxford, England: Oxford University Press, Inc.
Fischer, H. M., and T. G. Pollock. 2004. "Effects of Social Capital and Power on Surviving Transformational Change: The Case of Initial Public Offerings." Academy of Management Journal 47(4): 463-481.
Flickinger, M., M. Wrage, A. Tuschke, and R. Bresser. 2016. "How CEOs Protect Themselves Against Dismissal: A Social Status Perspective." Strategic Management Journal 36(6): 1107-1117.
Fogel, K., T. Jandik, and W. R. McCumber. 2017. "CFO Social Capital and the Cost of Private Debt." University of Arkansas working paper.
Fracassi, C., and G. Tate. 2012. "External Networking and Internal Firm Governance." Journal of Finance 67: 153-194.
Fredrickson, J. W., D. C. Hambrick, and S. Baumrin. 1988. "A Model of CEO Dismissal." Academy of Management Review 13(2): 255-270.
Freeman, L. 1977. "A Set of Measures of Centrality Based on Betweenness." Sociometry 40: 35-41.
Garg, V. K., B. A. Walters, and R. L. Priem. 2003. "Chief Executive Scanning Emphases, Environmental Dynamism, and Manufacturing Firm Performance." Strategic Management Journal 48(8): 725-744.
Hambrick, D. C., S. Finkelstein, and A. C. Mooney. 2005. "Executive Job Demands: New Insights for Explaining Strategic Decisions and Leader Behaviors." Academy of Management Review 30(3): 472-491.
Hambrick, D. C., and P. A. Mason. 1984. "Upper Echelons: The Organization as a Reflection of its Top Managers." Academy of Management Review 9(2): 193-206.
Hillman, A. J., A. A. Cannella, and R. I. Paetzold. 2000. "The Resource Dependence Role of Corporate Directors: Strategic Adaptation of Board Composition in Response to Environmental Change." Journal of Management Studies 37(2): 235-256.
Jackson, M. 2010. Social and Economic Networks. Princeton, NJ: Princeton University Press.
Jandik, D., T. Jandik, and W. Xu. 2017. "CEO Network Centrality and IPO Performance." University of Arkansas working paper.
Krause, R. 2017. "Being the CEO's Boss: An Examination of Board Chair Orientations." Strategic Management Journal 38: 697-713.
Krause, R., and M. Semadeni. 2013. "Apprentice, Departure, and Demotion: An Examination of the Three Types of CEO-Board Chair Separation." Academy of Management Journal 56: 805-826.
Larcker, D. F., E. C. So, and C. C. Wang. 2013. "Boardroom Centrality and Firm Performance." Journal of Accounting and Economics 55(2): 225-250.
Mazzola, E., G. Perrone, and D. S. Kamuriwo. 2016. "The Interaction Between InterFirm and Interlocking Directorate Networks on Firms' New Product Development Outcomes. "Journal of Business Research 69(2): 672-682.
McDonald, M. L., P. Khanna, and J. D. Westphal. 2008. "Getting Them to Think Outside the Circle: Corporate Governance, CEOs' External Advice Networks, and Firm Performance." Academy of Management Journal 51(3): 453-475.
Meulbroek, L. K. 2001. "The Efficiency of Equity-Linked Compensation: Understanding the Full Cost of Awarding Executive Stock Options." Financial Management 30(2): 5-44.
Mintzberg, H., D. Raisinghani, and A. Theoret. 1976. "The Structure of "Unstructured" Decision Processes." Administrative Science Quarterly 21: 246-275.
Newman, M. E.J. 2010. Networks: An Introduction. Oxford, England: Oxford University Press, Inc.
Padgett, J. F., and C. K. Ansell. 1993. "Robust Action and the Rise of the Medici, 1400\ AM." American Journal of Sociology 98: 1259-1319.
Proctor, C. H., and C. P. Loomis. 1951. "Analysis of Socioeconomic Data." In Research Methods in Social Relations. Eds. P. W. Holland and S. Leinhardt. New York, NY: Dryden Press.
Ryan, H. E., and R. A. Wiggins. 2004. "Who is in Whose Pocket? Director Compensation, Board Independence, and Barriers to Effective Monitoring." Journal of Financial Economics 73(3): 497-524.
Sabidussi, G. 1966. "The Centrality Index of a Graph." Phychometrika 31: 581-603.
Sauerwald, S., Z. J. Lin, and M. W. Peng. 2016. "Board Social Capital and Excess CEO Returns." Strategic Management Journal 37(3): 498-520.
Shleifer, A., and R. W. Vishny. 1997. "A Survey of Corporate Governance." Journal of Finance 52(2): 737-783.
Withers, M. C., and M. A Fitza. 2017. "Do Board Chairs Matter? The Influence of Board Chairs on Firm Performance." Strategic Management Journal 38(6): 1343-1355.
(1) The Bebchuck et al. (2009) Entrenchment Index gives one point for each of the following firm governance mechanisms for a range of zero to six: poison pills, classified boards, golden parachute provisions, shareholder restrictions on charter and bylaw amendments, and a supermajority requirement to approve a merger.
(2) The M-score is calculated using eight financial ratios: days sales in receivables, gross margin index, asset quality index, sales growth index, depreciation index, SSGA index, leverage index, and accruals to assets.
(3) There is voluminous literature on executive compensation and incentive alignment. See, for example, Shleifer and Vishny (1997) and Core et al. (2003).
Table 1 Summary Statistics This table presents summary statistics of executive centralities and firm variables. Centrality variables are in percentiles of all executives in North America, thus the median executive in the sample has a Betweenness Centrality greater than 81% of all North American executives. Centrality is the first principal component of the four centrality variables. Profitability is net income divided by assets (return on assets). Leverage is long term debt, inclusive of that in current liabilities, to assets. The Entrenchment Index is per Bebchuck et al. (2009), an index of firm governance quality. The M-score is per Beneish (1999), an index of reporting quality. mean Sd min 25th Betweenness Centrality 75.45 21.60 7 64 Closeness Centrality 67.46 24.47 1 50 Degree Centrality 76.18 18.19 3 63 Eigenvector Centrality 70.10 22.14 1 55 Centrality, principal component -0.01 1.89 -6.49 -1.32 Profitability (ROA) -3.36 23.29 -130.25 -1.86 log EBITDA 4.47 2.24 -5.81 2.95 Leverage 0.1771 0.2034 0.00 0.0044 Entrenchment Index 2.83 1.45 0 2 M-score -2.61 2.28 -18.32 -2.83 median 75th max Betweenness Centrality 82 92 100 Closeness Centrality 72 89 100 Degree Centrality 80 92 100 Eigenvector Centrality 75 89 100 Centrality, principal component 0.40 1.59 2.55 Profitability (ROA) 2.01 6.28 28.67 log EBITDA 4.44 5.95 11.73 Leverage 0.1066 0.2838 0.9276 Entrenchment Index 3 4 6 M-score -2.55 -2.21 5.89 Table 2 Pairwise Correlation Matrix This table presents correlation of centralities and firm control variables. Negative correlations are expressed in parentheses. P-values are reported below correlation coefficients. 1 2 3 4 1 Betweenness 1.0000 Centrality 2 Closeness 0.8563 1.0000 Centrality 0.0000 3 Degree 0.8442 0.9292 1.0000 Centrality 0.0000 0.0000 4 Eigenvector 0.7524 0.8946 0.8569 1.0000 Centrality 0.0000 0.0000 O.OOOO 5 Centrality, PC 0.9130 0.9750 0.9619 0.9273 0.0000 0.0000 0.0000 0.0000 6 Profitability (0.0211) (0.0509) 0.0030 (0.0261) (ROA) 0.0000 0.0000 0.2535 0.0000 7 log EBITDA 0.2689 0.3213 0.3835 0.3202 0.0000 0.0000 0.0000 0.0000 8 Leverage 0.0513 0.0654 0.0656 0.0615 0.0000 0.0000 0.0000 0.0000 9 Entrenchment 0.0214 0.0091 0.0359 0.0205 Index 0.0000 0.0652 0.0000 0.0000 10 M-score (0.0093) (0.0156) (0.0163) (0.0205) 0.0128 0.0000 0.0000 0.0000 5 6 7 1 Betweenness Centrality 2 Closeness Centrality 3 Degree Centrality 4 Eigenvector Centrality 5 Centrality, PC 1.0000 6 Profitability (0.0253) 1.0000 (ROA) 0.0000 7 log EBITDA 0.3425 0.1931 1.0000 0.0000 0.0000 8 Leverage 0.0646 0.0177 0.2342 0.0000 0.0000 0.0000 9 Entrenchment 0.0229 0.0173 (0.0565) Index 0.0000 0.0005 0.0000 10 M-score (0.0163) 0.1470 (0.0122) 0.0000 0.0000 0.0024 8 9 10 1 Betweenness Centrality 2 Closeness Centrality 3 Degree Centrality 4 Eigenvector Centrality 5 Centrality, PC 6 Profitability (ROA) 7 log EBITDA 8 Leverage 1.0000 9 Entrenchment (0.0022) 1.0000 Index 0.6558 10 M-score (0.0658) 0.0043 1.0000 0.0000 0.5146 Table 3 CEO Centrality and Firm Profitability This table presents results of regressions of CEO centrality and firm profitability, with controls. The dependent variable is firm return on assets (ROA), a proxy for firm profitability. Cash Flow is the log of EBITDA. Leverage is total debt to assets. The Entrenchment Index is per Bebchuck et al. (2009), an index of firm governance quality. The M-score is per Beneish (1999), an index of reporting quality. All models include industry and time fixed effects with robust standard errors clustered by firm. Significance is designated by ***, **, and * at die 1%, 5%, and 10% levels, respectively. ROA (1) (2) (3) Betweenness -0.0004 *** -0.0002 *** Centrality (0.0000) (0.0001) Closeness -0.0004 *** Centrality (0.0000) Degree Centrality Eigenvector Centrality Cash Flow 0.0133 *** 0.0167 *** 0.0137 *** (0.0005) (0.0011) (0.0005) Leverage -0.1236 *** -0.1141 *** -0.1240 *** (0.0066) (0.0115) (0.0066) Entrenchment 0.0002 Index (0.0009) M-Score 0.0047 *** (0.0010) Industry Effects Yes Yes Yes Year Effects Yes Yes Yes Observations 33,690 6,205 33,690 R-squared 0.155 0.326 0.155 ROA (4) (5) (6) Betweenness Centrality Closeness -0.0002 *** Centrality (0.0001) Degree -0.0006 *** -0.0003 *** Centrality (0.0001) (0.0001) Eigenvector Centrality Cash Flow 0.0168 *** 0.0143 *** 0.0170 *** (0.0011) (0.0006) (0.0011) Leverage -0.1140 *** -0.1243 *** -0.1141 *** (0.0115) (0.0066) (0.0115) Entrenchment 0.0001 0.0002 Index (0.0009) (0.0009) M-Score 0.0047 *** 0.0047 *** (0.0010) (0.0010) Industry Effects Yes Yes Yes Year Effects Yes Yes Yes Observations 6,205 33,690 6,205 R-squared 0.327 0.156 0.327 ROA (7) (8) Betweenness Centrality Closeness Centrality Degree Centrality Eigenvector -0.0004 *** -0.0002 *** Centrality (0.0001) (0.0001) Cash Flow 0.0134 *** 0.0167 *** (0.0006) (0.0011) Leverage -0.1239 *** -0.1145 *** (0.0066) (0.0115) Entrenchment 0.0001 Index (0.0009) M-Score 0.0047 *** (0.0010) Industry Effects Yes Yes Year Effects Yes Yes Observations 33,690 6,205 R-squared 0.154 0.326 Table 4 CEO Component Centrality and Firm Profitability This table presents results of regressions of CEO centrality and firm profitability, with controls. The dependent variable is firm return on assets (ROA), a proxy for firm profitability. Centrality is the first principal component of four centrality variables. Orthogonal centralities capture the unique element of each variable apart from the principal component. Cash Flaw is the log of EBITDA Leverage is total debt to assets. The Entrenchment Index is per Bebchuck et al. (2009), an index of firm governance quality. The M-score is per Beneish (1999), an index of reporting quality. All models include industry and time fixed effects with robust standard errors clustered by firm. Significance is designated by ***, **, and * at the 1%, 5%, and 10% levels, respectively. ROA (1) (2) Centrality -0.0054 *** -0.0028 *** (0.0006) (0.0007) Betweenness, orthogonal Closeness, orthogonal Degree, orthogonal Eigenvector, orthogonal Cash Flow 0.0138 *** 0.0169 *** (0.0006) (0.0011) Leverage -0.1239 *** -0.1140 *** (0.0066) (0.0115) Entrenchment Index 0.0001 (0.0009) M-Score 0.0047 *** (0.0010) Industry Effects Yes Yes Year Effects Yes Yes Observations 33,690 6,205 R-squared 0.156 0.327 ROA (3) (4) Centrality Betweenness, -0.0103 *** -0.0049 *** orthogonal (0.0010) (0.0014) Closeness, -0.0029 *** -0.0022 * orthogonal (0.0009) (0.0011) Degree, orthogonal -0.0056 *** -0.0018 (0.0009) (0.0013) Eigenvector, 0.0011 -0.0003 orthogonal (0.0008) (0.0011) Cash Flow 0.0143 *** 0.0170 *** (0.0006) (0.0011) Leverage -0.1244 *** -0.1141 *** (0.0066) (0.0115) Entrenchment Index 0.0002 (0.0009) M-Score 0.0047 *** (0.0010) Industry Effects Yes Yes Year Effects Yes Yes Observations 33,690 6,205 R-squared 0.157 0.327 Table 5 Chair Centrality and Powerful CEOs This table presents results of the effect of chair centrality on firm profitability when the CEO is highly central. Observations are limited to those wherein the CEO is in the top quartile of all executives in North America in Centrality, the first principal component of four measures of centrality. The dependent variable is firm return on assets (ROA), a proxy for firm profitability. CEO (Chair) Centrality is the first principal component of four measures of network centrality of the CEO (Chair). Powerful Executive (Independent) Chair is an indicator variable equal to one if the Executive (Independent) Chair is in the top quartile of all executives in North America in Centrality. All models include firm, industry and time controls with robust standard errors clustered by firm. Significance is designated by ***, **, and * at the 1%, 5%, and 10% levels, respectively. ROA (1) (2) CEO Centrality -0.1490 ** -0.0750 * (0.0628) (0.0399) Chair Centrality -0.0969 *** -0.0128 (0.0280) (0.0210) Powerful Executive 3.0961 ** Chair (1.3946) Powerful Independent -0.5200 Chair (1.0729) Firm, industry, time Yes Yes controls Observations 945 1,478 R-squared 0.384 0.353 Table 6 Margins Analysis of Chair Centrality in the Presence of Powerful CEOs This table presents margins analysis of the effect of chair centrality on firm profitability when the CEO is highly central. Observations are limited to those wherein the CEO is in the top quartile of all executives in North America in Centrality, the first principal component of four measures of centrality. Holding all other covariates at the means, the marginal impact of high chair centrality on firm profitability is reported as difference, with t-tests for differences in means designated by ***, **, and * at the 1%, 5%, and 10% levels, respectively. ROA top quartile 5.16 Executive Chair Centrality non-top quartile 4.51 difference 0.65 ** top quartile 3.39 Independent Chair Centrality non-top quartile 3.16 difference 0.23
|Printer friendly Cite/link Email Feedback|
|Author:||Walters, Bruce A.; McCumber, William|
|Publication:||Journal of Managerial Issues|
|Date:||Mar 22, 2019|
|Previous Article:||The Organizational Justice-Job Engagement Relationship: How Social Exchange and Identity Explain This Effect.|
|Next Article:||"Aren't Bosses Supposed to Manage You?" A Qualitative Study of Managing Your Boss.|