Governance in Executive Suites

45 Pages·2012·0.63 MB·English

Governance in Executive Suites

E. Han Kim and Yao Lu*

Abstract
This paper investigates how personal connections influencing governance in executive suites are
impacted by other governance mechanisms. We use the independent board requirement as an
exogenous shock reducing CEO influence in the boardroom. CEOs of the treated firms recoup the loss of
influence by increasing their influence in executive suites: The executive suites are filled with more of
current CEOs’ appointees with pre-existing social connections to the CEOs, leading to closer CEO
connectedness with other top executives. The closer connectedness seems to diminish the intended
benefits of the regulation. The improvement in the effectiveness of monitoring CEOs and shareholder
value are inversely related to the capacity to increase CEO connectedness. These findings highlight the
important role CEO connectedness in executive suites plays in determining the overall governance at the
firm level. They also demonstrate strengthening a specific governance mechanism can have spillover
effects to a seemingly unrelated governing body.

July 28, 2012

JEL classification: G34, G38, K22

Keywords: Corporate Governance, Social Connections, Monitoring CEOs, Unintended Consequences of
Regulation.

*Ross School of Business, University of Michigan, Ann Arbor, Michigan 48109 and School of Economics and Management,
Tsinghua University, Beijing, China. Emails: [email protected] and [email protected]. We have benefitted from useful
comments/suggestions from Jerry Davis, Jagadeesh Sivadasan, Denis Sosyura, Jim Westphal, Mike Weisbach, and participants at
the 2011 Ohio State University Finance Alumni Conference, the First Edwards Symposium on Corporate Governance in Canada,
and the 4th Five-Star Conference in Beijing, and seminars at the Ford School of Public Policy and the Ross School of Business at the
University of Michigan, Hong Kong Baptist University, Peking University, University of International Business and Economics,
University of Utah, and Xiamen University. We acknowledge excellent research assistance by Goudong Chen and Shinwoo Kang
and financial support from Mitsui Life Financial Research Center at the University of Michigan.
1
1. Introduction
Although a vast body of academic literature on various forms of governance exists, relatively little
attention has been paid to governance in executive suites. This is surprising because CEOs and top
executives are central in managing companies. They have substantial discretion to make decisions and to
advance their preferences, evidenced by their significant personal impacts on a wide range of firm policies
(Bertrand and Schoar, 2003; Jenter and Lewellen, 2011; Cronqvist, Makhija, and Yonker, 2012).
Two recent theoretical contributions advance our understanding of personal dynamics governing
executive suites: dissent among executives with preferences different from those of CEOs (Landier, Sraer,
and Thesmar, 2009) and internal monitoring of CEOs by non-CEO top executives (Acharya, Myers, and
Rajan, 2011). They provide important insights into how relationships and divergent self-interests among
top executives influence governance in executive suites. These insights also raise a new question. How is
governance in executive suites affected by board monitoring? We recognize both aspects of internal
governance are endogenous; as such, when one changes, the change may affect the other. Understanding
their interdependence is important because the overall governance at the firm level reflects the strength
of both.
In this paper we analyze the interplay between governance in executive suites and board
monitoring. We find an exogenous shock increasing board independence weakens governance in
executive suites. The empirical proxy for the strength of governance in executive suites is based on the
governance mechanism identified by Landier et al. (2009), wherein dissenting executives steer CEOs
towards more shareholder friendly decisions through “an efficient implementation constraint that
disciplines the decision-making process” (p.762). In their model, a CEO makes a decision, which is
implemented by an executive with preferences different from the CEO’s. The CEO anticipates the
dissenting executive will put forth more (unobservable) effort in implementing the decision if it is based
on objective information about potential profits, not on the CEO’s personal preferences. Thus, the
2
presence of more non-CEO top executives with different preferences and dissenting views strengthens
governance in executive suites.
The level of dissent in executive suites is measured by the depth of connections a current CEO has
developed with other top executives through personnel decisions. The measure is based on two metrics:
(1) the fraction of top four non-CEO executives appointed during a current CEO’s tenure (FTA), and (2) the
newly-appointed executives’ pre-existing network ties to the CEO. Most CEOs are highly engaged in
appointment decisions of their top lieutenants; hence, a newly-appointed executive is more likely to share
similar preferences with, and may be beholden to, the current CEO than executives appointed during a
previous CEO’s tenure (Landier, Sauvagnat, Sraer, and Thesmar, 2012).1 When a CEO has more of his own
appointees with whom he is connected through pre-existing network ties, his internal influence increases
through “social influence”—influence relying on norms of reciprocity, liking, and social consensus to shape
management’s decision making (Cialdini, 1984).
The exogenous shock we use to analyze the effects of a change in board monitoring is the
mandate for NYSE and NASDAQ listed firms to have a majority of independent directors by October 31,
2004. Although this regulation was promulgated around the same time as the enactment of the Sarbanes-
Oxley Act (SOX) of 2002, it is distinct and under separate purview from the SOX. Consider a firm lacking a
majority of independent directors. When it complies with the regulation, will the newly independent
board strengthen or weaken governance in executive suites? On one hand, the independent board may
intervene in executive personnel decisions and fill executive suites with more dissenting executives to
guard against the CEO making self-serving decisions. On the other hand, absent regulation, Hermalin and
Weisbach (1998) argue, board independence is determined by a bargaining process between CEOs and
boards, resulting in CEOs with greater bargaining power having more dependent boards. When the
negotiated outcome is nullified by the regulation, the CEO may attempt to recoup the loss of influence

1 Our FTA is similar, but more inclusive, than the measure used in Landier et al. (2012), which counts only new hires. We include
all new appointees, including those promoted from within the firm.
3
over the board by building a more closely aligned executive team by hiring and promoting more of “his
men”—circling the wagons against more independent directors. And he may be more successful in doing
so because he has more bargaining power against the board according to Hermalin and Weisbach.
Whether CEOs will be more or less connected to their top executives in the aftermath of the shock
depends on which governing arm has the effective control over non-CEO top executive personnel
decisions, the board or the CEO office. In a typical corporate organizational chart, the board is on the top
and has the power to appoint or dismiss CEOs; hence by extension, it may also influence personnel
decisions of other top executives. However, the board meets only a few times a year and its independent
directors work on a part-time basis. They have limited access to pertinent information and rely heavily on
management as their primary source of information (Dominguez-Martinez, Swank, and Visser, 2008;
Adams, Hermalin, and Weisbach, 2010). In contrast, CEOs work full time with employees at their disposal
to perform the necessary footwork to make their case to their boards. Thus, the CEO office has an
advantage by controlling the information channel. This advantage is especially important in personnel
decisions, which often require non-public information on why some executives are replaced, the
qualifications of their potential replacements, and the synergies each candidate can bring to the remaining
executives. Thus, actual control over top executive appointment decisions may differ from that indicated
by the organizational charts.
We find the board regulation significantly increases CEO connectedness in executive suites; more
top executives are appointed during a current CEO’s tenure and more of them are pre-connected to the
CEO. These findings are based on difference-in-differences estimation using the regulatory shock on board
independence. The variation comes from the pre-regulation board composition; firms without a majority
of independent directors prior to the regulation are the treatment group. To mitigate bias due to
differences between treatment and control groups, we use propensity-score matched samples and report
results based on both matched and unmatched samples. The evidence is robust to possible confounding
4
effects due to the SOX and other events; executive and/or CEO turnovers; heterogeneity in the degree of
the shock; or major structural changes. The results also are robust to alternative measures of CEO
connectedness and sample construction, and to outliers.
Does the closer CEO connectedness soften monitoring by the newly independent board? The
closer connectedness may help CEOs to control more effectively the information channel to the board and
put up a more united front to impede monitoring by newly independent boards. However, if the closer
CEO connectedness is the result of executive turnovers caused by newly independent boards demanding
new blood in executive suites, for example, it is unlikely to weaken the monitoring.
We investigate how CEO connectedness impacts the efficacy of monitoring CEOs by exploiting
cross-sectional variation in pre-existing conditions among treated firms. Due to differences in firm
characteristics, some firms’ optimal dissent in executive suites may be lower than others.2 For such firms,
FTA—our proxy for the inverse of dissent—will be high prior to the regulatory intervention. Hence, a
higher pre-regulation FTA implies weaker governance in executive suites at the time of the shock,
providing more targets for corrective measures by the newly independent board. A higher pre-regulation
FTA also means less capacity to increase FTA post-regulation because FTA is bounded at one, making the
regulation more binding. Thus, we predict if the closer CEO connectedness counteracts newly independent
boards, pre-regulation FTA is positively related to regulation-related improvements in monitoring of CEOs.
The proxies for the efficacy of monitoring are CEO compensation, CEO pay-for-performance sensitivity,
and the likelihood of forced CEO turnover for poor performance.
We find when pre-regulation FTA is high, CEO compensation decreases and the likelihood of CEO
dismissal for poor performance increases. In contrast, when pre-regulation FTA is low, CEO compensation
increases and the likelihood of CEO dismissal for poor performance does not change. These findings
suggest greater board independence enhances the efficacy of monitoring CEOs only when governance in

2 Landier et al. (2009) show the optimal dissent level is achieved by trading off the benefits of objective, less self-serving
decisions against the costs associated with impairing implementation efficiencies due to heterogeneous preferences.
5
executive suites is weak prior to the regulation, which limits the ability to further increase CEO
connectedness.
How do the regulatory impacts on CEO connectedness affect shareholder value? We conjecture if
closer connectedness is achieved by giving higher priority to building a more closely aligned team of
executives than to finding the best combination of experience and talent, it will hurt firm performance.
However, closer connectedness in executive suites has advantages. It may help expedite decision-making
and implementation processes by making coordination easier, resulting in more timely and efficient
outcomes. That is, the increase in CEO connectedness may be a double-edged sword with both benefits
and deleterious effects (Sah and Stiglitz, 1986, 1991).
We find a significant positive relation between pre-regulation FTA and the regulatory impact on Q.
When pre-regulation FTA is equal to one, the regulation-related changes in Q are greater, on average, by
1.042 than those for firms with zero FTA. In sum, when governance in executive suites is weak prior to the
regulation, the shock in board independence enhances the efficacy of monitoring CEOs and increases
shareholder value. But when the governance is already strong, the impacts are significantly less favorable.
This paper contributes to the literature on several fronts. It demonstrates that governance in
executive suites is endogenous and can be affected by changes in the strength of the other internal
governance mechanism—board independence. We also demonstrate the interplay between the two
internal governance mechanisms has important impacts on the overall efficacy of monitoring CEOs and on
shareholder value in ways that warrant careful consideration by governance specialists and regulators.
Our findings challenge the presumption that mandated independent boards are good for all firms.
This neither contradicts, nor supports, previous studies on how board independence is related to the
strength of board oversight or to firm performance.3 Rather, our evidence demonstrates that regulating

3 Past studies examining the relation between board independence and the strength of board oversight and/or firm
performance include Brickley and James (1987); Weisbach (1988); Rosenstein and Wyatt (1990); Byrd and Hickman (1992);
Brickley, Coles, and Terry (1994); Cotter and Zenner (1994); Borokhovich, Parrino, and Trapani (1996); Mayers, Shivdasani, and
6
board structure can have undesirable spillover effects. The evidence of harmful effects is not new; Larcker,
Ormazabal, and Taylor (2011) and Ahern and Dittmar (2012) document harmful effects of regulatory
interventions in corporate governance. Our contribution is to identify a specific channel used to
circumvent the regulation—strengthening CEO connectedness in executive suites. More broadly, our
evidence illustrates that when policy makers regulate one aspect of governance, some firms shift other
aspects of governance to circumvent the regulation. Thus, when regulators contemplate improving a
specific governance mechanism, they should not focus solely on the impact on the intended target. They
also should carefully consider indirect impacts on other governing bodies.
This study fills a void in the literature on CEOs’ influence and involvement in the selection of top
echelon players governing the firm. Previous studies examine CEOs’ influence in the selection of board
members and the outcomes (e.g., Shivdasani and Yermack, 1999; Hwang and Kim, 2009; Coles, Daniel, and
Naveen, 2011; Fracassi and Tate, 2012). We add to this literature by studying the appointment and
composition of another important group of top echelon players—top executives.
The next section describes the empirical design and data. Section 3 examines the effects of the
shock in board independence on CEO connectedness, and conducts a battery of robustness tests. Sections
4 and 5 relate the pre-regulation strength of governance in executive suites to the regulatory impact on
monitoring and shareholder value. Section 6 concludes.
2. Empirical Design and Data
2.1. Proxy for the Strength of Governance in Executive Suites
Landier et al. (2009) show the presence of more dissenting executives strengthens governance in
executive suites. Our proxy for (the inverse of) dissent is CEO connectedness with top four non-CEO
executives. CEO connectedness is measured by two metrics: (1) the fraction of top four non-CEO
executives appointed during a CEO’s tenure (FTA); and (2) pre-existing network ties the newly appointed

Smith (1997); Dahya, McConnell, and Travlos (2002); Dahya and McConnell (2007); Nguyen and Nielsen (2010); and Wintoki,
Linck, and Netter (2012).
7
executives have with the CEO. Top four non-CEO executives are identified from the ExecuComp, which
ranks executives by the sum of salaries and bonuses. To prevent changes in the reported number of
executives from affecting within-firm variation in FTA, we drop firm-year observations when ExecuComp
reports less than four non-CEO executives.4 FTA is the number of executives hired or promoted during
it
firm i’s CEO as of year t, divided by four; hence, it ranges from zero to one in increments of 0.25. We
assume the year a non-CEO executive first appears on the list of top four non-CEO executives is the year
she obtained the position. We compare this year with the year a current CEO took office to determine
whether the executive is appointed during the CEO’s tenure.
CEOs’ pre-existing network ties with top executives are obtained by manually matching individual
and company names in ExecuComp with those in BoardEx. BoardEx provides information for past
employment, education background, and membership in social organizations (e.g., philanthropic and
religious organizations, social clubs, and professional organizations). The sample period for network ties
starts from 2000 because BoardEx provides limited coverage prior to 2000. We count the number of
network ties established during overlapping years for each category of network ties (through past
employment, education, or membership in social organizations) to capture the depth of past connections.
Then we sum the three types of ties to arrive at the total number of ties. To avoid reverse causality, we
include only network ties formed prior to the executive and the CEO joining the company. Similar
measures of social connections have been used in previous papers (e.g., Cohen, Frazzini, and Malloy, 2008;
Engelberg, Gao, and Parsons, 2010; Fracassi and Tate, 2012; Duchin and Sosyura, 2012).
2.2. Empirical Methodology
We estimate the effects of the shock in board independence on governance in executive suites by
using difference-in-differences approaches. The deadline for compliance with the independent board

4Kim and Lu (2011) illustrate the importance of keeping the number of executives constant when constructing executive
variables for panel regressions with firm fixed effects. Cross-checking against proxy statements shows that missing executives in
ExecuComp are due to omission rather than to dismissal; hence, the restriction does not seem to introduce a selection bias.
8
requirement was October 31, 2004; however, many firms lacking a majority of independent directors
began to change their board structure once the recommendations were promulgated by the exchanges in
2002. The largest changes occurred in 2002 and 2003. Thus, we use 2001 as the base-year to define which
firms are affected by the regulation and 2003 as the first year of the post-regulation period. We treat 2002
as the transition period and exclude observations during that year.
2.2.1. The baseline specification
Our baseline specification is as follows:
Y = a + a + β Dep_Board2001*Post + β Post + β X + ε (1)
it i t 1 i t 2 t 3 it it
Y is a measure of firm i’s CEO connectedness with top executives as of year t, FTA or log of one plus the
it it
number of pre-existing social ties. Dep_Board2001 is the affected firm indicator, equal to one if firm i does
i
not have a majority of independent directors in 2001, and zero otherwise. This indicator is interacted with
Post, the post-regulation indicator, equal to one if year t is 2003 or thereafter. The regression includes
t
firm- and year fixed effects, a and a. Because of firm fixed effects, the regression does not contain a
i t
separate term for Dep_Board2001. X is a vector of time-varying control variables. When estimation is
i
based on an unmatched sample, standard errors are clustered at the firm-period level, where the period is
separated into pre- and post-regulation periods. Table I contains definitions of all variables.
2.2.2. Propensity-score matching
In a difference-in-differences estimation, the outcome variable of the control group is used to
calculate the expected counterfactual to control for time trend effects. It assumes the treatment and
control groups have the same time trend if there are no regulatory changes—the conditional
independence assumption (CIA). Since CIA may not be valid, our control group is identified by propensity
scores following Rosenbaum and Rubin (1983). We match firms based on information in 2001 because
whether a firm is affected or unaffected is determined by the composition of the board in 2001. Ideally,
the independent variables used to estimate the Probit model must include all factors affecting both the
9
likelihood of being affected by the regulation (board independence) and regulation outcome. Linck, Netter,
and Yang (2008) show that board independence is affected by firm complexity, costs of monitoring,
ownership incentive, and CEO characteristics. These factors are also likely to affect our measures of CEO
connectedness and the effectiveness of monitoring CEOs.
Firm complexity is captured by firm size, firm age,5 and the number of business segments within a
firm; costs of monitoring, by EBITDA/TA, Tobin’s Q, and board size; ownership incentive, by the
percentage share ownership held by a CEO; and CEO characteristics, by CEO tenure6, log of CEO age, an
indicator for CEOs hired from outside, CEOs chairing the board, and CEO gender. These are the common
matching criteria used throughout the paper. When the dependent variable is FTA or pre-existing social
ties, we add variables specifically correlated to FTA: the average tenure of top four non-CEO executives,
EXECSEN; the fraction of executives appointed during a CEO's first year in office, FTA_1Y; the fraction of
executives whose first year on the list of the top four non-CEO executives can be identified from
ExecuComp, KNOWN, which controls noise in FTA and EXECSEN arising from the ambiguity about the
precise year in which some of the top executives joined the list.7 Landier et al. (2012) find these variables
are related to the fraction of executives hired during a CEO’s tenure. We also add the fraction of top
executives who we cannot determine are appointed during a CEO's first year in office, FTA_1Y_Unknown,
which controls noise in FTA_1Y.
We employ one-to-one nearest-neighbor matching. Each unaffected firm is matched to the
nearest affected firm identified by the Mahalanobis distance metric between an affected firm and an
unaffected firm. We exclude all observations that do not satisfy the common support condition. Log
likelihood, Prob > Chi2, and Pseudo R2 for estimating the propensity scores are -243.94, 0.00, 0.15 for the

5 Boone, Field, Karpoff, and Raheja (2007) suggest that complexity increases with firm age.
6 If a CEO leaves the position and returns later, ExecuComp reports only the latest appointment date. Thus simply comparing
the CEO appointment date reported by ExecuComp with the current year may generate negative CEO tenure. We correct for
this problem by backtracking the previous appointment year using the CEO and company names.
7 If an executive is already one of the top four non-CEO executives at the firm’s first appearance in ExecuComp, we cannot
determine the year of her appointment. For such an executive, we use the year the executive joined the company as the year
she made the list. This understates FTA and overstates EXECSEN, which is why we include KNOWN as a control variable.
10