Professional Documents
Culture Documents
by Inside Succession
Author(s): Wei Shen and Albert A. Cannella Jr.
Source: The Academy of Management Journal , Dec., 2002, Vol. 45, No. 6 (Dec., 2002), pp.
1195-1206
Published by: Academy of Management
REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/3069434?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to
The Academy of Management Journal
WEI SHEN
University of Florida
stein, 1993). Further, this view implies that, because of these factors on the occurrence of CEO dismissal
of the fear of losing their jobs, the senior executives of followed by inside succession.
a firm have strong incentives to ally with its CEO and
defend their positions (Dalton & Kesner, 1985). Al-
though this managerial entrenchment argument has CEO Origin
received empirical support (Boeker & Goodstein, CEO origin refers to whether an incumbent CEO
1993; Weisbach, 1988), it neglects important interest was an employee of the firm he or she leads at the
conflicts and competition within top management time of appointment as CEO (Kesner & Sebora,
groups (Ocasio, 1994; Pfeffer, 1981). 1994). Origin has important implications for power
A primary cause of interest conflicts and compe- dynamics within top management groups. Building
tition among top executives lies in their desire for strong social networks and coalitions within a firm
power and career advancement. Senior executives is an essential task for those who aspire to be the
are ambitious individuals who have high needs for CEO (Jennings, 1971; Vancil, 1987). When they are
power and achievement (McClelland & Burnham, promoted to a firm's CEO position, inside succes-
1976). As they move up the corporate hierarchy, sors not only have the approval of outside direc-
their desire to become CEO and "run their own
tors, but also have support within the top manage-
show" becomes even stronger (Pfeffer, 1981; Van- ment group (though perhaps not complete support,
cil, 1987). The extraordinary prestige and material owing to internal competition). In contrast, when
benefits associated with the CEO title provide fur-outside CEOs take office, they lack the internal
ther incentives for senior executives to challenge social networks and coalitions of inside CEOs. Ac-
the CEO of their company and to participate incentuating
a this problem is the fact that senior ex-
power tournament in the firm's internal labor mar-ecutives from a firm's prior regime often have a
ket (Lazear, 1989; Lazear & Rosen, 1981). In addi-
hostile attitude toward outside CEOs because of the
tion, the external labor market generally evaluateschanges these outsiders may initiate (Boeker &
executives' talent on the basis of the performance of
Goodstein, 1993; Dalton & Kesner, 1985; Gouldner,
their employing firms. When firm performance suf- 1954). In addition, because outside successors are
fers, so does each senior executive's reputation andoften appointed in periods of poor performance
value in the external labor market (Fama, 1980). and are expected to turn their firms around
Senior executives thus, even if not direct power (Gabarro, 1987; Zajac, 1990), they are under pres-
contenders themselves, have incentives to monitor sure to take quick action in restructuring top man-
the CEO's leadership and join others in taking ac- agement groups (Gouldner, 1954; Helmich &
tion against the CEO when they perceive him or her Brown, 1972). This intensifies interest conflicts be-
to be less than capable. tween outside CEOs and senior executives. Thus,
The interest conflicts and competition between a compared to inside succession, outside succession
CEO and other senior executives put the CEO at increases tension within a top management group
risk of power contests with senior executives and places the outside CEO at a higher risk of
(Lazear, 1989; Ocasio, 1994). Although such con- power contests with senior executives. The high
tests are not easily observable to outsiders, we ar- expectations the board and other stakeholders have
gue that the power dynamics within top manage- toward outside CEOs make them more vulnerable
ment will affect the occurrence of CEO dismissal as
when challenges from senior executives emerge
well as choices of successors. When senior execu-
(Puffer & Weintrop, 1991). All else being equal, we
tives successfully challenge a CEO and gain the expected a higher rate of CEO dismissal followed
support of outside directors, the CEO will be dis-by inside succession among outside CEOs.
missed, and one of the contending executives will
likely be promoted as the successor (Pfeffer, 1981; Hypothesis 1. CEO origin as an outsider in-
Sonnenfeld, 1988). This scenario is very different creases the likelihood of CEO dismissal fol-
from the one widely discussed in much of the gov- lowed by inside succession.
ernance literature, in which outside directors initi-
ate a power contest with a CEO or an entire topCEO Tenure
management group in the context of poor perfor-
mance and subsequently appoint an outsider as the New CEOs confront significant challenges upon
successor (Boeker & Goodstein, 1993; Cannella & taking office. Promotion to the CEO position typi-
Lubatkin, 1993; Weisbach, 1988). We believe that cally leads to significant changes in both an exec-
four factors particularly reflect the power dynamics utive's responsibility and task environment (Kotter,
that arise between CEOs and senior executives. Be- 1982). New CEOs must adjust to their new roles
low, we develop hypotheses regarding the impacts and quickly develop good working relationships
followed by inside succession among firms with Journal and News Wires databases. Data on execu-
higher proportions of non-CEO inside directors. tive and director ownership were collected from
proxy statements. Data on firm size and perfor-
Hypothesis 3. The proportion of non-CEO in-
mance were gathered from COMPUSTAT and
side directors increases the likelihood of CEO CRSP.
dismissal followed by inside succession.
Measures
Non-CEO Executive Ownership
CEO dismissal followed by inside succession.
As an important source of power (Finkelstein,
Identifying CEO dismissals is a major challenge in
1992), stock ownership will affect both the incen-
succession and governance research because firms
tives senior executives have to challenge their CEO
seldom fully disclose the true reasons behind CEO
and their credibility in doing so. Stock ownership
resignations (Denis & Denis, 1995; Fredrickson,
by senior executives does not typically provide
Hambrick, & Baumrin, 1988; Weisbach, 1988).
enough voting power to remove a CEO, but it in-
Given their importance in our study, we took two
creases the credibility of these executives' concerns
approaches to identify CEO dismissals. Our first
about the CEO's leadership and helps them gain
approach relied on news reports (Parrino, 1997).
support from outside directors. Further, senior ex-
We searched Dow Jones databases to collect all
ecutives' stock ownership increases their influence
reports about each selected firm's CEO during a
in the selection of an inside successor following the
three-year period, from one year before to one year
dismissal (Boeker & Goodstein, 1993). In addition,
after the year of in which the CEO was succeeded.
stock ownership gives senior executives extra fi-
nancial incentives to monitor the CEO and thus There are 340 successions in our sample, and we
found news reports for 334 of these events. Of these
weakens the CEO's influence over them. Corporate
334, we first excluded successions that were con-
governance scholars have emphasized using stock
sequences of a CEO's death or of clear health issues
ownership to bond managerial actions to share-
holder interests (Walsh & Seward, 1990). Stock
(14 cases), of a CEO's acceptance of a similar posi-
tion at another firm (16 cases), or of a merger or
ownership by executives has been reported to re-
acquisition (3 cases). We also excluded from our
duce agency problems in corporate restructuring
analysis four CEOs who were reported as serving
(Johnson, Hoskisson, & Hitt, 1993). All else being on an interim basis.
equal, we expected a higher rate of CEO dismissal
Next, we analyzed the news reports and used the
followed by inside succession among firms with
following three criteria to identify CEO dismissals.
higher levels of non-CEO executive ownership.
First, a CEO was directly reported as having been
Hypothesis 4. Non-CEO executive ownership fired or forced out (11 cases). Second, a CEO was
increases the likelihood of CEO dismissal fol- reported as having resigned unexpectedly or imme-
lowed by inside succession. diately, owing to poor performance, undisclosed
personal reasons, or a desire "to pursue other inter-
METHODS ests" (35 cases). Lastly, we included each case in
which a CEO was reported as taking early retire-
ment and there was discussion of performance
The population for this study is large, publicly
traded U.S. corporations. We initially selected a
problems (18 cases) as a dismissal. Although a CEO
random sample of 512 publicly traded firms report-
may step down voluntarily when performance de-
clines,
ing at least $200 million in sales for 1988. After we our analysis of news reports suggests that
deleted firms with missing data, a total of 387 this
firmsis rare. Further, a CEO may not have any option
remained, comprising the final sample used in other
the than stepping down when performance pres-
sure is very high. Thus, we classified a CEO turn-
analysis. A t-test indicated no significant difference
in annual sales between the final sample andover
theas a dismissal if it met any one of the above
three criteria. We identified 64 dismissals in our
initial sample. Data were collected annually from
multiple sources for the years 1988 through 1997. A on the basis of the news report analysis.
sample
primary data source for CEO turnover, successor
In our second approach, we used both CEO age
selection, and board structure is the officer and
and continuity as a board member at the time of
director list provided in each firm's annual report
turnover to identify CEO dismissals. Because many
to shareholders. Additional descriptive data on firms have retirement policies, some previous re-
CEOs and their successors were collected from Dun search has relied on CEO age to separate dismissals
& Bradstreet's Reference Book of Corporate Man- from retirements (Ocasio, 1994; Puffer & Weintrop,
agement and Dow Jones Interactive's Wall Street 1991). The age of 64 is often used as a cutoff point
in making the decision: CEOs who leave office be- years of CEO tenure were also measured as a
fore the age of 64 are treated as dismissed, anddummy variable. In a qualitative study, Gabarro
those who depart later are treated as retired. An (1987) suggested that the process of CEOs' estab-
examination of our sample suggested that relying lishing authority in office generally takes from 30 to
on CEO age alone to classify a departure as a dis-48 months. Drawing on his study and our prelimi-
nary analysis, we selected five years of tenure as
missal or a retirement was too crude for our study.
Many CEOs in our sample relinquished the CEOthe cutoff in our coding. Early years of tenure was
title at the age of 62 or 63 but kept directorship orcoded 1 for the first five years of a CEO's tenure and
0 otherwise. Proportion of non-CEO inside direc-
the title of chairman of the board until the age of 64
or 65. These CEOs can hardly be treated as dis-tors was calculated as the number of directors who
missed because it is unreasonable to keep a dis-were current executives of a firm, excluding the
missed CEO on a board (Lorsch & MacIver, 1989). CEO, divided by the total number of directors (Dai-
Given this, we used continuity as a board member ly, Johnson, & Dalton, 1999). Non-CEO executive
as a second criterion, in addition to CEO age, to ownership was calculated as the proportion of a
identify dismissal. A CEO's departure from a cor- firm's outstanding shares held by all executives
poration was classified as a dismissal when the other than the CEO who were listed in the compa-
incumbent terminated his or her service as both ny's proxy statement.
CEO and board member before the age of 64 for Control variables. Firm performance impor-
reasons other than death or health, taking a similar
tantly influences the power dynamics at the top of
a firm in that poor performance weakens a CEO's
position at another firm, or a merger or acquisition.
We identified 75 dismissals using this approach.credibility and increases the likelihood of a power
contest waged against the CEO (Boeker & Good-
We compared the CEO dismissals identified un-
stein, 1993; Ocasio, 1994). However, because out-
der the above two approaches and found that the 75
dismissals identified with the second approach side directors rely more heavily on financial per-
(age and continuity observations) included all 64 formance to assess a CEO than do senior executives
dismissals identified with the first approach (news(Baysinger & Hoskisson, 1990), challenges to the
reports analysis). In other words, all 64 CEOs clas-CEO under conditions of poor performance may be
sified as dismissed on the basis of news reports left
initiated by outside directors rather than by senior
office before the age of 64 and simultaneously ter- executives. Given this concern, we did not develop
minated their directorships. This finding gave us hypotheses directly linking firm performance to
great confidence in the 64 dismissals identified viapower dynamics within top management and their
news reports. We then reexamined the news re- impact on CEO dismissal followed by inside suc-
ports on the 11 nonoverlapping cases. Eight of
cession. Instead, because firm performance is
these were reported as normal retirements in accor- widely believed to reflect a CEO's leadership abil-
dance with company policy; 2 were reported asity (Puffer & Weintrop, 1991; Weisbach, 1988), we
early retirements occurring after corporate restruc-included it as a control variable representing CEO
turing; and 1 was reported as an early retirement competence. We first chose an accounting measure
resulting from a need for restructuring. We elected of performance, return on assets (ROA), and calcu-
lated it as income before extraordinary items and
to add only the last case to the dismissals. Thus, we
identified a total of 65 CEO dismissals. We then discontinued operations divided by net assets. To
created a dummy variable-dismissal, insider-tocontrol for industry effects (Dess, Ireland, & Hitt,
measure CEO dismissal followed by inside succes- 1990), following Edwards's (1995) recommenda-
sion. It was coded 1 when a current executive of a tions, we included industry ROA in our analysis
firm became CEO after the predecessor's dismissaland measured it as the average ROA in the firms in
and 0 otherwise. There are 38 such events in our a focal company's primary two-digit SIC category.
sample. Of these, 36 involved successors whoWe also selected a market-based measure of perfor-
joined their firms before their dismissed predeces-mance, shareholder returns, and measured it by
sors had been appointed as CEO, and only 2 in-compounding the daily returns to shareholders re-
volved successors who had joined their firms dur-ported in the relevant CRSP tapes over each fiscal
ing their dismissed predecessors' tenures as CEO.year. To further distinguish the impact of power
Predictor variables. CEO origin as an outsiderfrom that of a CEO's human capital, we controlled
was also measured as a dummy variable. In linefor CEO's industry experience. Industry experience
with the discussion presented above, outsider CEOwas coded 1 when a CEO had more than five years
was coded 1 if a successor CEO was not an em- industry experience at the time of his or her ap-
ployee of the firm in which he or she took officepointment
at and 0 otherwise.
the time of succession, and 0 otherwise. The earlyThe corporate governance literature highlights
the view that boards of directors, and particularly from the perspective of the CEO who is being suc-
their outside members, play an important role in ceeded (Boeker, 1992; Ocasio, 1994). CEO tenure
CEO succession decisions (Walsh & Seward, 1990). thus was the duration measure in our event history
Two variables, the proportion of outside directors analysis. To permit annual updating of the time-
and outside director stock ownership, have been varying covariates, we divided CEO tenure into
extensively studied in previous research (Boeker & fiscal years (Ocasio, 1994, 1999). In our sample,
Goodstein, 1993; Johnson et al., 1993; Lorsch & CEOs appointed before 1988 were subjected to left
MacIver, 1989; Ocasio, 1994; Weisbach, 1988). Be- truncation. To reduce the risk of an upward bias in
cause we included the proportion of non-CEO in- the estimates, we collected tenure information for
side directors as a predictor, we chose to include all the left-truncated CEOs and included all CEO-
only outside director stock ownership as a control. years, beginning in 1988, in our analysis (Tuma &
This measure was calculated as the proportion of a Hannan, 1984). This procedure also reduced loss of
firm's outstanding shares owned by directors who information and increased the statistical power of
were not current or past employees of the firm. the analysis.
To control for the impact of CEO power, we col- A critical decision in event history analysis is to
lected information about CEO duality (Finkelstein choosing an appropriate parametric model for the
& D'Aveni, 1994) and CEO stock ownership data (Yamaguchi, 1991). We started with the gen-
(Boeker, 1992). Duality was measured as a dummy eralized gamma model. The survival and density
variable with a value of 1 when a CEO also held the functions for the three-parameter generalized
title of chairman of the board and a value of 0 gamma model are
otherwise. CEO stock ownership was calculated as
the proportion of the firm's outstanding shares S(t) = 1 - I{k, k exp[z/(kl/2)]},
owned by the CEO. Finally, firm size may influence
the power dynamics within top management fAt)
be-= [Ik/F (k-2)] [k-2]l/k*kexp[k-2(kz - ekz)],
cause the CEO succession process at a large corpo- if k 0; or
ration may have become formalized or institution-
alized (Ocasio, 1999; Vancil, 1987). We measured f(t) = 1/[(2 r) 1/2] exp[ - 2/2z]if k = 0,
firm size as the natural logarithm of total sales and
included it as our last control variable. where I (k, a) is the incomplete gamma function
and z = [In (t) - A]/a. This model is implemented
by parameterizing Aj = Xjf and treating the k and cr
Analytical Methodology
as ancillary parameters to be estimated from the
In response to frequent calls for longitudinal data (Stata, 1999: 440). The hazard function of the
studies in succession research (e.g., Datta & Rajago- generalized gamma distribution is extremely flexi-
palan, 1998; Finkelstein & Hambrick, 1996), weble, allowing for a large number of possible shapes,
selected a continuous-time event history analysisincluding as special cases the Weibull distribution
to test our hypotheses. Event history analysis is an when k = 1, the exponential distribution when k =
established methodology for analyzing dynamic 1 and a = 1, and the "log-normal" distribution
process when the outcome of that process is a dis- when k = 0. We then estimated other functional
crete event (Tuma & Hannan, 1984). It was appro-forms, including Gompertz, Weibull, log-normal,
priate for our study of the effects of the powerexponential, and "log-logistic" models. Our analy-
dynamics within top management on CEO dis-sis suggested that the generalized gamma model
missal followed by inside succession because it provided the best fit and had the largest log-likeli-
takes into account the effect of time (that is, CEO hood with the smallest value on the Akaike infor-
tenure) in estimation. We conducted our analysis mation criterion (AIC; Akaike, 1974).
using the software package Stata, which provides Previous research has consistently demonstrated
maximum likelihood estimation of both monotonic that CEO dismissal is frequently the outcome of a
and nonmonotonic parametric models of duration power struggle between a CEO and outside direc-
dependence (Stata, 1999). Further, Stata allows es- tors (Boeker, 1992; Boeker & Goodstein, 1993; Can-
timation of survival-time models with time-varyingnella & Lubatkin, 1993). To test our theory and
covariates and takes "right censoring" of data (loss convincingly demonstrate that power dynamics
within top management groups simultaneously in-
of data on events occurring after the end of a study
period) into account by using the information pro-fluence CEO dismissal and subsequent inside suc-
vided by the cumulative survival time of censored cession, we conducted analyses not only on CEO
dismissal followed by inside succession, but also
subjects. Although succession is a repeatable event
from the perspective of a firm, it is nonrepeatable on CEO dismissal followed by outside succession
and CEO dismissal in general. The purpose was to itive impacts, and firm performance and CEO own-
examine the similarities as well as the differencesership both display negative impacts. Neither the
between the factors influencing CEO dismissal fol- proportion of non-CEO inside directors nor non-
lowed by inside succession and those influencing CEO executive ownership is significant.
CEO dismissal followed by outside succession. By
comparing the results, we hoped to be able to better
understand how the power dynamics within top DISCUSSION
management groups contribute to CEO dismissal,
and the limits to that influence. This study helps advance understanding of the
role of senior executives in CEO dismissal. Previ-
a Pooled data, with n = 2,965, are shown. Correlation coefficients greater than .04 or l
Dismissal Followed by Inside Succession Dismissal Followed by Outside Succession All Dismissals Co
Outsider CEO 0.42* (0.21) 0.04 ( 0.20) 0.15 (0.29) 0.21 (0.28) 0.68** (0.26) 0.63*
Early years of tenure 1.07*** (0.23) 0.82*** ( 0.21) 1.03*** (0.17) 1.03*** (0.16) 0.97*** (0.18) 0.99
Proportion of non-CEO inside directors 5.75*** (1.66) 4.46*** ( 1.35) -0.64 (0.78) -1.03 (0.77) 1.54 (0.99)
Non-CEO executive ownership 0.98* (0.63) 1.98** ( 0.74) -2.70 (2.34) -2.60 (2.26) 0.17 (0.94) 0
Firm ROA -1.78t (0.96) -4.22** (1.57) -2.21* (1.03)
Shareholder returns -1.56*** ( 0.45) ' -1.62*** (0.43) -1.78***
Industry ROA -0.15 (2.82) 0.83 (5.18) 5.68 (4.89)
Industry experience 1.98 (8.32) 1.80 ( 4.87) -0.47 (0.66) 0.21 (0.75) 0.74 (0.70) 0.8
Outside director ownership -4.33*** (1.11) -4.93*** ( 0.95) 0.45 (1.06) 0.38 (0.90) -1.02 (1.03) 0
CEO duality 0.11 (0.21) 0.41** (0.14) -0.44* (0.20) -0.35* (0.18) -0.16 (0.22) -0.2
CEO ownership -25.28** (9.22) -34.68** (13.40) -1.55 (1.63) -1.32 (1.51) -6.34** (2.26) -5.79
Firm size -0.20*** (0.03) -0.21*** ( 0.04) 0.01 (0.06) -0.01 (0.06) -0.08 (0.05) -0.0
Constant -0.86* (0.42) -3.68*** ( 0.74) -3.06*** (0.59) -3.18*** (0.49) -2.71*** (0.69) -3.32*
Log-likelihood -131.45 -128.29 -98.39 -91.82 -211.96 -200.
Model chi-square 60.85*** 65.97*** 36.00*** 48.39*** 54.16*** 78.72
a Standard errors are in parentheses; t-tests were one-tailed for hypothesized eff
p < .10
* p < .05
** p < .01
*** p < .001
can limit the CEO's influence and increase execu- succession. This finding corroborates the proposi-
tives' incentives and ability to challenge the CEO. tion that CEOs are in weak positions and are at high
A competing explanation for our results is that risk of power contests in the early years of their
the positive impact of non-CEO inside directors on tenures (Ocasio, 1994).
dismissal followed by inside succession does not The two control variables for aspects of CEO
reflect the power dynamics occurring between a
power provide additional clues about the power
CEO and senior executives. Instead, outside direc- dynamics at the tops of firms. CEO ownership
tors may have taken the lead in dismissing the CEO shows a negative impact on dismissal followed by
and subsequently picked an inside successor be- inside succession, but it shows no impact on dis-
cause there were sufficient inside candidates avail- missal followed by outside succession. This result
able. This alternative explanation has some appeal. suggests that stock ownership increases a CEO's
However, if this scenario prevailed, we would ob- influence within top management. CEO duality
serve a significant, negative association between shows a negative impact on dismissal followed by
the proportion of non-CEO inside directors and outside succession and some positive impact on
dismissal followed by outside succession, rather dismissal followed by inside succession. This pat-
than the insignificant relationship found in this tern suggests that although being chairman of the
study. Similarly, without reference to power chal- board increases a CEO's influence over outside di-
lenges to a CEO by senior executives, it is difficult rectors, it may also increase the CEO's vulnerability
to explain the positive impact of non-CEO execu- to challenges from within the top management
tive ownership on dismissal followed by inside group, because duality makes it difficult for the
succession and its insignificant impact on dis-CEO to blame others for low performance (Harri-
missal followed by outside succession. We thus son, Torres, & Kukalis, 1988). Thus, we surmise
believe that the above results support our proposi- that stock ownership and duality play different
tion that senior executives are potential power con- roles in a CEO's dealing with senior executives and
tenders and play an important role in CEO dis- outside directors. Finally, the negative impact of
missal followed by inside succession. As to the firm size suggests that power contests within top
insignificant impact of senior executives on dis- management are more frequent in smaller firms.
missal followed by outside succession, our expla-
nation is twofold. First, because their careers and
Limitations and Directions for Future Research
personal wealth are at stake, senior executives may
not resist outside succession if it is the best choice This study advances understanding of the role of
for their firm. Second, and more likely, senior ex-senior executives in CEO dismissal, but it has lim-
ecutives may simply not have much influence itations. First, like most researchers taking a power
when outside directors take the lead in CEO dis- perspective, we did not directly observe the power
missal and successor selection. dynamics leading to CEO dismissal discussed in
Our results about the impact of outsider CEO andour theory. Instead, we examined the antecedents
early CEO tenure are also informative. The positiveand outcomes of that process. Although the results
impact of outsider CEO on dismissal followed by support our power-based arguments, it would be
inside succession disappears when shareholder re- desirable to have field research to corroborate the
turns replace ROA as the performance measure. findings of this study. Second, our identification of
This result and the significant, negative impact ofCEO dismissals was not based on firsthand infor-
shareholder returns in model 4 together suggest mation and may include some errors despite our
that the higher risk of CEO dismissal followed by efforts to avoid them. Third, given the focus of our
inside succession faced by outsider CEOs is primar-study and constraints on the availability of finan-
ily a consequence of their firms' low performance cial information, we divided CEO tenure into fiscal
in the financial market. However, when all CEO years. Fiscal year may be a relatively coarse mea-
dismissals are examined together, we still find a sure, despite its frequent use as a metric in organi-
higher risk of dismissal among outsider CEOs even zational research. A finer measure of CEO tenure
after controlling for firm accounting and market may be needed to facilitate closer examination of
performance and CEO industry experience. We the power dynamics within top management dur-
speculate that this higher risk of dismissal for out- ing the early years of a CEO's tenure. Such a finer
sider CEOs is a result of their lacking social net- measure would be especially useful to test for the
works and power bases within their new firms. existence of the "honeymoon" between a new CEO
Early CEO tenure shows a consistently strong and and a board that is often discussed in the business
positive impact on both dismissal followed by in- press.
side succession and dismissal followed by outside The internal monitoring argument of corporate
governance suggests that inside directors challengeDaily, C. M., Johnson, J. L., & Dalton, D. R. 1999. On the
a CEO when firm performance declines and the measurements of board composition: Poor consis-
CEO's capability is questioned (Fama, 1980; Oca- tency and a serious mismatch of theory and opera-
sio, 1994). In contrast, our results show that CEO tionalization. Decision Science, 30: 83-106.
dismissal followed by inside succession does not Dalton, D. R., & Kesner, I. F. 1985. Organizational perfor-
always derive from a CEO's lack of capability in mance as an antecedent of inside/outside chief ex-
that our power-related predictors displayed signif- ecutive succession: An empirical assessment. Acad-
icant main effects after we controlled for firm per- emy of Management Journal, 28: 749-762.
formance and CEO industry experience in analyses.Datta, D. K., & Rajagopalan, N. 1998. Industry structure
This finding suggests that outside directors need to and CEO characteristics: An empirical study of suc-
closely monitor the power dynamics within top cession events. Strategic Management Journal, 19:
management and make sure that the competition is 833-852.
under control and contributes to firm operations. Denis, D. J., & Denis, D. K. 1995. Performance changes
For management researchers, it will be important to following top management dismissals. Journal of
examine the interaction between CEO capability Finance, 50: 1029-1057.
and senior executive power in future research to
Dess, G. G., Ireland, R. D., & Hitt, M. A. 1990. Industry
fully identify the causal mechanisms underlying
effects and strategic management research. Journal
CEO dismissals. In addition, by demonstrating the
of Management, 16: 7-27.
effects of power dynamics within top management
Edwards, J. R. 1995. Alternatives to difference scores as
on CEO dismissal, our study suggests that there are
important differences among inside CEOs. To bet- dependent variables in the study of congruence in
ter understand the performance consequences of organizational research. Organizational Behavior
and Human Decision Processes, 64: 307-324.
inside succession, it is necessary to separate inside
CEOs appointed following their predecessors' dis- Fama, E. F. 1980. Agency problems and the theory of the
missal from those appointed following their prede- firm. Journal of Political Economy, 88: 288-307.
cessors' ordinary retirement (Shen & Cannella, Finkelstein, S. 1992. Power in top management teams:
2002). This implication may be equally applicable Dimensions, measurement, and validation. Acad-
to the study of the performance impact of outside emy of Management Journal, 35: 505-538.
CEO succession.
Finkelstein, S., & D'Aveni, R. A. 1994. CEO duality as a
double-edged sword: How boards of directors bal-
REFERENCES
ance entrenchment avoidance and unity of com-
Akaike, H. 1974. A new look at the statistical model mand. Academy of Management Journal, 37:
1079-1108.
identification. IEEE Transaction and Automatic
Control, AC-19: 716-723. Finkelstein, S., & Hambrick, D. C. 1996. Strategic lead-
Alderfer, C. P. 1986. The invisible director on corporate ership. New York: West.
boards. Harvard Business Review, 64(6): 38-52. Fredrickson, J. W., Hambrick, D. C., & Baumrin, S. 1988.
Barnard, C. I. 1938. The functions of the executive. A model of CEO turnover. Academy of Manage-
Cambridge, MA: Harvard University Press. ment Review, 13: 255-270.
Baysinger, B. D., & Hoskisson, R. E. 1990. The composi- Friedman, S., & Singh, H. 1989. CEO succession and
tion of board of directors and strategic control: Ef- stockholder reaction: The influence of organiza-
fects on corporate strategy. Academy of Manage- tional context and event content. Academy of Man-
ment Review, 15: 72-87. agement Journal, 32: 718-744.
Boeker, W. 1992. Power and managerial turnover: Scape- Gabarro, J. J. 1987. The dynamics of taking charge.
goating at the top. Administrative Science Quar- Boston: Harvard Business School Press.
terly, 37: 400-421.
Gouldner, A. W. 1954. Patterns of industrial bureau-
Boeker, W., & Goodstein, J. 1993. Performance and suc- cracy. Glencoe, IL: Free Press.
cessor choice: The moderating effects of governance
Hambrick, D. C., & Fukutomi, G. D. S. 1991. The seasons
and ownership. Academy of Management Journal,
36: 172-186. of a CEO's tenure. Academy of Management Re-
view, 16: 719-742.
Cannella, A. A., Jr., & Lubatkin, M. H. 1993. Succession
Harrison, J. R., Torres, D. L., & Kukalis, S. 1988. The
as a sociopolitical process. Academy of Manage-
ment Journal, 36: 763-793. changing of the guard: Turnover and structural
change in the top management positions. Adminis-
Cannella, A. A., Jr., & Shen, W. 2001. So close and yet so
trative Science Quarterly, 33: 211-232.
far: Promotion versus exit for CEO heirs apparent.
Academy of Management Journal, 44: 252-270. Helmich, D. L., & Brow, W. B. 1972. Successor type and
organizational change in corporate enterprise. Ad- pacts of successor type, postsuccession senior exec-
ministrative Science Quarterly, 17: 371-381. utive turnover, and departing CEO tenure. Academy
Jennings, E. E. 1971. Routes to the executive suite. New of Management Journal, 45: 717-734.
York: McGraw-Hill. Sonnenfeld, J. A. 1988 The hero's farewell. New York:
Johnson, R. A., Hoskisson, R. E., & Hitt, M. A. 1993. Oxford University Press.
Board of director involvement in restructuring: Stata. 1999. Stata reference manual, release 6, vol. 3.
The effects of board versus managerial controls and College Station, TX: Stata.
characteristics. Strategic Management Journal, 14: Stevenson, W. B., Pearce, J. L. & Porter, L. W. 1985. The
33-50.
concept of "coalition" in organization theory and
Kesner, I. F., & Sebora, T. C. 1994. Executive succession: research. Academy of Management Review, 10:
Past, present & future. Journal of Management, 20: 256-268.
327-372.
Tuma, N. B., & Hannan, M. T. 1984. Social dynamics.
Orlando, FL: Academic Press.
Kotter, J. P. 1982. The general managers. New York:
Free Press.
Vancil, R. M. 1987. Passing the baton. Boston: Harvard
Lazear, E. P. 1989. Pay equality and industrial politics. University Press.
Journal of Political Economy, 97: 561-580.
Walsh, J. P., & Seward, J. K. 1990. On the efficiency of
Lazear, E. P., & Rosen, S. 1981. Rank-order tournaments internal and external corporate control mechanisms.
as optimum labor contracts. Journal of Political Academy of Management Review, 15: 421-458.
Economy, 89: 841-864.
Weber, M. 1946. From Max Weber: Essays in sociology.
Lorsch, J. W., & MacIver, E. 1989. Pawns or potentates: New York: Oxford University Press.
The reality of America's corporate boards. Boston: Weisbach, M. S. 1988. Outside directors and CEO turn-
Harvard Business School Press.
over. Journal of Financial Economics, 19: 431-460.
McClelland, D. C., & Burnham, D. H. 1976. Power is the
Yamaguchi, K. 1991. Event history analysis. Newbury
great motivator. Harvard Business Review, 54(2): Park, CA: Sage.
100-110.
Zajac, E. J. 1990. CEO selection, succession, compensa-
Ocasio, W. 1994. Political dynamics and the circulation
tion and firm performance: A theoretical integration
of power: CEO succession in U.S. industrial corpo-
and empirical analysis. Strategic Management
rations, 1960-1990. Administrative Science Journal,
Quar- 11: 217-230.
terly, 39: 285-312.
Zajac, E. J., & Westphal, J. D. 1996. Who shall succeed?
Ocasio, W. 1999. Institutionalized action and corporate
How CEO/board preferences and power affect the
governance: The reliance on rules of CEO succes-
choice of new CEOs. Academy of Management
sion. Administrative Science Quarterly, 44:Journal,
384- 39: 64-90.
416.