% \
.'-"A
'x
'-
c-
/
Ut:
HD28
.M414
^
ALFRED
P.
WORKING PAPER
SLOAN SCHOOL OF MANAGEMENT
IMPERMANENT POWER, CONTESTED POWER: THE
CIRCULATION OF CHIEF EXECUTIVE OFFICERS
U.S.
IN
INDUSTRIAL CORPORATIONS, 1960-1990
Working Paper No. 3544-93
MASSACHUSETTS
INSTITUTE OF TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASSACHUSETTS 02139
IMPERMANENT POWER, CONTESTED POWER: THE
CIRCULATION OF CHIEF EXECUTIVE OFFICERS
U.S.
IN
INDUSTRIAL CORPORATIONS, 1960-1990
Working Paper No. 3544-93
William Ocasio
Assistant Professor of Strategy and Organization Studies
MIT Sloan School of Managennent
E52-562, 50 Memorial Drive
Cannbridge, Massachusetts
Tel.:
(617) 253-8412
E-Mail:
02139
wocasio@sloan.mit.edu
March 1993
This paper is based on my doctoral dissertation research at Stanford University. An
earlier, abridged version will appear in the Academy of Management Best Papers
would like to thank Jeff Pfeffer, Bill Barnett, Jim March, Joel
Proceedings 1993.
I
Podolny, Mauro Guillen, Paul Osterman and Marc Ventresca for invaluable advice,
suggestions, and encouragement.
|j
Ki.i,
-
-
im
3
1
1833
:
IMPERMANENT POWER, CONTESTED POWER:
THE CIRCULATION OF CHIEF EXECUTIVE OFFICERS
IN U.S. INDUSTRIAL
CORPORATIONS,
1960-1990
Abstract
This
paper develops a theoretical
circulation of power, and
compares
it
to the
model of
model of
political
dynamics, the
institutionalization in
an
event history analysis of CEO succession. The model of circulation extends earlier
theories of circulation of elites by Pareto, Mosca, and Michels.
to
a
liability
CEOs
are subject
of experience, as increased tenure and prior board experience
increases their obsolescence and the contestation by rivals.
The study of 225
succession events provides greater support for the model of circulation than for
institutionalization. Contrary to conventional views,
more
inside board
members
increases
CEO
succession.
under economic adversity,
INTRODUCTION
The concept of the business firm
as a political coalition (March, 1962; Cyert
and March,
1963) underlies a long stream of research that challenges unitary actor models and looks to the
of power
role
in
shaping firm behavior. According to political models of organization, firm
behavior responds to the interests and beliefs of the dominant coalition. Chief executive officers
(CEOs) of U.S.
their influence
The
ability
industrial corporations exert a central role in the
dominant
coalition, exercising
through both formal authority and their informal basis of power (Pfeffer, 1992).
of CEOs to exert and maintain
dominant coalition and
to
their
power determines
their ability to control the
shape the direction of the corporation.
Most research on power has
implicitly or explicitly studied
process. For example, in structural contingency
models power
is
alignment of the capabilities of individuals and subunits with the
it
as a static, or equilibrium
obtained by the equilibrium
critical
contingencies and
resource dependencies in the environment (Hickson, Hinnings, Lee, Schneck, and Pennings,
1971; Pfeffer and Salancik, 1978). But
if the
concept of power
is
to
have independent
explanatory power (March, 1966), an equilibrium model will not suffice. If the composition of
the
dominant coalition
is in
constant alignment or equilibrium with the firm' s environment,
power becomes an epiphenomenon, a mere
reflection of structural contingencies
dependencies. If the power and interests of the
and resource
CEO and the dominant coalition are to
have
independent explanatory force, the composition of the political coalition must become
decoupled, over time, from the firm's environmental contingencies. Consequently, an
understanding of the role of power and politics in shaping behavior requires that organizational
theory and research focus not just on equilibrium processes, but on the underlying political
dynamics
that determine the determination
dominant
coalition.
and maintenance of power of the
Contingency theories of power view
the firm's political coalition with
its
CEO succession as
a critical
CEO
and the
mechanism
for aligning
resource dependencies and critical contingencies (Pfeffer
and Salancik. 1978; Tushman and Romanelli, 1985). Failures of economic performance serve as
triggering devices for executive turnover, and for opportunities for realigning the firm with
critical
its
contingencies. But most studies of executive succession have found that the effects of
performance on succession are small (Weisbach, 1988; Fredrickson, Hambrick and Baumrin,
1988), nonexistent (Fizel, Louie, and Mentzer, 1990), or in
actually preceded
some
by abnormally good performance (Morck,
cases, that
Schleifer,
CEO turnover is
and Vishny, 1989). The
loose coupling of executive succession from the firm's economic performance suggests a role for
CEO power in shaping
firm behavior. Political dynamics offer an explanation for
economic performance often
fails to trigger
executive succession.
To explore
why poor
this hypothesis, this
paper studies political dynamics in an event-history analysis of CEO succession in U.S. industrial
corporations from 1960-1990.
studying
how
demands, and
The
The
analysis of
CEO
succession provides an opportunity for
political processes help to insulate executives
how these
political processes
limited research
institutionalization or
power
on
political
from economic and environmental
change over the CEO's tenure.
dynamics has followed,
(Pfeffer, 1981; Boeker, 1989).
for the
According
most
to this
part, a
model of
model, the power
of executives becomes entrenched over time. This paper presents and develops an alternative
model, that of circulation of power, where the power of executives decays over time, as
becomes subject
to obsolescence
it
and contestation. The model of circulation of power builds
upon early
political theories
of circulating
elites
developed by Mosca, Pareto, and Michels. But
unlike most elite theories, the present formulation uses the concept of circulation to describe the
continuing pattern of change and replacement of individual executives, rather than of elite
groups.
The empirical study compares and
models of political dynamics
contrasts a set of alternative hypotheses from the
in the context
of individual
CEO
succession.
two
The findings of the
study provide greater support for the model of the circulation of power.
THEORY AND HYPOTHESES
This paper compares and
institutionalization
tests
two models of power
in U.S. corporations: the
model of
of power, and the model of circulation. These two models provide alternative
conceptualizations of intraorganizational politics and of the ability of CEOs to develop and
maintain stable and cohesive political coalitions that support their power base. The
first
model
has been more prevalent in contemporary organizational theory (Salancik and Pfeffer, 1977;
Pfeffer,
1981) and posits the ability of CEO's to build upon their power to entrench themselves
and perpetuate
the
their
power. The second model highlights the contestation and impermanence of
CEO's power, and
the increasing obsolescence of and opposition to his executive tenure in
the corporation. This paper gives particular attention to the political
dynamics of the CEO's
power within the board of directors (Boeker, 1992).
1)
Model of institutionalization
interrelated processes that
may
of power. Pfeffer (1981) presents three underlying,
lead to the institutionalization and perpetuation of power. First,
an escalation of commitment to a course of action (Staw, 1976)
may
lead decision makers (e.g.,
boards of directors) to beliefs that sustain activities that have been selected
selection).
Second, beliefs and practices associated with those in power
(e.g.,
executive
may become
institutionalized (Selzniclc. 1957)', with the incumbent's actions acquiring the nature of
Rowan, 1977; Zucker, 1977)
"taken-for-grantedness. (Meyer and
incumbents
into question. Third,
may
and establish networks of influence
mechanisms
at the
disposal of the
developing close, stable
ties
in
"
with his" power not called
use their power to expend resources,
ways
CEO to
that consolidate
institutionalize
with existing board
make appointments,
and perpetuate
and perpetuate
members and with
their
their
power. The
power include
corporate officers,
influencing the selection of directors under his control, and exerting influence over subordinate
officers that are also
likely to increase
members of the
board. These
mechanisms suggest
that the
CEO's power
is
over the period of his incumbency, and that appointments of board members
serve to strengthen the
CEO's
influence over corporate decisions, and to insulate
him from
the
pressures of economic performance.
The model of institutionalization
that lead to greater
incumbent
of the
CEO
CEO,
is
used to develop a
set
of hypotheses on the conditions
entrenchment of the CEO's power. The institutionalization of power of the
in the corporate
board of directors
is
directly related to (a) the length of tenure
(b) prior tenure in the board of directors, (c) the proportion of directors appointed
during his incumbency, and inversely related to (d) the proportion of outside directors. The
power of the
'
CEO will
become most evident under conditions of economic
In this paper, the concept of institutionalization
which focuses on the
role
individual organizations.
of executive leadership
As such
it
differs
is
adversity, as
more
closest to that of Selznick (1957),
in attaining institutionalization
from the new institutionalism
within
in organizational
sociology (Powell and DiMaggio, 1991), which highlights institutionalization processes
occurring at the level of the organizational environment, and the organizational
"
There are few female
sample. For convenience,
to
CEOs.
I
CEOs
in U.S. industrial
will use the masculine
companies and none
pronoun throughout
this
field.
in this study's
paper in
all
random
references
powerful
CEOs
will
be able
to use their sources
of power
power and position
to maintain their
within the corporation.
The focus on
the institutionalization of the
CEO's power
within the board of directors
is
consistent with managerial control theories of organization (Berle and Means, 1967), and with
approaches that view corporate boards of directors as "pawns" of management (Lorsch, 1989).
secondary objective of this paper
whether, how, and
when CEOs
is
to use the
exert their
two models of political dynamics
power over
to explore
the board of directors, and inhibit the
capacity of directors to realign the corporation with environmental contingencies.
institutionalization
assumes
that
CEOs
A
are able to extend their
power over
The model of
the boards through
board appointments, by their control over insider directors, and by their prior network of contacts
with board members before becoming CEOs.
2)
Model of circulation of power. The model of circulation of power
alternative
view of political dynamics
in organizations,
among members of top management, and
challenges the view that
CEOs
power
and
political obstacles to the continuation
enemies and
likely to erode
rivals.
As
the obsolescence of a
of their power and to an increase
the organization faces changes in
rivals
its
It
directly
It
argues instead
power
is
in the
to technical
number of
environment, executive capabilities
and enemies will emerge, and the likelihood
great that those in position of power will eventually lose their
circulation of
CEO's power.
and dissipate over time. Powerfijl executives are subject
and programs will be called into question,
The
an
political contests
are able to entrench and perpetuate their power.
that
is
which highlights the
offers
power
(Pfeffer, 1992).
created by the interplay of two underlying mechanisms:
obsolescence and contestation. Chief executives develop strategies for matching problems,
is
solutions,
and choice opportunities
over the formal control,
new
strategies
organization's alignment with
stable
and
inert
in organizational
its
with decreasing
and solutions gain ascendancy which
environment. But early choices
fit
When
decision-making.
between the CEO's
strategies
a
new CEO
takes
alter the
made by CEOs
tend to
become
and programs and
environmental contingencies. In a study of Canadian corporations. Miller (1991) found that the
fit
between environmental demands and organizational structures and
tenure of the
CEO. CEOs become
strategies declined over the
"stale in the saddle" over time, tied to their past policies
programs, and unable to adapt adequately to environmental contingencies.
CEOs
competency
tied to skills
and March, 1988) as
traps (Levitt
their cognitive
patterns of behavior that led to success in the past, but
current environmental conditions.
their technical obsolescence, as
The
become
politically obsolete, with
from
suffer
and
are obstacles to understanding
resulting pattern of inertia in executive decisions leads to
incumbent
solutions to organizational problems.
which
schemas are
and
CEOs
are increasingly unable to provide satisfactory
As CEOs become
an increasing
technically obsolete, they
may
inability to control political conflict
also
and
to
maintain stable and cohesive political coalitions. Schemas and practices that were effective in the
past for gaining control over the political coalition
may
cease to be appropriate as the
composition and interests of the coalition changes. Technical and
political
opposition to his tenure both within and outside the corporation and
According
to the
model of circulation,
accompanied by a second
as directly controlled
the obsolescence of the
force, the contestation
corporations are rivals for the
its
CEO's power and
obsolescence leads to
board of directors.
CEO's
tenure
is
of his power. Executive officers of the
position. Rather than seeing other top executives
by the CEO, the model of circulation argues
that other executive
members
of the dominant coalition have
power and
position.
interests
The degree of contestation
his power, both within the corporation
CEOs
to
power and
independent from the
the
and
emergence of new
latent but are not eliminated.
is
in its
CEO, and
a function of the
are potential rivals to his
number of potential
board of directors. With the ascendancy of
political coalitions, political struggles
The power of the
rivals to
CEO and his coalition
is
may become
always subject to
contestation, with periods of political stability being only temporary interruptions of an
underlying pulling and tugging of contestants for power, position, and privilege. With a more
prolonged tenure for the
latent conflicts
and
CEO, and
as his technical and political obsolescence increases, the
political contests
may come
maintain a working political coalition
to the foreground
may become
threatened.
particularly fertile condition for triggering political contests.
firm deteriorates, latent conflicts
his sources of
power becomes
may become
and the
CEO to
As
the
economic performance of a
CEO to maintain
threatened.
the shifting political coalitions and the
incessant political struggles prevalent in organizations. This
model builds upon
theories of circulating elites developed by Mosca, Pareto, and Michels.
ruled by political elites, or dominant coalitions, these "...elites do not
is
of the
Economic adversity provides a
manifest, and the ability of the
The model of circulation of power emphasizes
of man
ability
early political
While organizations are
last.
Hence
—
the history
the history of the continuous replacement of certain elites: as one ascends, another
declines (Pareto, 1968, p. 36)." Michels (1962) extends Pareto's analysis by recognizing that
ruling oligarchies are characterized
general fraternity
is
is
by struggle among the leaders themselves:
conspicuously lacking;
we do
"...a spirit
of
not see sincere and cordial mutual trust; there
a continual latent struggle, a spirit of irritation determined by the reciprocal mistrust of the
leaders (p.
1
76)." Instead of a simple replacement of one elite group by another, Michels argues
for the process
elite
of amalgamation of power, with old elements intermixing with the new, and new
members intermixing with
existing ones. This conceptualization of elites and their
circulation stresses intraelite conflict as a driving force for change (Putnam, 1976).
It
contrasts
with the model of institutionalization of power, which stresses solidarity and cohesiveness
among group members,
rather than
with political change characterized by
by the amalgamation of existing
The model of circulation
is
used
elite
to derive a set
stability
of power of the incumbent
(a) the length
of tenure of the
CEO,
CEO
elites,
ones.
of conditions under which greater
contested and unstable and decays over time, as
is
replacement of existing
members with new
impermanence and contestation of the executive's power
CEO
full
is
to
new
in the corporate
be expected. The power of the
coalitions emerge.
board of directors
is
The
strength and
inversely related to
(b) prior tenure in the board of directors, (c) the
number of
directors.
The model of circulation of power provides
model of institutionalization regarding
CEO. Given
that political
a set of explanations that
compete with the
the effects of tenure and the political influence of the
dynamics may
insulate the
CEO from the effects of economic
adversity, the effects of political processes will interact with
economic adversity
of CEO succession. The two models of political dynamics are used to develop a
hypotheses that compare and contrast the effects of tenure and the
Given the emphasis on dynamic, disequilibrium process,
their effects
on the
rate
all
CEO's
to affect the rate
set
of
political influence.
hypotheses are presented in terms of
of CEO succession. The interaction effects of economic adversity and
political variables will
be highlighted, as the effects of CEO's power become more salient under
conditions of adversity.
Hypothesis la: The length of tenure of the incumbent
CEO
will
decrease the rate of
CEO succession.
Hypothesis lb: The length of tenure of the incumbent
CEO will
increase the rate of CEO
succession.
The
effects
of duration, or length of tenure of the CEO, provides the
contrasting predictions under the
two models. The
first set
of
institutionalization of power implies that
length of tenure will result in a decreased rate of CEO succession. Increased tenure leads to an
escalation of commitment to the
having been expended
prediction
is
in
maintaining and perpetuating the
CEO's programs and
increase over the duration of the
come
a taken-for-grantedness of his power, and
CEO's power. The
associated with a model of circulation of power, which
obsolescence of a
to
CEO,
to the forefront
CEO's
the
CEO to
is
likely to
to
become openly
contested.
will interact with the length of prior board tenure of
the rate of executive succession.
Hypothesis 2b: Economic adversity
the incumbent
characterized by the
impermanence of his power. Obsolescence
and for the CEO's power
CEO to decrease
opposite
tenure, with increased opportunities for latent opposition
Hypothesis 2a: Economic adversity
the incumbent
is
more resources
will interact with the length
of prior board tenure of
increase the rate of executive succession.
Prior board tenure measures the experience the
CEO has had with the board
before
assuming the top executive position. Does prior experience increase the CEO's power? Or does
prior board experience prove a liability, increasing the
enemies of the
CEO?
This issue remains
unexplored
CEO
in past studies
power
of CEO succession. Again we find a case wliere the two models of
yield opposite predictions. According to the
CEO
board experience increases the opportunities for the
to increase
group cohesiveness, and
According
to the
model of circulation, greater
power, and to expose him
prior board tenure are
more
to
expend resources,
likely to be
prior board experience serves to increase the
CEO,
to provide opportunities to
to a greater
degree of contestation.
committed
to past strategies
may
suffer
from a
liability
develop enemies and
CEOs
with longer
and programs of the
corporation, and less likely to undertake significant organizational change.
board tenures
to build alliances,
power can be maintained.
to increase the probability that his
technical and political obsolescence of the
rivals for his
model of institutionalization, prior
CEOs
of experience, as increased exposure
with longer
to the
board
increased their obsolescence and the probability of contestation.
Hypothesis
during the
3:
Economic adversity
CEO's incumbency
CEO's may
Chandratat, 1991
).
will interact with the proportion of directors appointed
to decrease the rate
of CEO succession.
exert social influence through board appointments
One mechanism CEO's
utilize to
(Wade, O'Reilly, and
maintain and preserve their power
is
through the appointments of executives and board members more favorable to his position.
While board members are nominated by the board
typically
most
itself
and elected by shareholders,
influential in their selection (Lorsch, 1989).
The power of the
institutionalized as the proportion of board appointments he has
no countervailing hypothesis
Hypothesis
4:
is
made
CEOs
CEO becomes
are
more
increases. In this instance
provided by the circulation of power model.
Economic adversity
will interact with the proportion of outside directors to
increase the rate of CEO succession.
10
Both agency theory (Fama and Jensen. 1983; Weisbach, 1988) and the
of power model argue that outsiders are able
The supposition
his power.
to the
CEO
is
power and would
who have
that both inside
hypothesis other than the null hypothesis
the rate of
CEO
is
Economic adversity
contestability of the
4. It
does not necessarily imply the alternative
CEOs
than outside directors, so no competing
presented.
will interact with the
number of directors
to increase
CEO's power. The
control highlights the limits, impermanence, and
stability
and cohesiveness of the governing coalition
CEO can be best contested when there are a larger number of directors on the board. A
larger board is
the
The model of circulation of
succession.
The model of contested power and
under the
CEO, and
more
his power.
No
likely to
be able
to generate alternative political coalitions that
take control over the corporation.
CEO exerting social
A
can challenge
larger board also limits the possibility
of the
influence to maintain a cohesive, stable coalition in the board to maintain
competing hypothesis on board
institutionalization
size is presented for the
model of
of power.
METHOD
Sample.
A
random sample of 120 U.S.
Industrial Directory for 1980
company
limit
and outside directors serve as constraints on the CEO's
hypothesis, that inside directors are less loyal to
5:
CEO and
the
of the management team are more loyal
greater independence.
therefore reject Hypothesis
Hypothesis
more independence from
that inside directors as part
than outside directors,
power would argue
to exert
institutionalization
was
industrial corporations listed in the
selected for the analysis.
year, covering the years 1960-1990.
The
unit of observation
Given lack of financial data
11
Moody's
for six
is
the
of the
companies
had data
Many
in the original sample, the
for the
complete period.
others merged,
sample was reduced
Many were founded
became bankrupt, went
to
1
14 companies. Not
total
companies
and/ or became publicly held after 1960.
private, or otherwise ceased to
companies during the decade of the 1980s. The
all
be publicly held
sample used included 2,391 company-years
of data.
The methodology
ownership as competing
implicitly treats bankruptcy, acquisition,
risks to executive succession. This
and change
to private
means, of course, that the effects
being measured relate exclusively to normal forms of succession within the current ownership
and organization of the firm. Succession due
competing
risk
which
is
The sample was
to acquisition or private
ownership
is
treated as a
excluded from the current analysis.
selected in 1980 to permit firms founded since 1960. including
high-techology companies, to become part of the sample. This creates some sample selection
bias as firms that died
selection bias
between 1960 and 1980 were excluded
would have been eliminated
if
ft-om the sample.
While sample
sampling had been undertaken in 1960,
this
have excluded newer firms fi'om the sample, and the resulting sample would have been
representative of existing firms. Sampling in 1980
would reduce sample
selection bias, at the
was undertaken
same time
that the
as a
would
less
compromise solution
that
sample would be more closely
representative of industrial firms in 1990.
Independent variables and succession events.
were coded from Standard and Poor
on changes
in the
names of the
's
CEO
Turnover.
CEO
succession events
Directory of Corporations, Officers, and Director
relevant officers.
A total of 225
succession events
's
based
was observed
during the 2,391 company-years of data in the sample. Performance. Data on return on assets
12
(ROA), adjusted
for industry averages,
used as a performance measure given
were obtained from
its
widespread use
1990). Tenure. This variable measures the
CEO's
used to measure the duration of the
CEO tenure
is
recorded for
all
incumbents
in prior
number of years
tenure.
COMPUSTAT.
succession studies (Zajac.
the incumbent serves as
To address
the
in 1960. or for the first
incumbent
joined the board of directors was recorded to obtain a measure of the
becoming CEO. This
variable measures the
variable
is
measured
in logarithmic form.
number of board members, both
CEO, and
is
problem of left-censoring, prior
each company. lOKs. and proxy statements. Prior Board Tenure. The year
first
Return on assets was
sample for
in the
when
the
CEO's board
CEO
first
tenure before
Number of Directors. This
insider directors
and outside
directors.
Proportion of Outside Directors. This variable measures the number of outside board members
divided by the total
number of directors.
CEO Appointees.
of all directors
appointed during the
CEO's incumbency.
first
Control variables. Age. The age of the
have a positive effect on the
number of employees
rate
CEO
of CEO succession.
as reported in
This variable measures the proportion
during the current year.
Size.
COMPUSTAT. No
Measured
account increased pressures on
if
the
CEO
is
the founding
CEOs
CEO, and
otherwise,
is
on the
rate
of CEO succession.
13
expected to
is
960, and
is
hypothesized.
is
intended to
expected, taking into
A dummy
variable coded as
included in the analysis. This variable
expected to capture the differential power of a founding
effect
1
A positive trend
during the 1980's. Founder.
is
as the logarithm of the
specific effect of size
Time Trend. This variable measures the calendar years elapsed since
capture historical trends in the rate of CEO succession.
Age
CEO, and
is
1,
is
expected to have a negative
Data sources. Data on performance and
other data were obtained from Standard
and Poor
's
Directors, and supplemented by proxy statements.
Industry
and Finance.
were obtained from
size
1
COMPUSTAT.
Directory of Corporations. Officers, and
OKs, and annual reports, and
All variables, except financial and
analysis, estimated
succession
by
any given year or
for
last
two methods
test the
maximum
be preferred
is to
1
set
The use of event
A
at the
fiscal year.
CEO
section analysis either
are subject to specification bias, due to sample censoring,
newness
is
for
which can be
CEOs.
Implicit in almost
the assumption of
study of the effects of political dynamics on succession must explicitly account
for the likelihood that adaptations to
economic forces
extended period of disequilibrium
adjustment
in
may
will not be instantaneous,
follow.
The
utilization
for dealing with
dynamic processes
that has
and
that
an
of
duration-dependent event history analysis provides an established methodology
Hannan, 1984)
in
of years, or alternatively, sampling only for the year of turnover. These
past empirical models on the determinants of CEO succession
equilibrium.
Who
history analysis for
more common method of using cross
particularly critical given the hypothesized vulnerability of
all
's
hypotheses by specifying continuous-time, event history
likelihood methods.
to the
Who
employment data were recorded
beginning of the year. Financial and employment data used were lagged one
Modeling Procedure.
All
(Tuma and
been almost unexplored
in the
CEO
succession literature. The models are estimated using Rate (Tuma, 1980).
Several alternative specifications of duration-dependence were tested, including the
exponential, Gompertz, and the Weibull models. Although the models are not nested, the
chi-squares of the estimated Weibull models were significantly larger than the exponential or
Gompertz, indicating a better
statistical
fit.
Consequently, the Weibull model was used.
14
0.090
Descriptive Statistics and Correlation Matrix
Table
1
presents the sample means, standard deviations, and the
maximum and minimum
values for the variables and interaction terms used in the analysis. The unit of observation
company
values.
A
is
the
year. Table 2 presents the Pearson correlation coefficients, with the corresponding p-
succession variable that takes the value of
otherwise
is
1
if
a succession event occurs, and
included for comparison. Note that the correlation between
adjusted return on assets measure
is
-0.06, with a p- value of .003.
between market return and succession
p-value of .966. Consequently,
all
is
Note
CEO
succession and the
that the correlation
-0.0009, clearly not statistically significant, with a
reported hazard rate models will use return on assets as a
measure of economic adversity.
An
examination of the correlation matrix reveals a high degree of multicollinearity
between the measures of economic performance and the multiplicative interactions of
performance with other independent variables. The correlations between return on assets and
its
interactions with proportion of outside directors, proportion of directors appointed under the
CEO's
tenure,
number of directors, and
the
CEO's
tenure
exceed
all
.92. Individual correlations
between the interaction terms are also typically very high. The resulting multicollinearity has the
effect
of increasing the standard errors of the coefficients and
statistical test
power of the
of hypotheses. Multicollinearity however does not bias the estimate.
increase the failure to reject the null hypotheses in those cases
the null hypothesis.
level
restricting the
Given the existence of multicollinearity
when
may
the true effect differs
from
in the data, a statistical significance
of .10 will be chosen for hypothesis testing and model specification.
16
It
Table 2
Pearson Correlation Coefficients
Variables
Table 2 (Continued)
Pearson Correlation Coefficients
Variables
RESULTS
Table 3 presents the results of the hazard
model specifications are presented. Model
on
1
rate
model of CEO succession. Six
presents the baseline model, which includes return
assets (adjusted for industry average), all the control variables,
Models 2-5 successively add both
the
main
effects
and the
effects
all
is
included, so
interaction effects with
when
all
Model
1
The
is
the corresponding
main
effect.
economic performance. multicoUinearity
model, which
Given the large number of
is
likely to limit the
power of
variables are included in the full model.
shows a negative
economic adversity leading
level.
full
variables that have achieved a .10 level of significance in previous models. If an
interaction effect
the test
of tenure.
and interaction effects of variables
hypothesized to interact with economic performance. Model 6 presents the
includes
alternative
variable age
is
to
coefficient for return
CEO succession.
on
assets, consistent
This effect
is statistically
with a net effect of
significant at the .01
positive and statistically significant at the .01 level, consistent with the
expectation that the rate of CEO succession increases with age. The variable time trend
positive and statistically significant at the .05 level. This implies that the rate of
CEO
is
succession
has shown an increasing trend over the period 1960-1990. The size of the age and time trend
effects
and
their statistical significance are stable across all the alternative specifications
hazard rate model. The size of the firm has no
The
level in
coefficient for the effect of tenure
Model
1,
as
increases the rate of
it
is
statistically significant effects in the
is
model.
positive and statistically significant at the .01
in all subsequent models. Hypothesis lb, that the length
CEO
of the
succession receives clear support, while Hypothesis la
19
of tenure
is
rejected.
Table 3
Maximum
Model
Likelihood Estimates of the Hazard Rate of CEO Succession
The
results
show a
positive duration effect of
age and for time trend. This provides the
first
CEO tenure,
even
after controlling for
CEO
evidence against the institutionalization of power
model, and in favor of that of circulation of power. Rather than increasing their power and their
ability to
maintain their jobs over time, the positive duration effect shows that as the
tenure increases, the
CEO's
ability to
both increased obsolescence of the
increases over time, as he
is
maintain his position decreases. This
CEO and contestation of his power.
is
consistent with
Opposition to the
unable to satisfy the demands of the organization and
constituents, leading to an increase in the rate of
A positive effect for CEO tenure on CEO
CEO's
CEO
diverse
its
CEO succession.
succession was also found by Puffer and
Weintrob (1991). Their sample covered only succession events when the CEO's age was
less
than 64, but no additional control for age was put in place. Puffer and Weintrob treated tenure as
a control variable and attributed the positive tenure effect
getting closer to retirement age.
The
effects in the current study hold
age and calendar time, consistent with a view that the
contestable.
on succession
Given the finding of stability
in
CEO's
CEO's power
to the effect
even
is
of CEOs
after controlling for
impermanent and
reactions to environmental change (Miller,
1991), increased tenure indicates increased obsolescence and decreased ability to face the
environmental contingencies facing the corporation. This obsolescence decreases the
power and decreases
his opportunities for holding
Hypothesis 2b
effect
is
performance has larger effects when
is
to his position.
supported in Model 2 at the .01 levels of significance. The interaction
of prior board tenure and performance
board tenure
on
CEO's
to increase the rate
CEOs
is
negative, showing that poor
economic
have longer board tenures. The main effect of prior
of succession, but
21
this effect is not statistically significant.
The main
effect
of return on assets
is
now positive,
interaction effect of founder and return
immune
to
on
assets
but
is
not statistically significant. The
negative,
is
showing
that founders are
poor performance effects than those with no prior board experience, but
more
this effect is
also not statistically significant.
The
effects
of prior board experience on the
graphically in Figure
1.
The
rate
of CEO succession are shown
four curves correspond to four levels of board experience: none. 5
years. 10 years,
and 20 years. The curves measure the combined
Model
on
2: return
assets, prior
effects
of three variables
in
board experience and the interaction of these two variables.
These effects are measured independently of other variables. The horizontal axis measures the
multiplier of the hazard rate of
experience
is
upward
CEO
succession. Note that the curve for no prior board
sloping, corresponding to decreased succession under poor performance.
The remaining curves show a downward
slope, with the effects of
on succession the longer the CEO's
greater impact
prior board experience.
Prior board experience, rather than increasing the
immune from
the effects of poor performance
greater susceptibility to the effect of
from a
liability
to the
(i.e..
CEO's power and making him more
Hypothesis 2a),
CEO
is
economic adversity (Hypothesis
of experience, where greater prior board tenure
of enemies, and more
economic adversity having a
is
more
2b).
likely to lead to a
CEOs may
suffer
likely to lead to a larger
number
succession under conditions of adversity. This result leads credence
obsolescence explanation that underlies the model of circulation of power. Prior board
tenure serves as a measure of the obsolescence of the
tenure are
more
they were
members of the
likely to continue the policies
board, and
may
CEO. CEOs with more
extensive board
and programs of the corporation
instituted while
be less receptive to organizational change. According
99
Figure
to
z
z
CE
3
O
o
I
"1
g
O
in
iri
I
1.
2
23
in
O
d
to
Model
more
2. the
CEO
negative effects of prior board tenure and obsolescence on
likely to be felt
prior board tenure
succession are
under conditions of poor economic performance. The interaction effects of
and return on assets remains negative
in
Model
6,
with a significance level of
.10.
The
directions of the coefficients in
results are not statistically significant.
directors appointed under the
CEO's
Model
3 are consistent
The estimated
coefficients
with Hypothesis
show
assets,
shows
that the effect
but the
that the proportion
of poor performance
is
for
dampened. The
its
interaction with return
size
of the coefficient
approximately 2/3 the size of the main effect of economic performance. This means that
CEOs
CEO
have appointed
all
of
tenure decreases the rate of CEO succession, as predicted
by the institutionalization of power model. The positive coefficient
on
3,
is
when
the directors under their tenure, the negative effects of performance
on
succession are reduced by 2/3.
Boeker (1992) tested a similar hypothesis on the
and performance (measured
in
interaction
between board appointments
terms of decrease in sales) in a study of CEO dismissals in
semiconductor companies and also found no
statistically significant effects.
however, that a significant effect existed with respect
appointed. Boeker argued that board loyalty
may
to the proportion
be especially intense
Boeker did
find,
of inside directors
when board members
appointed are insiders. Wade, O'Reilly and Chandratat (1990) in their study of golden
parachutes, found that the proportion of outside board
tenure
was
members appointed under
the
positively related to the incidence of golden parachutes, arguing that this
the social influence of CEO's over outside board appointments.
One
was due
to
potential reason for the lack
of statistical significance of the measure of proportion of director's appointed,
24
CEO's
may be
that
CEOs
may have more
al.).
To
influence over insiders (as in Boeker's results), or on outsiders (as in
test for these alternative
appointed under the
as
main
effects
The
CEO's
hypotheses, measures of proportion of inside board
tenure,
and as interaction
results
compared with the
Wade
et.
members
and of proportion of outsiders appointed were recorded both
effects with return
on
assets.
of these alternative hypotheses are presented as Model 3a, 3b
original results of
Model
3.
The
variables
all
show
that the
in
main
Table
4,
and
of
effects
both inside board appointments and of outside board appointments are to reduce the rate of
succession, and the interaction effects are positive, so that the greater the board loyalty the
limited will be the effect of adversity
the original
on increasing the
of succession. But,
rate
more
like in the case
measure of total proportion of board members appointed, neither the main nor the
and
interaction effects are statistically significant. Because Boeker's study only covers dismissals
his
sample applies
that
of
to only
one industry, the
while board appointments
succession, there
the coefficients.
help limit the
may be
The
may
results are not strictly
in general increase
loyalty of inside board
succession. Another factor that
possibility
is
board loyalty and decrease the rate of
may
members appointed by
similar effects
may
the
CEO may,
participated in the original selection of the
participated in the
CEO.
CEO's
members not appointed
If the process
selection
is
for example,
not hold for other forms of CEO
decrease the significance of this effect
escalation of commitment (Staw, 1976) of board
who
One
sufficient exceptions to this process to create large standard errors in
number of dismissals, but
board members
comparable.
to
may be due
by the CEO, who
of escalation of commitment by
simultaneously at work with that of
board loyalty, the two forces could cancel each other. In any case, the evidence that the
25
to the
CEO can
Table 4
Maximum
Likelihood Estimates of the Hazard Rate of CEO Succession:
Effects of Proportion of Inside vs. Outside Directors
Model
Appointed
help peq^etuate and institutionalize his power in the board through board appointments cannot
be demonstrated in the current study.
Regarding Hypothesis
statistically significant in
Again
I
tlnd
directors
little
4, not
Model
only are the results on the proportion of outside directors not
4, but they are in the opposite direction
support for the model of institutionalization of the
do not reduce the power of the CEO, nor are they
from those predicted.
CEO's power. Outside
better able to respond to the interests
of shareholders as predicted by agency model of the firm.
Hypothesis
level.
As
5 receives support in
number of directors
the
Model
5,
with the null hypothesis rejected
increases, the effects of economic adversity
at the .05
become more
pronounced, as posited by the model of circulation of power. The main effect of board size
positive, so that larger boards lead to greater succession, but this effect
significant.
CEOs
likely to hold
size
is
is
not statistically
with larger boards are faced with greater contestation, with the effect more
under conditions of poor economic performance The interaction effect of board
and economic performance
retains the
same sign
in
Model
6,
but the null hypothesis can no
longer be rejected.
Figure 2 shows graphically
how the
affect the rate of succession.
10, 15,
The
size
of the board interacts with economic performance to
five curves
shown correspond
and 20 board members, respectively. The curve
so that
basically
flat,
adversity
on
adversity
become
CEOs
their rates
to five selected
that corresponds to a
facing very small boards are
immune from
board
sizes: 3, 5,
board size of 3
the effects of
economic
of succession. But as board size increase, the effects of economic
quite noticeable.
27
is
Figure 2
in
.
o
-in
o
Table 5
Maximum
Model
Likelihood Estimates of the Hazard Rate of CEO Succession:
Effects of Number of Inside vs. Outside Directors
In
examining the
effects
CEO
of board size on
effect of an additional inside board
member
is
succession,
we
assuming
are implicitly
that the
equivalent to the effect of an additional outsider.
Table 5 estimates hazard rate of CEO succession separating the effects of number of inside board
members (Model
5a), the
number of outside board members (Model 5b) and
their interactions
with the economic performance measure; Model 5c includes the effects of number of insider and
number of outsider simultaneously. These
includes the total size of the board and
The main
.10 level in
effect of
Model
return
on
assets
interaction with
5b.
These
is
interaction with
number of inside
main
5a, while the
statistically significant in
its
Model
5b.
effect
The
directors
is
which
positive and statistically significant at the
is
negative and not
interaction effect of number of inside directors
directors
is
results indicate that the contestation
likely to take place
5,
economic performance.
of number of outside directors
negative and statistically significant
number of outside
Model
are contrasted with the original
at the .05 level in
Model
5a,
and
while the
Model
negative but not statistically significant in
of power of board members
by other executive officers who function as
rivals
is
stronger and
CEO and
of the
more
serve to
limit his
power. This result contradicts the long-standing assumption that inside board members
serve as
pawns of the
likely to limit the
CEO (Herman,
CEO's power and
1981; Weisbach, 1988). Inside board
to contest
it,
particularly under conditions of
adversity, rather than to uphold
it.
Model 6 presents
effects of board size, prior board tenure
the
combined
with economic performance, as well as the main effects of the
statistically significant
improvement
in
members
CEO's
30
CEO
1,
more
economic
their interactions
tenure.
explanatory power over Model
consistent with the view that the effects of economic adversity on
and
are
Model 6 provides a
at the .10 level,
succession are mediated
by
political processes.
at the .01 le\el,
remain
of tenure on succession are positive and
effects
number of directors
results in
statistically significant
supporting the model of circulation. The interaction effects of prior board tenure
statistically significant at the .10 level, also
effects of
The
The
Model 6
retains the
same
direction, but
is
no longer
statistically significant.
are broadly consistent with the circulation of power
existence of obsolescence and contestation.
statistical results in
supporting the model of circulation. The
Model 6
Due
to the existence
are not particularly strong.
model and with
the
of multicollinearity, the
But the consistency of results
in direction
supporting the model of circulation and opposite to those predicted by the model of
institutionalization
of power lend credence to the interpretation
impermanent and contestable, and
circulation of
Note
that political
CEO's power
is
dynamics are best characterized by the
CEOs.
that both the size
and direction of the main effect for return on assets varies
significantly across the various
negative in Models
1
and
statistically significant in
3,
models estimated. The
relationship in fact
effects are statistically significant
negative, but not significant in
Models
2, 5,
and
6.
model
4,
and
and positive, but not
This indicates that while for the average
corporation the net effect of poor performance
the political
that
is
a failure-induced change in the
masks a heterogeneous response
dynamics within the corporation and
to
its
economic adversity
CEO,
that is
mediated by
board of directors. The political dynamics
within the corporation, as modeled by the circulation of power, serve to decouple the
position from the effects of economic adversity.
31
this
CEO's
CONCLUSIONS
This paper provides a theoretical and empirical contribution to the study of political
dynamics, an area that has received limited attention
succession.
I
in the past,
and
of CEO
to the study
develop a model of circulation of power, which invokes and builds upon earlier
political theories
of circulation of elites by Mosca. Pareto, and Michels, and apply
of political dynamics in U.
circulation of
S. industrial corporations.
CEOs: obsolescence and
are subject to a liability
I
posit
two mechanisms
contestation. According to the
of experience, as greater
to
it
to the study
account for
CEOs
model of circulation,
familiarity with past practices
and
politics,
increases both technical and political obsolescence, and increases the potential for contestation of
the
CEO's power. The model of circulation of power
is
contrasted and tested against the
model
of institutionalization, which had been previously formulated, but had been infrequently tested.
By
focusing on political dynamics in an event-history analysis of CEO succession, this
study permits us to explicate
directors
how the power
becomes decoupled from
over the
is
CEO's
tenure.
economic adversity
While
results are consistent with the
for the average corporation, poor
that is
this results in fact
shaped by the
None of the hypotheses
be
The
shaped by the dynamics of the CEO's power, and
an increased rate of CEO succession,
to
CEO over the corporation and
statistically significant,
its
board of
the pressures of economic performance and, implicitly,
the firm's environmental contingencies.
succession
of the
political
view
that
from
CEO
that these political forces
change
economic performance leads
masks a heterogeneous response
to
to
dynamics of the corporation.
derived from the institutionalization of power model were found
and
the opposite predicdon, the latter
in those cases
where the model of circulation of power offered
model was supported. The
32
results
of the models clearly
contradict the view that
CEOs
are, in general, able to institutionalize their
A
corporation, and that boards of directors are simply their pawns.
ability
of CEOs
to increase their
power
in the
possible exception
power through board appointments, but
this result
the
is
was not
statistically significant.
The
results are
impermanent. The
more
political
consistent with a view that the
power of the
dynamics within the corporation and
by the model of circulation of power. Strong support was found
power, with a positive duration effect on the
decays over time. Obsolescence
may
is
rate
its
board
is
is
contested and
best characterized
for obsolescence in the
associated with a political liability of experience, as
The
effects
CEOs, but during
coalitions to challenge the coalition headed by the incumbent
number of inside board members, who
CEO's power and
positions,
CEO
CEOs
their prior
of contestation increase with the number of potential
opponents within the board. Larger boards increase the possibilities for forging new
the effects of the
CEO's
of CEO succession, as the power of the
increase their enemies with increased experience, not only as
tenure in the board of directors.
CEO
position, as sources of information
and as sources of social comparison
political
CEO. Of particular importance
is
serve as potential contestants to the
and validation of the CEO's programs and
in the board's evaluation
of the CEO's
effectiveness.
The
effects
of the number of inside board members on the
interaction effects with
inside board
members
economic adversity
serve as
directors helps to limit the
board
may
call into
that increasing the proportion
CEO over the board. A
be required for board members
to
of succession, and
its
question long-standing assumptions that
pawns of the CEO, and
power of the
rate
large
number of insiders
have adequate information
33
of outside
to evaluate the
in the
CEO's
explanations for the adequacy of the firm's performance. The small, net effects of performance
on
CEO
succession suggest that the effects of adversity are not automatic, but are dependent on
the perception
task. Inside
by the board on whether the CEO's policies and
board members serve several functions that
may
capabilities are adequate for the
increase
CEO
succession under
adversity: they are readily-available candidates for the position, they possess intensive
information on the company's operation, and
opposition to the
The
in
when economic
CEO may become manifest as contestants
results
of this study serve not only
adversity
is
faced, their rivalry and
for his position.
to highlight the significance
of political dynamics
executive succession, but to challenge conventional assumptions and beliefs regarding the
nature of those dynamics. Recent highly-publicized instances of CEO turnover in General
Motors and
IBM have
been interpreted by the business press as indications of the need for an
increased role for outside directors in corporate governance, particularly under conditions of
economic
adversity.
As
"decide" to remove the
removal of a
CEO by
a crisis
is
faced by the corporation, outside directors meet in private and
CEO. But such
observations ignore contrary evidence, such as the
an insider-dominated board in
unexplored role of inside directors
in
(Vancil, 1987
)
and the
supplying information, serving as sources of comparison,
and or contesting the CEO's position. The
results
question the basis for assuming that outside board
CEO's power, and
Dow Chemical
of this study, while not definite, lead us
members
are an effective control over the
lead us to look instead toward alternative explanations based on the
obsolescence and the contestation of his power.
34
to
CEO's
REFERENCES
Berle, Adolf, Jr.,
1967
The
and Gardiner C. Means
Modem Corporation and
Private Property.
New
York: Harcourt, Brace, and World.
Boeker, Warren
1
989
"The development and
institutionalization
of subunit power
in organizations."
Administrative Science Quarterly. 34: 388-410.
1992
"Power and managerial
dismissal: Scapegoating at the top." Administrative Science
Quarterly. 37:400-421.
Cyert, Richard M., and James G.
1963
A Behavioral
Fama, Eugene.
1983
L.,
March
Theory Of The Firm. Englewood
Cliffs. N.J.: Prentice-Hall.
and Michael C. Jensen
"Separation of ownership and control." Journal of Law and Economics. 26: 301-325.
John L., Kenneth K. T. Louie, and Marc S. Mentzer
1990
"An economic, organizational and behavioral model of the determinants of CEO
tenure." Journal of Economic Behavior and Organization. 14: 363-379.
Fizel,
Frederickson, James W., Donald C. Hambrick, and Sara Baumrin
1988
"A model of CEO
Herman, Edward
1981
dismissal."
Academy of Management Review.
255-271.
S.
Corporate Control, Corporate Power. Cambridge, U.K.: Cambridge University Press.
Hickson, D., C. R. Hinnings, C. A. Lee, R. E.
1971
13:
"A
strategic contingencies theory
,
Schneck, and
J,
M. Pennings
of intraorganizational power." Administrative Science
Quarterly. 19:22-44.
Levitt,
1988
Barbara and James G. March
"Organizational learning." Annual Review of Sociology. 14: 319-340.
Lorsch, Jay
1
989
W.
Pawns
or potentates. Boston: Harvard Business School Press.
March, James G.
1962
"The business firm
1966
as a political coalition." Journal of Politics. 24: 662-678.
"The power of power."
In
David Eaton,
[Ed.] Varieties
of Political Theory.
New York:
Prentice-Hall.
Meyer, John W., and Brian Rowan
1977
"Institutionalized organizations:
Formal structure as myth and ceremony." American
Journal of Sociology. 83: 340-363.
35
Michels, Robert
1962
Miller,
1991
Political parties.
New
York: The Free Press.
Danny
CEO
"Stale in the saddle:
Management
tenure and the
mach between
organization and environment."
Science. 37: 34-52.
Morck, Randall, Andrei Shieifer, and Robert W. Vishny
"Alternative mechanisms for corporate control." American Economic Review.
1989
9:
842-852.
Pareto, Vilfredo
1968
"The Rise And
Fall
Of The
Totowa,
Elites."
N.J.:
The Bedminster
Press.
Pfeffer, Jeffrey
1
98 1
1992
Power
in organizations. Marshfield. Mass.:
Managing with power:
Politics
Pitman.
and influence
in organizations.
Boston:
Harvard Business School Press.
Pfeffer, Jeffrey,
1978
The
and Gerald R. Salancik
external control of organizations.
Powell, Walter W., and Paul
1991
The
New
J.
DiMaggio
New
York: Harper and Row.
(Eds.)
Institutionalism in Organizational Analysis. Chicago:
The University of
Chicago Press.
Puffer, Sheila. M.,
1991
and Joseph B. Weintrop
"Corporate performance and
CEO turnover:
The
of performance expectations."
role
Administrative Science Quarterly. 36: 1-19.
Putnam, Robert D.
1976
The Comparative Study of Political
Elites.
and Jeffrey Pfeffer
power and how they hold on
Englewood
Cliffs, N.J.: Prentice-Hall.
Salancik, Gerald R.,
1977
"Who
gets
—
power." Organizational Dynamics.
5:
to
it:
A strategic-contingency model
of
3-21.
Selznick, Philip
1957
Leadership in Administration. Evanston,
111.:
Row,
Peterson.
Staw, Barry M.
1976
"Knee-deep
in the big
muddy:
A
study of escalating
action." Organizational Behavior and
Tuma, Nancy
1980
Human
commitment
Performance.
B.
Invoking Rate. Menlo Park. Calif: SRI International.
36
to a
16: 27-44.
chosen course of
Tuma, Nancy
1984
B.,
and Michael T. Hannan
Models And Methods. Orlando:
Social Dynamics:
Tushman, Michael
1985
Vancil, Richard
in
Press.
and Elaine Romanelli
A
metamorphosis model of convergence and reorientation.''
Organizational Behavior. 7: 171-222. Greenwich, Ct.: JAI Press.
"Organizational evolution:
Research
1987
L.,
Academic
Fla.:
M.
Passing the Baton: Managing the Process of CEO Succession. Cambridge, Mass.:
Harvard Business School Press.
Wade, James, Charles A.
1990
O'Reilly, and Ike Chandratat
"Golden parachutes:
CEOs
and the exercise of social influence." Administrative
Science Quarterly. 35: 587-603.
Weisbach, Michael
1988
Zajac,
1990
S.
"Outside directors and
Edward
"CEO
CEO turnover." Journal
of Financial Economics. 20: 431-460.
J.
selection, compensation,
empirical analysis." Strategic
and firm performance:
Management
Zucker, Lynn G.
1977
"The role of institutionalization
Journal.
1
A
1:
in cultural persistence."
Review. 42: 726-743.
37
theoretical interpretation
217-230.
American Sociological
and
5939 060
Date Due
Lib-26-67
MIT
3
TDflD
LIBRARIfS DUPl
DOflMbSbO
S