The Ecology of Organizational Demography: Managerial Tenure …sorensen/EOD.pdf · 2006. 7. 12. · The Ecology of Organizational Demography 4 2 Among the voluntary organizations

1 Funds from the University of Chicago Graduate School of Business supported thewriting of this paper. Comments from Ron Burt, Diane Burton, Glenn Carroll, Michael Hannan,Rakesh Khurana, Joel Podolny, Olav Sorenson and Toby Stuart have been very helpful. Direct allcorrespondence to Jesper B. Sørensen, Graduate School of Business, University of Chicago, 1101East 58th St, Chicago, IL 60637, or [email protected]

The Ecology of Organizational Demography:Managerial Tenure Distributions and Organizational Competition1

Jesper B. SørensenUniversity of Chicago

July 1999

Forthcoming, Industrial and Corporate Change

Copies of this paper may be downloaded from:http://gsbwww.uchicago.edu/fac/jesper.sorensen/research

The Ecology of Organizational Demography:Managerial Tenure Distributions and Organizational Competition

Abstract

This paper argues that turnover processes in top management teams createinterorganizational interdependencies. I argue that managerial capabilities are shaped by theirexperiences in a given competitive context; differences in managerial tenure therefore lead todifferences in managerial capabilities. A comparison of the tenure distribution of a topmanagement team with those of its competitors therefore captures the extent to which a firmrelies on managerial capabilities similar to those of its competitors. Overlap in managerialcapabilities will lead to greater competition for resources, since managers shape a firm’s pattern ofresource utilization. I hypothesize that organizational growth rates will decline to the extent thatcompetitors are crowded around a firm’s location in the tenure distribution. Analyses of growthin viewership among commercial television stations support this claim.

The Ecology of Organizational Demography:Managerial Tenure Distributions and Organizational Competition

Sociologists and organizational theorists have long been interested in understanding how

the behavior and beliefs of individuals affect organizational outcomes. Yet the shift in

organizational theory toward macro-level, open systems theories (Hannan and Freeman 1977;

Meyer and Rowan 1977) and their emphasis on the organization’s external environment has led to

a decline in theorizing about how individuals influence organizational outcomes. As an exception

to this trend, theories of organizational demography have emerged as a powerful means of linking

easily observable demographic characteristics of management teams to organizational outcomes

(Pfeffer 1983; Hambrick and Mason 1984). Theorists in this tradition argue that the diversity of

ideas and perspectives in a decision-making group reflects the degree of demographic

heterogeneity of group members. This heterogeneity in outlooks is in turn thought to affect

decision-making and organizational behavior.

Drawing on cohort arguments (Ryder 1965; Pfeffer 1983) organizational demographers

have primarily studied the effects that managerial tenure distributions have on organizational

behavior. Some studies suggest that increased demographic diversity is beneficial to

organizations because it exposes the firm to a variety of influences and makes it more responsive

to environmental changes. Along these lines, both Eisenhardt and Schoonhoven (1990) and

Hambrick, Cho, and Chen (1996) have shown that heterogeneity in the tenure distribution of top

management teams is associated with higher growth rates. Others argue that the diversity of

perspectives that accompany demographic heterogeneity in fact impedes effective decision making

The Ecology of Organizational Demography 2

and reduces group integration. For example, Ancona and Caldwell (1992) found that

heterogeneity in firm tenure was negatively associated with adaptive change among electronics

firms.

A core assumption in this research tradition is that the values, beliefs and capabilities of

individuals are shaped by their past experiences and therefore by the cohort to which they belong

(Hambrick and Mason 1984; Finkelstein and Hambrick 1996; cf. Ryder 1965). This implies, as a

corollary, that an organization’s capabilities are a function of the tenure distribution within the

firm. The organizational demography argument thus consists of two interrelated claims: 1) the

tenure distribution (or other demographic characteristics of the group) affects the distribution of

managerial capabilities; and 2) the distribution of managerial capabilities influence organizational

performance.

A common feature of organizational demography research is the adoption of a focal

organization perspective. In other words, demographic effects are seen as internal to the firm; in

particular, organizational demographers attribute no causal or mediating force to the demographic

characteristics of other organizations. In this respect, such studies run counter to the prevailing

open-systems emphasis on mutual organizational interdependence. More importantly, this

atomistic approach seems internally inconsistent, given the central theoretical emphasis on the role

of managerial experiences in shaping organizational capabilities. Organizations facing a common

environment will be exposed to a common pattern of environmental change. People working with

these organizations will therefore also have overlapping experiences. Unless one assumes that all

relevant managerial experiences are firm-specific, this implies that organizations whose managers

have similar tenures (in a common competitive setting) will have similar capabilities.


The full competitive implications of an organization’s tenure distribution therefore only

can be understood in an ecological framework that incorporates the tenure distribution of its

competitors. From an ecological perspective, firms should benefit, on average, to the extent that

they possess unique capabilities; in other words, to the extent that their managerial tenure

distributions are distinctive. Conversely, the greater the extent to which firms are led by managers

with similar capabilities, the more intensely they will compete for resources (Hannan and Freeman

1989). This paper explores two primary implications of an ecological approach to organizational

demography processes. First, I focus on the effects of overlap in the tenure distribution among

competitors. I argue that the greater the extent to which a firm’s tenure distribution matches

those of its competitors, the greater the competitive pressure the firm will experience, which in

turn leads to lower growth rates. Second, I explore the consequences of increases in an

organization’s mean tenure relative to the mean tenure level of its competitors. Arguments in

both the organizational demography (Katz 1982) and organizational ecology (Baron, West and

Hannan 1994) literatures suggest that increases in the mean tenure of management teams leads to

declines in organization-environment fit over time. In this case, an ecological perspective

suggests that such a decline in fit will be particularly detrimental to performance if other firms are

better aligned with environmental demands.

This argument builds on key insights by McPherson and his colleagues concerning the

“ecology of affiliation” among voluntary associations (McPherson 1983; McPherson, Popielarz

and Drobni… 1992). Like McPherson, I conceive of organizations as arrayed in a social space

defined by the attributes of their members, and argue that firms in densely populated regions of

this space will fare worse than firms in sparsely populated regions. My argument differs from


2 Among the voluntary organizations studied by McPherson, of course, an excellentmeasure of performance is the organization’s ability to attract and retain members.

McPherson’s in several key respects, however. First, in applying these ideas to a population of

for-profit, employing organizations, my interest is not in how the social similarity of members

affects competition for members, but rather in how it affects competition for other resources, in

this case television viewers.2 I thereby link two dimensions of the resource space by showing how

a firm’s performance in its product market depends on its position in the social space defined by

its member’s attributes. Furthermore, I capture similarity between managers not in terms of

socio-demographic characteristics such as age or education, but rather in terms of the timing of

entry into top management teams. The central idea is that managerial capabilities are shaped by

experiences in a given competitive context; in other words, managerial decision making and

cognitive frameworks depend on past experiences (Daft and Weick 1984; Porac and Thomas

1990).

This paper more generally makes contributions to the organizational ecology literature, in

particular the growing bodies of work on niche crowding or overlap (Podolny, Stuart and Hannan

1996) and localized competition (Hannan, Ranger-Moore, and Banaszak-Holl 1990; Baum and

Mezias 1992). Elaborating on density dependence theory (Hannan and Carroll 1992) and

network theories of competition (Burt 1992), researchers have developed increasingly

sophisticated measures of the proximity of organizations in resource space. In line with this

tradition, I argue that overlap among firms along the tenure distribution implies overlap in

managerial capabilities; this overlap in capabilities in turn increases interorganizational

competition. The focus on overlap in managerial capabilities is unique and important, since


managers play a central role in determining a firm’s resource position. If a manager’s experiences

shape which strategic options are considered and deemed feasible for the firm (Geletkanycz and

Hambrick 1997), then the relative locations of firms in this experience distribution can have a

fundamental influence on the structure and dynamics of organizational competition.

The remainder of the paper is organized as follows. In the next section, I lay out in more

detail the implications of an ecological approach to the effects of organizational demography on

firm performance. From this argument, I derive a series of testable hypotheses concerning the

effects of overlap in tenure distributions and mean tenure levels on firm growth rates. I then test

these hypotheses on a longitudinal sample of commercial television stations in the United States.

Managerial Tenure Distributions and the Ecology of Competition

In past studies of organizational demography, the tenure distribution has received the

lion’s share of attention. Much of this research has focused on the demographic characteristics of

top management teams, under the assumption that organizational behavior reflects “the values and

cognitive bases of powerful actors in the organization” (Hambrick and Mason 1984: 193). This

perspective assumes that the presence of bounded rationality and multiple, conflicting goals means

that organizational decisions come to reflect the “givens” that each decision maker brings to the

situation (March and Simon 1958; Cyert and March 1963). Hambrick (1994) argues that top

management team characteristics are consistently better predictors of organizational outcomes

than CEO characteristics. For these reasons, I follow previous research and focus on the

demography of the top management team, as opposed to the demographic characteristics of any

one individual or of the entire organization.


Most research in organizational demography has focused on the consequences of tenure

for group processes within organizations. With respect to tenure, two distributional

characteristics have garnered attention: the mean level of tenure in a group, and the degree of

tenure heterogeneity.

Several researchers have emphasized the consequences of mean tenure levels for

organizational outcomes. Pfeffer (1983) argued that increased tenure improves employee

performance as understandings of organizational routines improve and the employee adjusts to the

new organization (cf. McNeil and Thompson 1971). Katz (1982) argued that increased group

tenure reduces goal conflict and improves social integration; this is in part due to the fact that

increases in group tenure also lead to restricted information processing. Finkelstein and Hambrick

(1990) argue that tenure increases top managers’ commitment to the organizational status quo,

makes individuals more risk averse and less open to new information, and increases the likelihood

that they will emulate the behavior of other organizations in their industry. Lant and Mezias

(1992) argue that higher levels of tenure improve first order learning (e.g., the mastery of

established routines) but hinder second-order learning (e.g., learning from sudden environmental

changes).

Others have emphasized tenure heterogeneity as a driver of organizational outcomes. The

consequences of tenure heterogeneity for organizational outcomes are also mixed, however

(Carroll and Harrison 1998; Pfeffer 1997). On the one hand, heterogeneity may be beneficial to

organizations, as it implies a greater diversity of perspectives and exposure to more sources of

information. Heterogeneity may therefore be particularly beneficial in situations where creativity

and innovation are required (Hambrick, Cho and Chen 1996). Eisenhardt and Schoonhoven


(1990), for example, show that growth among semiconductor firms is positively related to

heterogeneity in top team tenure. Hambrick, Cho and Chen (1996) also find a positive effect of

tenure heterogeneity on growth among airlines; moreover, they find that heterogeneous teams are

more likely to initiate new strategic moves. On the other hand, tenure heterogeneity has also been

found to have negative effects. In particular, a diversity of perspectives may make communication

and social integration more difficult (Ancona and Caldwell 1992). O’Reilly, Snyder and Boothe

(1993) found that heterogeneity in firm tenure was negatively associated with adaptive change

among electronics firms. Smith et al. (1994) found a negative effect of tenure heterogeneity on

informal communication among team members. Hambrick, Cho and Chen (1996) found that

tenure heterogeneity had a negative effect on an airline’s propensity to respond to the competitive

moves of other airlines.

As this review indicates, the focus of research in organizational demography has been on

the consequences of the tenure distribution for group processes in the top management team.

While these group processes are rarely studied directly (for exceptions, see O’Reilly, Caldwell and

Barnett 1989; Glick, Miller and Huber 1993; Smith et al., 1994), most arguments rely on them in

order to link demographic characteristics to firm outcomes. However, the tenure distribution of a

top management team has implications for firm performance beyond its consequences for internal

group processes. In particular, it affects how the firm interprets its environment and competitive

situation, and therefore shapes resource allocation decisions.

A long tradition in organizational research suggests that the perceptions and

interpretations a manager brings to decision-making situations will reflect his or her “cognitive

base” (March and Simon 1958; Hambrick and Mason 1984). These cognitive frameworks restrict


managerial attention and influence the ways in which information is perceived and processed. An

individual’s mode of understanding and interpreting is shaped by a variety of experiences,

including education, training, and on-the-job experiences. Thus cognitive perspectives on

managerial decision making suggest that the ways in which managers interpret and react to

competitive situations will depend on their past experiences (March and Simon 1958; Hambrick

and Mason 1984; Daft and Weick 1984; Porac and Thomas 1990). A manager’s on-the-job

experiences are a form of on-the-job training which shapes the nature of the manager’s human

capital.

By extension, qualitative differences in managerial capabilities derive in part from

differences in experiences, which are captured in the relative tenures of managers: the degree to

which two individuals will have similar experiences in a role (in this case, top manager in a

particular market) depends on their relative times of entry into the role. If managers draw lessons

from past experiences, then the more similar those experiences are, the more likely it is that two

managers will draw similar conclusions and have similar managerial capabilities. This suggests

that a firm’s behavior depends on the pool of events and experiences its managers have to draw

upon in assessing new situations.

When we turn to compare tenure distributions among firms in a market, it becomes

apparent that organizations (in the same competitive context) with similar tenure distributions will

have management teams with similar experience profiles; this isomorphism in managerial

experiences implies that the firms draw on similar managerial capabilities. Note that I am not

claiming that managers with similar tenures will have identical experiences. For example,

differences between organizations (in culture, competitive position, etc.) may lead managers in


different firms to interpret the same events differently. However, managers with exposure to a

common set of events are more likely to have similar perspectives than managers with wholly

different experiences.

In ecological terms, the tenure distribution of the top management team defines the firm’s

niche in the managerial labor market. A firm’s tenure distribution is therefore not only an

organizational attribute, but also a relational attribute: it defines an organization’s position with

respect to other organizations in its environment. Furthermore, since managers decide which

resources to employ (and how they should be employed) based on their interpretations of the

environment, the organization’s position in the market for managerial talent affects its overall

competitive position.

In this respect, tenure as an organizational attribute is similar to organizational size.

Organizational size not only has implications for organizational structure (Blau and Schoenherr

1971), but also for a firm’s relation to its environment. Hannan and Freeman (1977: 945)

suggested that large organizations, in part due to the changes in structure caused by growth,

depend on a different mix of resources than small firms. Basic ecological assumptions imply that

the more similar the resource requirements of organizations, the greater the level of (perhaps

latent) competition between them (Hannan and Freeman 1977, 1989). This reasoning suggests

that competition will be localized along the size distribution in a population; organizations will

compete most intensely with other organizations of similar size. Previous research has indeed

uncovered evidence of such size-localized competitive processes. Baum and Mezias (1992), for

example, find evidence of size-localized competition in the Manhattan hotel industry; Baum and

Mezias (1993) found similar effects in a population of day care centers.


3 Like the size-localized competition argument, my argument depends on the underlyingassumption that there are heterogeneous resources that potentially support different competitivepositions. Differentiation from competitors is only a viable strategy to the extent that theenvironment supports the new competitive position.

By a similar logic, I expect that firms with similar top management tenure distributions

will compete more intensely, due to their overlap in managerial capabilities.3 Overlap in

managerial capabilities may increase competitive pressures in two ways. First, firms that depend

on similar managerial capabilities may pursue similar strategies. Geletkanycz and Hambrick

(1997) show that the diversity of experiences in a top management team is an important

determinant of strategic conformity in an industry: firms that draw heavily on intraindustry

experiences are more likely to conform to prevailing strategic practices in the industry. In

particular, their research shows that ties to people outside the firm’s competitive context increases

exposure to new ideas and practices and lowers strategic conformity; ties to people within the

same competitive context encourages the development of common understandings of the

competitive environment and hence increases strategic conformity. By a similar argument, when

comparing top management teams across firms, we can expect that firms with similar tenure

distributions will be more likely to operate from a common understanding of the competitive

situation and hence more likely to adopt similar strategies. By contrast, firms with dissimilar

tenure distributions should be more likely to pursue dissimilar strategies and thus compete less

intensely with each other.

Even among firms pursuing similar strategies, however, there may be competitive

advantages to relying on different managerial capabilities. The experiences of managers may also

affect the ways in which strategies are executed (Gunz and Jalland 1996). If firms relying on


4 This decision could be characterized as a strategic one, as opposed to a reflection of achosen implementation. But a focus on local crime can also be implemented in a number ofdifferent ways. The point here is not the proper level at which to define strategy, but rather thatdifferences in managerial capabilities can have broader consequences for organizationalperformance, even in situations where strategies are highly institutionalized. This also suggeststhe difficulty of designing a study comparing the strategic positions of firms.

5 This lack of competition is no guarantee that people will tune in.

similar managerial capabilities are more likely to implement their strategies in a similar fashion,

they will compete more intensely for resources. For example, two television stations may both

pursue the same strategy of showing local news at 11, but they may execute these strategies quite

differently depending on the experiences and capabilities of their management team. To the extent

that the two stations adopt approaches to news broadcasts that draw on the same environmental

resources, they will compete more intensely. Thus both stations may decide to place a heavy

emphasis on coverage of local crime.4 If so, they will compete for access to at least two types of

resources – criminal events and viewers who prefer this type of news. A station emphasizing local

politics and education would by contrast experience relatively less competition for resources.5

Thus overlap in managerial capabilities may increase competitive pressures even in settings where

strategic positions are highly institutionalized.

Both arguments suggest that the level of competition between organizations should be a

function of the degree to which their managerial tenure distributions overlap; the total competitive

pressure experienced by the firm will consequently be a direct function of the crowding around its

position in the tenure distribution. Firms in more crowded regions of the distribution will

experience greater competitive pressure than firms in sparsely populated regions.


The level of competition between organizations is difficult to measure directly; however, it

can be observed indirectly by considering its implications for different organizational outcomes.

Ecological studies have demonstrated that organizational growth rates vary inversely with the

degree of competition for resources (Barron, West and Hannan 1994; Podolny, Stuart and

Hannan 1996; Barnett 1994). As the number of competitors for a fixed set of resources increases,

growth becomes more difficult. I therefore expect that crowding around a firm’s position in the

tenure distribution will have a negative effect on growth rates. This crowding can be measured in

two ways, depending on how one chooses to capture a firm’s location in the tenure distribution.

The first alternative is to use a measure of central tendency, such as the mean, to represent the

firm’s position. In this case, crowding around the firm’s position can be measured (inversely) by

the average Euclidean distance between the focal firm’s mean tenure and that of its competitors.

This leads to the following hypothesis concerning the effects of a firm’s location in the tenure

distribution across firms:

Hypothesis 1a: Growth rates will increase with a firm’s distance from its competitorsalong the tenure distribution.

(Note that since distance is an inverse measure of crowding, it is expected to have a positive

effect on growth.)

Measures of central tendency such as the mean may conceal distributional differences

between firms in their recruitment patterns. A mean tenure of 10 years, for example, may come

about through having three members each with 10 years of tenure, or from having two members

with 5 years of tenure and one with 20. Such a difference may clearly be important: for the first

firm, its experiences are entirely redundant with those of the second firm, while the same is not

true for the second firm. The Euclidean distance measure does not take this asymmetry into


account. For this reason, I also use the variance in team tenure to capture a firm’s position along

the tenure distribution. Specifically, I define a firm’s niche as the two standard deviation window

centered around the mean tenure (cf. McPherson 1983). Crowding around a firm’s position is

then a question of the extent to which other firms occupy the same niche (McPherson 1983;

Podolny, Stuart and Hannan 1996).

Hypothesis 1b: Growth rates will be a negative function of a firm’s overlap with itscompetitors along the tenure distribution.

Organizational demography research suggests that as mean tenure levels increase, the

likelihood that a firm’s managerial capabilities will be misaligned with its environment increases.

For example, organizational demographers argue that increases in mean tenure are accompanied

by increasingly restricted information processing and an unwillingness to take risks (Katz 1982).

Lant and Mezias (1992) argue that as average tenure increases, the management team becomes

increasingly adept at carrying out established routines (first-order learning) but decreasingly able

to learn from new, unfamiliar environmental conditions (second-order learning). Consistent with

this, Finkelstein and Hambrick (1990) find a strong positive relationship between mean tenure and

strategic persistence or a firm’s resistance to engage in strategic change. In the face of

environmental change, such strategic persistence can cause a decline in fit between managerial

capabilities and environmental demands. This inertia can be a particular liability if a firm’s

competitors experience greater levels of turnover and thereby remain in better alignment with the

environment; thus long-tenured management teams should be at a particular disadvantage when

crowding in the market occurs around low levels of tenure. Conversely, any liability due to

misalignment with the environment is mitigated to the extent that other management teams suffer

from the same degree of misalignment. Thus for long tenured management teams, crowding may


6 Long-tenured management teams may – regardless of the level of crowding – stillexploit environmental opportunities less efficiently than short-tenured teams. If such resourcesare sufficiently attractive, other firms will generally enter (Porter 1980). It is important in thisrespect that television stations are to a large extent sheltered (by FCC regulation of broadcastlicenses) from threats of entry by new competitors.

be preferable to its absence.6 By the same logic, a short-tenured management team should find

advantages in a setting where its competitors primarily have long-tenured management teams.

Crowding around high levels of tenure implies that most firms are pursuing relatively

inappropriate strategies and hence leaving resources unexploited. If short-tenured management

teams are more open to new ideas (Finkelstein and Hambrick 1990) and more willing to take risks

(Katz 1982), then they should be better able to exploit such new niches. This suggests the

following related hypotheses:

Hypothesis 2a: Distance from competitors will be relatively advantageous for managementteams with low average tenure and disadvantageous for management teams with highaverage tenure.

Hypothesis 2b: Overlap will be relatively advantageous for management teams with highaverage tenure and disadvantageous for management teams with low average tenure.

These hypotheses can be tested using an interaction effect between mean tenure and the

respective crowding measures.

Data and Methods

The data for most research in organizational demography have come from random

samples of organizations selected according to different criteria. For example Smith et al. (1994)

selected a sample of technology intensive firms. Finkelstein and Hambrick (1990) chose a sample

of firms in the computer, chemical, and natural-gas distribution industries. A sampling approach


7 The rapid spread of cable television, and the attendant emergence of “super-stations”,has weakened the boundaries between markets somewhat. Cable television grew rapidly in the1980s, so this weakening occurred at the tail end of the sample analyzed in this paper.

does not allow for a consideration of the ecological consequences of management team

demography; the data requirements for this paper are therefore correspondingly higher. In order

to test the ideas behind Hypotheses 1 and 2, a longitudinal record of a complete population of

competing firms is required, including data on the executive personnel employed by each firm.

I have collected this type of data for the commercial television broadcasting industry.

Archival sources contain rich information on organizational characteristics, personnel and

performance of local television stations, in addition to data on environmental conditions. By

tracking the personnel lists over time, I am able to compute the mean tenure for management

teams at each station in a market, every year. By collecting information on all competitors, I am

able to construct the tenure distribution across competing firms for each year of data. This is task

is simplified by another aspect of local television broadcasting: the regulation of competition by

the Federal Communications Commission. Because the FCC defines local broadcast markets and

assigns broadcast licenses by market, the geographical dispersion of television stations is a

fundamental structural feature of the industry. These broadcast markets segregate competition

for viewers; a station in New York does not compete for viewers with a station in Chicago.

Therefore, a station’s direct competitors can readily be identified as the other stations in the focal

station’s market.7


8 The Factbook was not published in 1962, 1975 and 1983.

9 The choice of 1988 as a cutoff was dictated largely by the availability of resources and istherefore arbitrary.

The Sample

I analyze a complete record of commercial television stations in 15 broadcast markets

from 1961 to 1988. Data from 1953 to 1961 were also collected and used in the computation of

the organizational demography measures. The data on station characteristics were collected from

yearly volumes of the Television Factbook (subsequently entitled the Television and Cable

Factbook), an industry publication used as a reference work by industry members and television

advertisers.8 For each station in existence in a given year, the Factbook includes information on

ownership, technical facilities, network affiliations and executive personnel. Starting in 1961, the

Factbook includes circulation information for each station, indicating the number of households

reached by each station according to surveys by the market research firm Arbitron.9 Some of the

information in the yearbooks is culled by the publishers from the broadcast licenses on file with

the FCC. The stations themselves provide the remainder of the information, including the

personnel listings.

Data were collected on the characteristics of all commercial television stations in the 15

broadcast markets listed in Table 1; these are the 15 largest markets (in terms of the number of

television households) in the United States in 1988. These markets reached approximately 40%

of all viewers in the United States at that time. There are no obvious reasons to expect that the

demographic and ecological processes investigated vary by the size of the broadcast market.


Only data on local, commercial broadcast stations were collected; public and educational

television stations were excluded since the Factbook does not include information on the station

personnel or on the circulation levels of public television stations. In addition to other broadcast

television stations, a station also competes with cable franchises and (to a lesser extent) radio

stations. Data on cable operators and radio stations were not collected. The models presented

here therefore do no account for the full range of competitive influences that television stations

are subject to. Nonetheless, the main competitors faced by an individual station are the other

stations in its market (Eastman, Head and Klein 1989); the number of other commercial television

stations is a conservative measure of the competitive pressure faced by a station.

The hypothesized effects of overlap in the tenure distributions of television station

management teams will only be found if the behavior of television station managers can be

expected to affect firm performance. With respect to the outcome studied here – viewership –

managerial choices concerning programming and promotion are important determinants of

success. Since independent stations (i.e., those not affiliated with a major broadcast network) are

responsible for all of their own programming and promotion, it is clear that the management team

has an impact on station performance. However, the majority of stations are affiliates of the

major broadcast networks; affiliates receive over half of their programming from the networks.

Given this, what do (affiliate) station managers do? While the managers of network affiliates in

particular might seem to be little more than order-takers, they in fact play an important role in

responding to local competitive conditions (Eastman, Head and Klein 1989). Affiliate managers

remain responsible for local promotion, including community relations and tie-ins with local

events. Affiliate managers also choose programming for the parts of the schedule that are not


covered by network programming. This includes designing local news coverage and choosing

syndicated programming, and thus requires an understanding of local tastes. A station’s news

broadcast is an important source of viewer loyalty and brand identity in the local market. Affiliate

managers must also be effective in choosing programming that leads in and out of network feeds,

such that the local station is able to retain the audience generated by network programming. Thus

while the audience size of a network affiliate is in part determined by the network’s programming,

the decisions of local station managers are also important in drawing and retaining viewers

(Eastman, Head and Klein 1989).

Models of Organizational Growth

In order to test the hypotheses regarding the effects of mean team tenure and a

organization’s location in the tenure distribution, I estimate models of organizational growth. As

noted earlier, organizations that experience greater competition for resources will have lower

growth rates on average. I estimate models of growth in a station’s net weekly circulation,

defined as the total number of homes that tuned in to a station at least once during a given week.

Measures of organizational size should measure both the success of an organization in utilizing

environmental resources and the burden an organization places on the environmental carrying

capacity. The circulation data are measures of both an organization's success in exploiting

resource opportunities and an organization's impact on its environment. Station viewership is of

intense interest to advertisers (as well as the stations themselves), inducing firms to provide

widely accessible data on viewership.


Si,t%1

Sit

' S "&1it e

$xit%,i,t%1 (1)

ln(Si,t%1) ' "ln(Sit)%$xit%,i,t%1 (2)

We can express the growth of organizations as a function of an organization's size and a

number of covariates characterizing organizational and environmental characteristics:

where S is a time-varying measure of organizational size, " is an adjustment parameter that

indicates how growth rates depend on organizational size, and $ is a vector of parameters

characterizing the effects of organizational and environmental covariates.

If we take the log of Equation 1 and rearrange terms, we have the following log-linear

model suitable for estimation with linear methods:

The data is arranged in the form of a pooled cross-section/time series data set, with each

television station contributing a time series of observations. The length of each station's time

series may differ because of missing data, or because a station is founded or fails during the

observation period. To correct for autocorrelation, I use a fixed effects or least squares with

constants estimator (Balestra and Nerlove 1966; Tuma and Hannan 1984), which includes a

dummy variable for each station in the sample.

Measures

The tenure distribution of the top managers with each firm is the measure of central

interest here. At least three types of tenure might be used: time with the top management team

(team tenure), time with the organization (organizational tenure), and time in the industry


(industry tenure). Each of these measures can be found in previous research, sometimes in

combination. In the present study, I compare organizations in terms of their distributions of team

tenure. In part this choice is guided by practical considerations, since organizational and industry

tenure measures are considerably more difficult to collect. However, it is reasonable to believe

that differences in team tenure distributions will capture the processes of interest here. Clearly,

the reasoning laid out here suggests that two managers with similar organizational and industrial

tenures will have similar capabilities. Measures based of team tenure therefore potentially

underestimate the extent to which managers are similar. However, experiences as a member of a

top management team seem clearly relevant to the development of managerial capabilities. While

the capabilities of top managers are shaped by their entire career histories, their experiences as

members of a top management team may be particularly salient, both because they are more

recent and because it is only in that role that individuals have a responsibility for evaluating data

from the perspective of the firm’s overall interests (Daft and Weick 1984). Team tenure arguably

captures the experiences that are most relevant to the task at hand for top managers, namely the

ones related to leading the firm in a particular competitive context.

For each television station, the yearly Factbooks include a listing of top managerial

personnel. The Factbook asks each station to provide a list of the management personnel.

Inspection of the job titles suggests that they are quite standardized across organizations. A total

of 6,321 unique names were collected in the fifteen markets studied. The mean number of

employees declined steadily over time. In 1961, the mean number of team members listed was

9.2. The average team size declined gradually, eventually reaching a mean of 7.4 in 1987. The

decline in team sizes is largely attributable to the growth of independent stations, which have


10 This measure is potentially sensitive to outliers. In particular, the presence ofcompetitors far from the focal firm’s tenure will inflate the measure, independently of thedistribution of firms in the focal firm’s tenure neighborhood. However, this problem is mitigatedhere by using the log of mean tenure in Eq. 3; the range of log mean tenure is 3.5 years. Experimentation with alternative constructions of the distance measure (including the use of the

Dit'

jj…i

(xit&xjt)2

Nmt&1(3)

smaller staffs on average for a variety of reasons. For example, independent stations often do not

have news departments.

For each station, the tenure values are averaged across all of the team members in a given

year. It seems plausible that an additional year of average experience should have a greater

impact on performance for young teams than for older teams. Therefore, I use the natural

logarithm of mean team tenure in the models presented below; however, models using the

untransformed mean tenure resulted is substantively similar findings

The first measure of crowding is the average Euclidean distance between the focal firm’s

mean tenure and the mean tenure of each of its competitors. Euclidean distance is a commonly

used measure in studies of localized competition (Hannan, Ranger-Moore and Banaszak-Holl

1990; Baum and Mezias 1992, 1993). This measure captures the extent to which the focal

station’s mean tenure differs from that of its competitors; firms in more crowded regions of the

tenure distribution will have low values of this measure while isolated firms will have high

distance scores. The average distance score for firm i in year t is defined as

where xit is the (logged) measure of mean team tenure and N is the number of stations in the focal

firm’s market.10 Since competition in the television broadcasting industry is highly segmented by


inverse of mean tenure in Eq. 3) yielded substantively similar results.

"ijt'min(bit,bjt)&max(ait,ajt)

2Fit

(4)

Ait'

jj…i

"ijt

Nmt&1(5)

broadcast market, I only compare the mean tenure of a station to that of its direct competitors for

viewers. As a consequence, the distance measure in Equation 3 is computed by comparing a

station’s mean tenure to the mean tenure of other stations in the same market only.

I define each station’s tenure niche as a two standard deviation window centered on the

mean level of team tenure. Following McPherson (1983), I compare the tenure niches for each

pair of stations in a market to compute pairwise overlap scores. If we define ai as the lower

boundary of the niche (i.e., the mean tenure minus one standard deviation), and bi as the upper

boundary (i.e., the mean tenure plus one standard deviation), then the extent to which firm j

overlaps on the niche of firm i is given by:

If the numerator in this equation is less than zero, "ijt is set to zero. A crowding measure can then

be constructed by computing the average overlap across all other competitors in the focal firm’s

market:

Note that while the Euclidean distance measure defined in Equation 3 is symmetric (i.e., the

distance from firm i to j is the same as from j to i), the niche overlap measure in Equation 4

relaxes this assumption. The crowding experienced by a focal firm due to a given amount of


overlap will depend on the width of the firm’s niche — i.e., the heterogeneity of its tenure

distribution.

One feature of the crowding measures used here is that overlap between two firms is an

indication that they have tended to recruit at the same times. A firm that experiences a high

degree of crowding, in other words, has been unfortunate enough to recruit at the same times as

its competitors. If labor markets are very tight, such a firm may be forced to settle for managers

of lower quality; this would lead the average quality of recruits to be lower for firms in crowded

conditions. A negative relationship between crowding along the tenure distribution and growth

may therefore be due to inferior management skills, and not overlap in managerial capabilities.

However, it does not seem likely that this will occur, for a number of reasons. First, this scenario

depends heavily on the assumption that the supply of qualified managers is tight and unresponsive

to demand. Given that these stations are members of the highest status broadcast markets, it

seems reasonable to assume that they would be able to find qualified talent. Furthermore, the

construction of the crowding measures allows firms to overlap even if they did not recruit in the

exact same year. Finally, separate analyses (not shown here) suggest that the results presented

below are not spurious. For example, the number of managers hired by the other stations in a

market does not significantly affect individual-level turnover rates, a crude measure of the fit

between employer and employee. Also, inclusion of the same measure (aggregated to the team

level) in the growth models below does not affect the conclusions drawn with respect to the

effects of crowding.

Finally, there is some endogeneity in the relationship between tenure distributions and

population-level competitive processes (Haveman 1995). Births and deaths of organizations are


accompanied by the wholesale creation and destruction of jobs. By influencing the supply and

demand for labor, these processes affect mean tenure levels and tenure heterogeneity in existing

organizations. In a study of California thrifts, Haveman (1995) finds that population dynamics

indeed have complex effects on demographic characteristics; in particular, the impact of births and

deaths on a firm’s tenure distribution appears to depend on the size of the firm. However, the

impact of any such endogeneity would appear to be minimal in this study. First, the television

broadcasting industry experiences relatively little population turnover, as entry is regulated by the

FCC and firms rarely fail. Moreover, the number of employees varies little across firms. Finally,

it should be kept in mind that tenure distributions are also importantly shaped by the employment

and career decisions of employers and employees.

Control Variables

I control for a number of organizational and environmental characteristics in the growth

models. At the organizational level, I control for station age and the dispersion (standard

deviation) in team tenure. Most ecological studies of growth have found negative age

dependence in growth rates when controlling for size (e.g., Barnett 1994; Barron, West and

Hannan 1994), reflecting either the increasing obsolescence of an organization’s capabilities or an

increasing inability to respond rapidly to environmental changes. More importantly, station age

and team tenure are strongly correlated when stations are young, since the age of the station sets

an upper bound on the mean tenure measure.

The standard deviation in team tenure is included as a measure of the heterogeneity of the

top management team. Social integration arguments would suggest that growth will decline with


increases in heterogeneity as group cohesiveness suffers. On the other hand, the diversity of

perspectives in more heterogeneous teams may benefit the firm, particularly in rapidly changing

environments such as broadcast television.

Two measures of environmental conditions are included in the models. Given the local

nature of competition among television stations for viewers, it is best to specify these factors at

the level of broadcast markets. A natural measure of the carrying capacity in television broadcast

markets is the number of households with a television. Yearly estimates of the number of

television households are taken from the “spot television market” volumes of the Standard Rate

and Data Service. Finally, the models include a measure of market density, the number of

commercial television stations in a focal firm’s market.

I also include a number of control variables specific to the television industry. I control

for each station’s visual broadcast power. The increased resources that derive from membership

in ownership groups may also promote circulation growth. Finally, the performance of network

affiliates can in part be attributed to the performance of the network they are affiliated with;

affiliates suffer if a network’s prime time schedule is unpopular, and reap the benefits of successful

network programming strategies. To control for network affiliation, I include a separate dummy

variable for each of the major networks (ABC, CBS, and NBC) for each year a station is affiliated

with a given network. Finally, the models include two additional control variables. Data gap

indexes years that follow a missing year of data due to the unavailability of the Factbook. The

Same Network dummy is coded one in those rare instances where an affiliate is exposed to

competition from another local station affiliated with the same network.


Descriptive statistics for all variables are presented in Table 2. Because the fixed effects

estimator utilizes the within-station variance, Table 2 breaks down the total variance for each

variable into between and within components. Within-station bivariate correlations are presented

in Appendix A.

Results

The first column of Table 3 presents estimates from a baseline model of organizational

growth, including measures of organizational demography. As expected, the coefficient for

lagged circulation is significantly less than unity, indicating the growth rates decline with size.

This finding is in line with previous (disconfirming) tests of Gibrat’s law (Barron, West and

Hannan 1994; Barnett 1994; Baum and Mezias 1993). Growth rates also decline with

organizational age, membership in ownership groups and network affiliation. A station’s growth

rate is higher when the size of its broadcast market exceeds the historical average for that station,

but the number of stations in a market has no significant effect on growth rates.

The dispersion in team tenure does not have a significant effect on growth rates among

television stations. However, mean tenure has a highly significant negative effect on growth rates.

Since the estimate is from a fixed effects model, it indicates that a station’s growth rate is lower in

those periods when its mean team tenure exceeds its historical average. As the team tenure

increases, station performance suffers. This is consistent with the notion that increases in tenure

lead to increased strategic persistence and – in the face of environmental change – misalignment

with environmental demands (Finkelstein and Hambrick 1990).


The next model tests Hypothesis 1a by adding the Euclidean distance measure, which

measures (the lack of) crowding around a station’s location in the tenure distribution across firms.

As expected, this distance measure has a significant, positive effect on firm performance. In other

words, firms that are further removed from their competitors in terms of their level of mean tenure

have higher growth rates. Model 3 uses niche overlap as a measure of crowding and leads to the

same conclusion: a station’s average overlap with other stations along the tenure distribution has

a significant, negative effect on growth rates. These findings support the claim that a firm’s

competitive position is affected by its location in the tenure distribution across firms; the greater

the overlap in managerial experiences, the greater the competition between stations for resources

and the lower the growth rates.

Model 4 includes the interaction effect between mean team tenure and the Euclidean

distance measure. The coefficient estimate is significant and negative, supporting Hypothesis 2a.

In Model 5, Hypothesis 2b is also supported through a significant positive interaction effect

between niche overlap and mean team tenure. In order to ease interpretation of these effects,

Figure 1 graphs the multipliers of the growth rates produced by the combined effects of niche

overlap and mean team tenure, for three levels of overlap. (The combined effects of tenure and

distance are substantively the same.) Several things are apparent from this figure. First, for a

management team with low levels of tenure, there are strong advantages to being distant from

one’s competitors. In other words, short-tenured teams are more likely to be successful if their

competitors are clustered around high tenure levels. Conversely, teams with high levels of tenure

are at a disadvantage if they find themselves relatively isolated from their competitors – in other

words, if crowding in the market occurs around low levels of tenure. This isolation of long-


tenured management teams is a consequence of their experiencing low turnover rates while their

competitors replace managers more frequently. The firm’s (short-tenured) competitors are

therefore updating their strategies and understandings of the environment while the focal (long-

tenured) station is drifting out of alignment and relying on inappropriate managerial capabilities.

Note that at high levels of mean tenure, overlap may be preferable to non-overlap: if you are

going to rely on inappropriate managerial capabilities, better to have your competitors do so as

well.

As an example of the liabilities associated with the combination of high tenure levels and

distance from competitors along the tenure distribution, consider the case of WMUR, a station in

the Boston broadcast market. This station experienced no turnover in its four-person top

management team between 1964 and 1980, and recruited no new members until 1983. By the end

of this period, the mean team tenure at WMUR was over 20 years; the average across other

stations in the same market was never greater than four years. Thus the distance along the tenure

distribution between WMUR and its competitors grew; its overlap with other stations was zero

throughout the 1970s. The evidence suggests that WMUR pursued outmoded strategies during

this period. As an example, WMUR was best known for "Uncle Gus" Bernier, who divided his

time between hosting a children’s show, doing the weather (dressed as a gas station attendant due

to the program’s sponsorship by an oil company) and occasionally reading the news. WMUR

continued the “Uncle Gus Show” until 1981, long after other stations had abandoned similar

programming and developed more professional newsrooms (Grove 1987). Similarly, WMUR did

not make the transition to color broadcasts until the mid-1970s. Not surprisingly, in the early

1980s, WMUR had the lowest ratings in its market (Grove 1987).


11 If firms assemble management teams after changing strategy, levels of mean tenure willreflect the recency of strategic change, and the effect of tenure on growth would be a spuriouseffect of strategy. However, this ignores the fact that most strategic changes are conceived of andimplemented by the top management team itself, or with their consent; it is unlikely that most topmanagement teams would go along with strategies that required their own dismissal. This issupported empirically by the fact that dramatic year-to-year changes in mean tenure are rare inthis sample; in only 3% of station-year spells is the mean tenure less than half of the previousyears tenure.

The interaction effects may also result from mutual adjustment and collusion processes

between competitors. Consider the case of long-tenured firms with high levels of crowding. The

managers of these firms have, essentially, been interacting with each other over an extended

period of time. This history of interaction may facilitate tacit collusion as competitors become

better able to predict how other firms will react to different situations. The longer the managers

have been interacting – that is, the higher the tenure level – the easier such mutual adjustment

becomes. Note however, that this arrangement will only work for mutual benefit as long as none

of the parties experiences substantial turnover and crowding decreases. It is not possible to

adjudicate between these alternative scenarios with this data; however, future research may be

designed to address this question.

The interaction effects raise the possibility, however, that differences in tenure

distributions reflect the strategic positions of firms, and not vice-versa.11 For example, turnover

dynamics may differ systematically according to a firm’s competitive position as the relative

power of employers and employees shifts (Phillips 1998; Phillips and Sørensen 1999). In this

case, levels of crowding along the tenure distribution would reflect the firm’s competitive

positioning and the results in Table 3 would be spurious. I approach this problem by using

network affiliation status as a measure of the niche occupied by each station. The presence of the


major broadcast networks (ABC, CBS, NBC) effectively creates two distinct strategic groups

within most markets: network affiliates and independent stations. As a consequence of the

network programming they carry, network affiliates pursue generalist strategies in trying to appeal

to a broad spectrum of the television audience. Independent stations, by contrast, typically pursue

specialist strategies by targeting their programming at particular audience segments, for example

by broadcasting local sports teams. Network affiliates are more constrained in their programming

choices, but their generalist strategies and the resources provided by the networks also provide

them with more slack to cope with fluctuations in audience tastes (Hannan and Freeman 1989).

Network affiliation status therefore defines two types of programming niches — network

affiliate and independent station. I use this distinction to create two different niche overlap

measures: overlap with stations of the same network affiliation status as the focal station (“own

overlap”), and overlap with stations of a different network affiliation status as the focal station

(“other overlap”). These variables measure crowding along the tenure distribution while

controlling for each station’s programming niche. The “own overlap” variable is therefore a

measure of the extent to which stations occupying the same programming niche as the focal firm

are crowded around its position in the tenure distribution. Conversely, the “other overlap”

measure captures the extent to which stations of a different affiliation status are crowded around

the focal firm’s position in the tenure distribution.

The models in Table 4 examine the effects of these measures. In the first column of Table

4, we see that “own overlap” has the same pattern of effects as the overall crowding measure.

Growth rates decline as firms of the same network affiliation status as the focal firm are crowded

around its position along the tenure distribution. Furthermore, the interaction effect between


12 A t-test of the difference in the mean circulation levels of network affiliates andindependent stations confirms that network affiliates have higher circulation levels (t=12.96).

team tenure and “own overlap” is positive and statistically significant as before. The estimates in

the second column of Table 4 show that crowding by stations of a different network affiliation

status has no significant effect on the focal station’s growth rate. The final model in Table 4

includes both measures; again, the effects of crowding by stations of the same type of the focal

firm follow expectations, while the crowding by stations of a different type have no effect on

performance.

Finally, the robustness of the overlap effects in Table 4 suggest that the overlap measure is

not simply capturing size-localized competition effects (e.g., Hannan, Ranger-Moore and

Banaszak-Holl 1990). This is due to the fact that network affiliation status is an effective proxy

for size differences among television stations: independent stations are typically smaller than

network affiliates, in part because they have less original programming to offer viewers.12 The

results in Table 4 therefore suggest that even among firms that occupy similar positions in the size

distribution, overlap in managerial tenure distributions increases competitive pressures.

Discussion

Since a firm’s overall competitive positioning is to a large extent a reflection of the

decisions made by top managers, its position in the market for managerial talent has wide-ranging

consequences for its competitive position. The tenure distribution, I have argued, captures

qualitative differences in managerial capabilities. The estimates in Tables 3 and 4 provide strong

evidence to support the claim that competition increases with crowding along the tenure


distribution of top management teams. Thus the timing of turnover, relative to competitors,

shapes organizational outcomes.

These findings reinforce the central insights of organizational demography research,

namely that managerial capabilities are shaped by past experiences (as captured by demographic

characteristics), and that organizational behavior is affected by the distribution of these

capabilities. The goal of this paper, however, has been to demonstrate the ecological implications

of this insight. If it is true that managerial behavior is influenced by past experiences, then we

should expect commonality in behavior across firms to the extent that managers in these firms

have been exposed to similar environmental conditions. In this sense, an organization’s

demographic composition is a relational characteristic that defines its position with respect to

other organizations. Moreover, the impact of demographic characteristics on firm performance

appears to be mediated by a firm’s location in the demographic distribution across firms. A full

understanding of organizational demography effects requires an ecological perspective.

The findings in this paper also support and extend previous research on the effects of niche

overlap or crowding on organizational outcomes (McPherson 1983; Podolny, Stuart and Hannan

1996) and studies of localized competition more generally (Hannan, Ranger-Moore and

Banaszak-Holl 1990; Baum and Mezias 1992; Baum and Mezias 1993). The findings reinforce

the ecological notion that the differential distribution of organizations through resource space is a

central determinant of the ecology of competition. However, the focus on overlap in managerial

capabilities raises a number of unique and interesting theoretical issues. First, this focus highlights

the importance of ecological processes at multiple levels of analysis: the labor market is a primary

arena in which organizations and individuals meet, and the nature of this interaction has both


macro- and micro-level consequences. In the model I have developed here, managerial

capabilities are shaped by the experiences individuals have as they progress through their careers –

a progression which consists of moving between organizations situated in various competitive

contexts. At the same time, these capabilities play an important role in shaping the structure and

dynamics of organizational interrelationships. Future research is needed to explore this reciprocal,

jointly constitutive relationship. For example, how do different types of labor market structures –

including internal labor markets, occupational labor markets and professions (Althauser and

Kalleberg 1981) – affect the dynamics of interorganizational competition and the evolution of

populations? How do the dynamics of competition shape the formation of different labor market

structures?

Second, greater attention should be paid to other sources of similarity and differences in

managerial capabilities. As noted, this study is limited in that it only considers overlap in terms of

a narrow range of managerial experiences and does not consider similarity in organizational or

industry tenure. Other sources of isomorphism in managerial capabilities may not depend on

common work experiences, suggesting that other demographic attributes (such as age, education,

race, etc.) should be explored as well as sources of overlap among firms.

The interaction effects between mean tenure and the crowding measures raise a number of

interesting issues as well. The results suggest that the detrimental effects of crowding around a

firm’s position in the tenure distribution decline as the mean tenure of the team increases. In line

with these findings, I have argued that long-tenured management teams are particularly vulnerable

when their competitors primarily have lower levels of tenure, because the short-tenured firms are

able to exploit the lack of fit between the long-tenured teams capabilities and environmental


demands. Thus the relationship between the quality of organization-environment fit and a firm’s

life chances depends on the relative fitness of other competitors. This suggests more generally

that the impact of demographic characteristics on firm outcomes is mediated by the broader

context, particularly the demographic attributes of other firms. The exploration of such mediating

effects should be an important avenue for future research.

Conclusion

Human resources are unlike other firm resources in that they have volition. Ultimately,

the tenure distributions of top management teams are driven by turnover processes, and hence are

shaped in large part by the individual interests of managers pursuing their careers. The career

choices made by individuals result in structures that constrain organizations. While it is possible

for firms to influence turnover rates in various ways, employment relationships can ultimately be

terminated by either party. Changes in a firm’s tenure distribution – particularly relative to its

competitors – are thus a partially exogenous source of change in their competitive position and

life chances. Of course, there is some endogeneity in these career choices, as the propensity of

managers to leave (or be fired) may depend in part on the competitive positions of their

employers. Moreover, individual-level turnover rates depend on the top management team’s

tenure distribution; for example, turnover rates decrease with increases in the mean tenure of the

team and with demographic heterogeneity experienced during an individual’s tenure (Sørensen

1999).

This points to the complexities involved in studying ecological dynamics involving

multiple levels of analysis. To some, this may suggest steering clear of the complications raised


by incorporating managers into macro-level theories, yet these issues are present in other setting

as well. From a research perspective the uniqueness of human resources should not be overstated:

other resources can also be withdrawn at the behest of the resource provider. This points to the

importance of the structure of relationships that firms have with resource providers (Burt 1992)

and how changes in these relationships drive the structure and dynamics of competition.

An organization’s managerial capabilities reflect its interactions with the environment.

How an organization relates to this part of the environment plays a crucial role in shaping

organizational outcomes and behavior. In most open systems theories of organizations, managers

play a peripheral role; organizational decision makers are largely seen simply as an extension of

the organization, and little attention has been paid to the role that variations in managerial

capabilities play in determining organizational outcomes. For this reason, open systems theories

have been faulted for an anti-managerial bias, or for having “beheaded” organizations (Finkelstein

and Hambrick 1996). However, this should not lead us back to a closed-system view in which

managerial effects are conceptualized exclusively in terms of internal firm processes. The

arguments in this paper demonstrate that taking managerial discretion and strategic choice

seriously does not require a return to closed-system theories. Rather, open-systems theories must

see the characteristics of a firm’s management team as a product of its interaction with its

environment and see managerial capabilities as in part a reflection of a firm’s history of turnover

among top managers. The issues involved are complex; addressing them carefully should lead to

a substantial enrichment of organizational theory.


Works Cited

Althauser, Robert P., and Arne L. Kalleberg. 1981. “Firms, Occupations, and the Structure ofLabor Markets: A Conceptual Analysis.” Pp. 119-149 in Ivar Berg (ed.), SociologicalPerspectives on Labor Markets: New York: Academic Press.

Ancona, D.G., and David F. Caldwell. 1992. “Demography and Design: Predictors of NewProduct Team Performance.” Organization Science 3: 321-341.

Balestra, Pietro, and Marc Nerlove. 1966. “Pooling Cross Section and Time Series Data in theEstimation of a Dynamic Model: The Demand for Natural Gas.” Econometrica 34: 585-612.

Barron, David N., Elizabeth West and Michael T. Hannan. 1994. “A Time to Grow and a Timeto Die: Growth and Mortality of Credit Unions in New York City, 1914-1990.” AmericanJournal of Sociology 100: 381-421.

Barnett, William P. 1994. “The Liability of Collective Action: Growth and Change among EarlyAmerican Telephone Companies.” Pp. 337-354 in J.A.C. Baum and J.V. Singh (eds.),Evolutionary Dynamics of Organizations. New York: Oxford University Press.

Baum, Joel A. C., and Stephen J. Mezias. 1992. “Localized competition and organizationalfailure in the Manhattan hotel industry, 1898-1990.” Administrative Science Quarterly37: 580-604.

———. 1993. “Competition, Institutional Linkages, and Organizational Growth.” SocialScience Research 22: 131-164.

Blau, Peter M. 1977. Inequality and Heterogeneity: A Primitive Theory of Social Structure. New York: Free Press.

Blau, Peter M., and Richard A. Schoenherr. 1971. The Structure of Organizations. New York:Basic Books.

Burt, Ronald S. 1992. Structural Holes. Cambridge, MA: Harvard University Press.

Carroll, Glenn R., and J. Richard Harrison. 1998. “Organizational Demography and Culture:Insights from a Formal Model and Simulation.” Administrative Science Quarterly 43:637-667.

Cyert, Richard M., and James G. March. 1963. A Behavioral Theory of the Firm. EnglewoodCliffs, NJ: Prentice-Hall.


Daft, Richard L., and Karl E. Weick. 1984. “Toward a Model of Organizations as InterpretiveSystems.” Academy of Management Review 9: 284-295.

Eastman, Susan Tyler, Sydney W. Head, and Lewis Klein. 1989. Broadcast/CableProgramming: Strategies and Practices. Belmont, CA: Wadsworth Publishing.

Eisenhardt, Kathleen M., and Claudia Bird Schoonhoven. 1990. “Organizational Growth:Linking Founding Team, Strategy, Environment, and Growth among U.S. SemiconductorVentures, 1978-1988.” Administrative Science Quarterly 35: 504-29.

Finkelstein, Sidney, and Donald C. Hambrick. 1990. “Top Management Team Tenure andOrganizational Outcomes: The Moderating Role of Managerial Discretion.” Administrative Science Quarterly 35: 484-503.

———. 1996. Strategic Leadership: Top Executives and Their Effects on Organizations. Minneapolis: West.

Gelatkanycz, Marta A., and Donald C. Hambrick. 1997. “The External Ties of Top Executives:Implications for Strategic Choice and Performance.” Administrative Science Quarterly42: 654-681.

Gibrat, Robert. 1931. Les inégalites économiques. Paris: Sirey.

Glick, William H., C.C. Miller, and George D. Huber. 1993. “The Import of Upper-EchelonDiversity on Organizational Performance.” Pp. 176-214 in G.D. Huber and W.H. Glick(eds.), Organizational Change and Redesign. New York: Oxford University Press.

Grove, Lloyd. 1987. “A Shift in the Mediascape: Scrappy TV Station Influences Coverage ofNew Hampshire Primary.” The Washington Post, December 8, 1987. Page A6.

Gunz, Hugh P., and R. Michael Jalland. 1996. “Managerial Careers and Business Strategies.” Academy of Management Review 9: 193-206.

Hambrick, Donald C. 1994. “Top Management Groups: A Conceptual Integration andReconsideration of the ‘Team’ Label.” Pp. 171-214 in B.M. Staw and L.L. Cummings(eds.), Research in Organizational Behavior. Greenwich, CT: JAI Press.

Hambrick, Donald C., Theresa Sung Cho and Ming-Jer Chen. 1996. “The Influence of TopManagement Heterogeneity on Firms’ Competitive Moves.” Administrative ScienceQuarterly 41: 659-684.

Hambrick, Donald C., and Phyllis A. Mason. 1984. “Upper Echelons: The Organization as aReflection of Its Top Managers.” Academy of Management Review 9: 193-206.


Hannan, Michael T., and Glenn R. Carroll. 1992. Dynamics of Organizational Populations. NewYork: Oxford University Press.

Hannan, Michael T., and John Freeman. 1977. “The Population Ecology of Organizations.” American Journal of Sociology 82: 929-964.

———. 1989. Organizational Ecology. Cambridge, MA: Harvard University Press.

Hannan, Michael T, James Ranger-Moore, and Jane Banaszak-Holl. 1990. “Competition and theEvolution of Organizational Size Distributions.” Pp. 246-268 in Jitendra V. Singh (ed.),Organizational Evolution: New Directions. Newbury Park: Sage.

Haveman, Heather A. 1995. “The Demographic Metabolism of Organizations: IndustryDynamics, Turnover, and Tenure Distributions.” Administrative Science Quarterly 40:586-618.

Katz, Ralph. 1982. “The Effects of Group Longevity on Project Communication andPerformance.” Administrative Science Quarterly 27: 81-104.

Lant, Theresa K., and Stephen J. Mezias. 1992. “An Organizational Learning Model ofConvergence and Reorientation.” Organization Science 3: 47-71.

March, James G., and Herbert A. Simon. 1958. Organizations. New York: Wiley.

McNeil, Kenneth, and James D. Thompson. 1971. “The Regeneration of Social Organizations.” Administrative Science Quarterly 36: 624-637.

McPherson, J. Miller. 1983. “An Ecology of Affiliation.” American Sociological Review 48:519-535.

McPherson, J. Miller, Pamela Popielarz, and Sonja Drobni…. 1992. “Social Networks andOrganizational Dynamics.” American Sociological Review 57: 153-170.

Meyer, John W., and Brian Rowan. 1977. “Institutionalized Organizations: Formal Structure asMyth and Ceremony.” American Journal of Sociology 83: 340-63.

Mundlak, Yair. 1978. “On the Pooling of Time Series and Cross Section Data.” Econometrica:46: 69-85.

O’Reilly, Charles A., David F. Caldwell, and William P. Barnett. 1989. “Work GroupDemography, Social Integration, and Turnover.” Administrative Science Quarterly34:21-37.


O’Reilly, Charles A., Richard C. Snyder, and Joan N. Boothe. 1993. “Effects of OrganizationalDemography on Organizational Change.” Pp. 147-175 in G.D. Huber and W.H. Glick(eds.), Organizational Change and Redesign. New York: Oxford University Press.

Pfeffer, Jeffrey. 1983. “Organizational Demography.” Pp. 299-357 in L.L. Cummings and BarryM. Staw (eds.), Research in Organizational Behavior (Volume 5). Greenwich, CT: JAIPress.

———. 1997. New Directions for Organizational Theory: Problems and Prospects. NewYork: Oxford University Press.

Phillips, Damon J. 1998. “The Promotion Paradox: The Relationship Between OrganizationalMortality and Employee Promotion Chances in Silicon Valley Law Firms, 1946-1996.” Unpublished manuscript, University of Chicago.

Phillips, Damon J., and Jesper B. Sørensen. 1999. “Promoting from a Position of Weakness:Firm Strength and the Rate of Internal Promotion.” Unpublished manuscript, Universityof Chicago.

Porac, Joseph F., and Howard Thomas. 1990. “Taxonomic Mental Models in CompetitorDefinition.” Academy of Management Review: 224-240.

Porter, Michael E. 1980. Competitive Strategy. New York: Free Press.

Podolny, Joel M., Toby E. Stuart, and Michael T. Hannan. 1996. “Networks, Knowledge, andNiches: Competition in the Worldwide Semiconductor Industry, 1984-1991.” AmericanJournal of Sociology 102: 659-689.

Ryder, Norman B. 1965. “The Cohort as a Concept in the Study of Social Change.” AmericanSociological Review 30: 843-861.

Smith, Ken G., Ken A. Smith, Judy D. Olian, Henry P. Sims Jr., Douglas P. O’Bannon and JudithScully. 1994. “Top Management Team Demography and Process: The Role of SocialIntegration and Communication.” Administrative Science Quarterly 39: 412-438.

Sørensen, Jesper B. 1999. “Changes in Group Composition and Turnover: A LongitudinalStudy.” Forthcoming, American Sociological Review.

Standard Rate and Data Service Spot Television Source. Various years. Willemette, IL:Standard Rate and Data Service.

Television Factbook. Various years. Washington, DC: Television Digest, Inc. (Also entitledTelevision and Cable Factbook).


Tuma, Nancy Brandon, and Michael T. Hannan. 1984. Social Dynamics: Models and Methods.San Francisco: Academic Press.

Table 1: Size Measures of Sampled Broadcast Markets, 1961-1988

Stations Television Households

Market 1961 1988 1961 1988

New York 8 13 4,431,045 6,686,700

Los Angeles 7 15 2,214,861 4,609,200

Chicago 5 13 2,153,474 3,050,700

Philadelphia 4 9 1,772,729 2,564,700

San Francisco 5 14 1,034,656 2,133,000

Boston 5 12 1,231,557 1,980,500

Detroit 3 6 1,174,445 1,614,100

Dallas-Fort Worth 4 9 620,869 1,584,700

Washington D.C. 4 7 716,273 1,550,700

Houston 3 7 504,514 1,469,700

Cleveland 4 12 1,167,944 1,379,400

Atlanta 3 7 464,179 1,207,500

Minneapolis-St. Paul 5 8 640,771 1,186,900

Seattle-Tacoma 6 12 475,530 1,180,900

Miami-Ft. Lauderdale 4 9 381,074 1,156,000

Table 2: Descriptive Statistics

Variable Mean Overall F Between F Within F Spells Stations

Net Circulation (log) 13.579 1.275 1.501 0.419 2,218 150

ABC 0.196 0.397 0.306 0.156 2,474 164

CBS 0.169 0.374 0.296 0.112 2,474 164

NBC 0.180 0.384 0.311 0.135 2,474 164

Same Network Dummy 0.181 0.385 0.288 0.203 2,474 164

Station Age 18.101 11.033 9.924 6.810 2,474 164

Visual Power (000) 0.963 1.382 1.657 0.490 2,474 164

Ownership Group Member 0.711 0.453 0.399 0.291 2,474 164

Market TV Households (log) 14.318 0.665 0.611 0.188 2,474 164

Market Density 7.962 2.949 2.641 1.685 2,474 164

Tenure Dispersion (F) 3.349 2.419 2.030 1.525 2,474 164

Log Mean Team Tenure 1.281 0.783 0.761 0.527 2,458 163

Distance (Dit) 0.404 0.217 0.135 0.188 2,458 163

Overlap (Ait) 0.478 0.227 0.197 0.182 2,458 163

Circulation Ratio 0.946 0.084 0.110 0.024 2,148 145

Own Tenure Overlap 0.329 0.218 0.180 0.138 2,458 163

Other Tenure Overlap 0.451 0.342 0.276 0.230 2,272 163

Table 3: Effects of Team Tenure and Crowding on Circulation GrowthModel

Variable (1) (2) (3) (4) (5)

Lag Net Circulation 0.720† 0.722† 0.723† 0.723† 0.724†(0.013) (0.013) (0.013) (0.013) (0.013)

ABC -0.058† -0.057† -0.056† -0.059† -0.055†(0.022) (0.022) (0.022) (0.022) (0.022)

CBS -0.117† -0.118† -0.118† -0.120† -0.118†(0.031) (0.031) (0.031) (0.031) (0.031)

NBC -0.038 -0.037 -0.034 -0.037 -0.035(0.026) (0.025) (0.026) (0.025) (0.026)

Year 0.005† 0.005† 0.005† 0.005† 0.005†(0.002) (0.002) (0.002) (0.002) (0.002)

Data Gap 0.020† 0.022† 0.021† 0.024† 0.022†(0.010) (0.010) (0.010) (0.010) (0.010)

Same Network Dummy -0.014 -0.015 -0.015 -0.014 -0.015(0.017) (0.017) (0.017) (0.017) (0.017)

Log Station Age -0.059† -0.057† -0.058† -0.055† -0.060†(0.020) (0.020) (0.020) (0.020) (0.020)

Visual Power (000) 0.035† 0.036† 0.035† 0.037† 0.035†(0.009) (0.009) (0.009) (0.009) (0.009)

Ownership group member -0.041† -0.042† -0.042† -0.043† -0.041†(0.013) (0.013) (0.013) (0.013) (0.013)

TV Households (log) 0.220† 0.213† 0.220† 0.222† 0.227†(0.040) (0.040) (0.040) (0.041) (0.040)

Market Density -0.004 -0.004 -0.007* -0.004 -0.005(0.004) (0.004) (0.004) (0.004) (0.004)

Team Tenure Dispersion (F) 0.005 0.004 0.004 0.003 0.004(0.003) (0.003) (0.003) (0.003) (0.003)

Log Mean Team Tenure -0.035† -0.033† -0.037† -0.005 -0.056†(0.010) (0.010) (0.010) (0.018) (0.014)

Distance (Dit) 0.051† 0.119†(0.021) (0.042)

Log Mean Team Tenure*Dit -0.048*(0.025)

Overlap (Ait) -0.047‡ -0.109†(0.024) (0.037)

Log Mean Team Tenure*Ait 0.057‡(0.026)

F-test relative to Model 1 5.67 3.99 4.63 4.34

df 1;1940 1;1940 2;1939 2;1939

p > F 0.01 0.05 0.01 0.05

Estimates from fixed effects models estimated on 137 stations and 2,092 station-year spells. Standard errors in parentheses.† p < 0.01 (one-sided) ‡ p < 0.025 (one-sided) * p < 0.05 (one-sided)

Table 4: Effects of Overlap as Computed by Network Affiliation Status

Model

Variable (6) (7) (8)

Lag Net Circulation 0.722† 0.709† 0.710†(0.013) (0.014) (0.014)

ABC -0.062† -0.064‡ -0.064‡(0.022) (0.029) (0.029)

CBS -0.124† -0.110† -0.111†(0.031) (0.042) (0.042)

NBC -0.044* -0.048 -0.053(0.026) (0.033) (0.033)

Year 0.005† 0.004‡ 0.004‡(0.002) (0.002) (0.002)

Data Gap 0.022‡ 0.025‡ 0.026‡(0.010) (0.011) (0.011)

Same Network Dummy -0.015 -0.014 -0.016(0.017) (0.022) (0.022)

Log Station Age -0.055† -0.047‡ -0.042*(0.020) (0.022) (0.022)

Visual Power (000) 0.033† 0.034† 0.032†(0.009) (0.009) (0.009)

Ownership group member -0.042† -0.042† -0.045†(0.013) (0.014) (0.014)

Television Households (log) 0.232† 0.263† 0.274†(0.040) (0.047) (0.048)

Market Density -0.004 -0.004 -0.004(0.004) (0.004) (0.004)

Team Tenure Dispersion (F) 0.003 0.002 0.001(0.003) (0.003) (0.003)

Log Mean Team Tenure -0.056† -0.030‡ -0.042†(0.012) (0.013) (0.014)

Own Tenure Overlap -0.121† -0.144†(0.046) (0.060)

Log Mean Team Tenure*Own Tenure Overlap 0.089† 0.096†(0.030) (0.039)

Other Tenure Overlap -0.029 0.029(0.032) (0.039)

Log Mean Team Tenure*Other Tenure Overlap -0.016 -0.047*(0.022) (0.025)

F-test relative to Model 1 4.44 2.94 3.11

df 2;1939 2;1759 4;1757

p > F 0.01 0.05 0.01

Estimates from fixed effects models estimated on 137 stations and 2,092 station-year spells. Standard errors in parentheses.† p < 0.01 (one-sided) ‡ p < 0.025 (one-sided) * p < 0.05 (one-sided)

0HDQ RYHUODS �� V�G�

0HDQ RYHUODS

0HDQ RYHUODS

�� V�G�

Mul

tiplie

r

Mean Team Tenure0 5 10 15 20

.85

.9

.95

1

Figure 1 Combined Effects of Tenure and Overlap on the Growth Rate (Predicted values from Model 5)

Appendix A: Within-Station Bivariate Correlation Matrix

Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1. Net circulation 1.000

2. ABC -0.189* 1.000

3. CBS -0.204* 0.280* 1.000

4. NBC -0.153* 0.274* 0.451* 1.000

5. Same Network -0.155* 0.366* 0.245* 0.364* 1.000

6. Station Age (log) 0.693* -0.171* -0.146* -0.115* -0.156* 1.000

7. Visual Power 0.405* -0.074* -0.179* -0.008 -0.040 0.422* 1.000

8. Ownership Group 0.176* -0.021 -0.037 0.079* 0.065* 0.277* 0.116* 1.000

9. TV Households 0.563* -0.114* -0.086* -0.034 -0.132* 0.726* 0.195* 0.309* 1.000

10. Market Density 0.417* -0.092* -0.078* -0.067* -0.167* 0.639* 0.183* 0.210* 0.672* 1.000

11. Tenure Dispersion 0.304* -0.172* -0.074* -0.036 -0.017 0.469* 0.168* 0.117* 0.412* 0.330* 1.000

12. Log Tenure 0.299* -0.184* -0.119* -0.070* -0.060* 0.424* 0.192* 0.031 0.170* 0.143* 0.637* 1.000

13. Distance (Dit) -0.302* 0.056* 0.063* 0.080* 0.107* -0.401* -0.174* -0.073* -0.277* -0.256* -0.139* -0.234* 1.000

14. Overlap (Ait) 0.174* 0.040 0.012 0.027 -0.015 0.171* 0.089* 0.038 0.100* -0.133* -0.079* 0.001 -0.467* 1.000

15. Own Overlap 0.171* 0.005 -0.006 -0.001 -0.046* 0.203* 0.022 0.048* 0.217* 0.140* 0.020 0.012 -0.372* 0.634* 1.000

16. Other Overlap 0.208* 0.022 0.001 -0.025 -0.039 0.283* 0.097* 0.061* 0.195* 0.077* 0.007 0.047* -0.500* 0.744* 0.386* 1.000

Note: All correlations are computed using within-station mean-deviated data, as in the fixed effects models.* p < .05 (two-sided)

Documents

The Ecology of Organizational Demography: Managerial Tenure …sorensen/EOD.pdf · 2006. 7. 12. · The Ecology of Organizational Demography 4 2 Among the voluntary organizations