1

HETEROGENEITY SHIFTS DUE TO MEMBER

MOBILITY: NEW INSIGHTS ON DIVERSITY AND

PERFORMANCE

August 14, 2005

Johannes M. PENNINGS Filippo Carlo WEZEL

Department of Management Wharton School

Locust Walk Suite 2000 Steinberg Hall-Dietrich Hall

Phone +1 215 898 7755 Pennings@wharton.upenn.edu

j.m.pennings@uvt.nl

Department of Organization and Strategy

Tilburg University P.O. Box 90153, 5000 LE Phone +31 13 466 3260

f.c.wezel@uvt.nl

Acknowledgements The authors contributed equally. Much of the original data was collected as part of a research project (with financial support from the Limpberg Institute) conducted by Willem Bujink, Steven Maijoor, Arjen van Witteloostuijn, and Maurice Zinken, NIBOR, Department of Economics, Maastricht University, the Netherlands. A special thank goes to Rahul Patel, Drexel University, who provided excellent data management assistance. Gino Cattani, initially involved in the project, greatly contributed to the development of this paper. We also thank Paul Almeida, Sigal Barsade, Donald Hambrick, Jackson Nickerson, Gordon Walker, the participants of the 19th 2003 EGOS conference, the 2004 Organization Science Winter Conference, and the 2004 Academy of Management. All errors remain our responsibility.

2

HETEROGENEITY SHIFTS DUE TO MEMBER MOBILITY:

NEW INSIGHTS ON DIVERSITY AND PERFORMANCE

ABSTRACT Organizational demography is often invoked as a driver of firm performance. Heterogeneity

taxes the collective decision-making of members, undermines the firm’s esprit de corps and

impedes its effective functioning. Three main limitations have characterized the literature on

demographic diversity. First, the theory remains inherently static. Second, its empirical

assessment have been confined short run predictions. Third, the consequences of within-firm

organizational demography for between-firm competition have rarely been explored. This

paper addresses these shortcomings by focusing on inter-firm mobility to detect the intra

and inter-firm implications of demographic diversity on long run performance – which we

measure in terms of hazard of organizational dissolution. At the organizational level, we do

not find seemingly symmetrical effects in that an increase in heterogeneity is harmful, while a

reduction is not; at the inter-firm level the up-tick versus reduction in demographic overlap

among firms does reveal a symmetric effect.. By and large, inter-firm effects appear more

robust than the intra-firm consequences in predicting long run organizational performance.

Key words: Organizational Demography; Inter-firm Mobility; Competitive

Interdependence.

3

INTRODUCTION

The study of organizational demography in general, and that of top management

teams in particular, has often been examined as a major precursor to strategic positioning

and organizational outcomes (e.g., Finkelstein and Hambrick 1996; Hambrick et al. 1996).

Similarly, turnover, that is, exit and entry of members as additions or depletions of human

capital has received a good deal of attention as determinants of firm performance (e.g.

Pfeffer, 1994; Harris and Helfat, 1997). Demography and member mobility have rarely been

combined, however, in some additive or interactive way to explain organizational outcomes.

The investigation of the long-term performance (e.g., survival) implications of diversity, both

at the firm and the inter-firm levels, due to member mobility is the theme of the present

empirical study.

Specifically, we focus on three under-explored issues in organizational demography

research: (i) the dynamics of demographic diversity variations within firms, (ii) the influence

of demographic variation on the competitive overlap among firms, and (iii) the long term

consequences of the abovementioned intra and inter-firm demographic variations. First,

much of the research to date investigating the performance consequences of demographic

diversity is static and remains silent on shifts or discontinuous changes diversity due to

absorption of new members. Empirical research has focused on the study of the levels (i.e.,

stock) of diversity, disregarding that on going recruitment may reduce or increase the

current stock. Researchers have noticed that focusing demographic diversity flows allows a

more appropriate truthful investigation of the underlying mechanisms of the theory (e.g.,

Carroll and Harrison, 1998). A dynamic theory should then focus on the impact of

demographic diversity variations (either positive or negative) on organizational performance.

In the present paper examines inter-firm mobility of key members (e.g., CEOs, top

managers) as shocks to the established levels of diversity.

4

Second, firms in fact do not exist in a vacuum but are part of a sector, market and

strategic group. Inter-firm level research has not been paired with demographic aspects even

as one would presume that shifts in heterogeneity would have “implications for firm

performance beyond its consequences for internal group processes” (Sørensen, 1999b: 718).

Much of the pertinent work to date has dealt with the composition of top management

teams ( teams) As “managers operate on mental representations of the world and those

representations are likely to be of historical environments rather than current ones” (Kiesler

and Sproull, 1982: 557), overlapping management teams demographic attributes are likely to

proxy strategic similarity among competitors. For instance, if firms are similar

demographically, their managers are likely to exhibit similar mental models and thus to

converge in decision-making (Pegels et al., 2000). Thus, several inferences regarding

demographic variables within firms could also be used to study the implications of

competitive overlap between them. In this case too, the between-firm consequences of

overlapping managerial characteristics are likely to vary depending on the type of variation

(either positive or negative) induced by the mobility event.

Third, top management team research has mainly focused on short-run outcomes as

proxied by accounting measuressuch as ROI, failing to include long-run organizational

performance measures such as growth and survival. To move towards a dynamic theory of

demographic diversity, longitudinal research designs investigating the long-term implications

of variations in demographic heterogeneity are needed (Boone, Wezel and van

Witteloostuijn, 2006) Although Hambrick and Mason’s model (1984) speculated on the

impact of within-firm diversity on long run performance, empirical research has rarely

investigated this issue (for two exceptions see Hambrick and D’Aveni, 1988 and 1992). Note

that diversity might either hamper current coordination and alignment of members seeking a

common frame of reference or joint decisional commitment. It might also have long term

5

repercussions in that heterogeneous teams are less likely to settle into a set of lasting,

persistent routines, practices and a collectively ingrained body of knowledge.

We claim that demographic diversity variations may affect organizational survival

also by reducing or augmenting competitive overlap among firms. A set of teams that differ

significantly form each other are more likely to diverge from each other and will focus on

rather different strategic issues and challenges. Consider in fact that ecological theories of

organizations inform us that competition unfolds when overlap (measured along different

characteristics among which socio-demographic ones, see e.g., Rotolo and McPherson, 2001)

increases.

The present paper elaborates on these three points to theorize on the survival

consequences of intra- and inter-firm demographic of demographic variations of Dutch

accounting firms over the period 1880-1986. The paper is organized as follows. The next

section presents the theory. Section 3 describes the empirical setting, the data, and the

independent and control variables, respectively. Section 4 describes the model and the

method we employ to test the hypotheses. Results are shown in Section 5. The discussion of

the main implications of the analysis and the conclusions are presented in Section 6.

2. THEORY Organizational Implications

The focus on “people” as a critical component of firms is hardly new (e.g., Cyert and

March, 1963; Pfeffer, 1994). However, firm membership embodies a configuration of skills,

experience, and social contacts that are specialized towards the market segments or sectors

in which a firm operates. Differences among members partly reflect differences in their

backgrounds, such as prior career development across firms. This diversity matters because

it can at once hinder or enhance organizational performance. For example demographic

differences among the members might challenge the alignment to such a degree that much

6

energy gets diverted form the tasks at hand, and that substandard outcomes on a variety of

performance criteria ensues. Diversity challenges coordination and alignment of the

membership. Diveristy also produces differences in cognitive perspectives and behavioural

dispositions that render the firm strongly embedded in its strategic position and long term

trajectory. The short term and long term implications of team diversity formed one of the

key motivations of this paper. .

Research on upper echelons (Hambrick and Mason, 1984) has elaborated on the

diversity-performance link under the assumption that organizations mirror the values, goals,

and experiences of their pre-eminent members. Managers’ demographic profiles enter into

the performance argument because their characteristics are presumed to occasion

psychological dispositions and subsequent strategic choices (e.g., Hambrick and Mason,

1984). Characteristics such as gender, age, and tenure affect managers’ decisions and actions

through three different filters (see Hambrick and Mason, 1984). First, their background

delimits the problems and information by which their attention is attracted. Second, selective

perception occurs because managers devote disproportionately more attention to the stimuli

in their field of vision. Finally, the information they receive is filtered through their cognitive

lenses. The degree of heterogeneity in demographic characteristics typically amounts to a

“proxy for cognitive heterogeneity, representing innovativeness, problem-solving abilities,

creativity, diversity of information sources and perspectives, openness to change and

willingness to challenge and be challenged” (Finkelstein and Hambrick, 1996: 125). As a

result, teams with high variability in their demographic attributes typically have different

schemata or ways of seeing the world (e.g., Michel and Hambrick, 1992).

The literature on managerial cognition supports this claim, as documented by

empirical studies regarding the influence of managerial cognition on strategic decision-

making (e.g., Tripsas and Gavetti, 2000). A common conclusion of this research holds that

“managerial work experience shapes managerial cognition” (Adner and Helfat, 2003: 1022).

7

Demographic heterogeneity enters into the fray because divergent mental frames produce

cognitive and attitudinal heterogeneity among organizational members by alternatively

interfering with intra-group communication (e.g., Lawrence 1997; Zenger and Lawrence,

1989); exacerbating internal conflict (Jehn et al., 1999); increasing the time required to reach

decisions (Hambrick, Cho and Chen, 1996); lowering adherence to budget and schedule

(Ancona and Caldwell, 1992) or decreasing strategic consensus (Knight et al., 1999).

Furthermore, since tenure heterogeneity diminishes social integration, organizations might

also experience higher turnover rates (e.g., O’Reilly et al., 1989; Wiersema and Bird, 1993).

Recent reviews of the literature have underscored the multifaceted nature of the

relationship between demographic heterogeneity and performance (Guzzo and Dickson,

1996). A positive heterogeneity effect was found for highly complex, uncertain tasks, such as

those facing a top management team, for which “informational diversity should theoretically

be more beneficial than in routine tasks” (Barsade et al., 2000: 809). Others studies found a

positive relation between managers’ diversity and (firm or group) performance for tasks

requiring creative problem solving and innovation (e.g., Bantel and Jackson, 1989;

Eisenhardt and Schoonhoven, 1990). While this discrepancy in research on the effects of

team diversity on performance reflects the more general issue of whether demographic

variables are good proxies for cognitive variables in predicting team outcomes (Lawrence,

1997; Kilduff et al., 2000), the specific outcome of interest might partly explain these

seemingly conflicting findings.1 Most of the extant research, however, has yielded results

slanted towards a negative conclusion: demographic heterogeneity is detrimental to firm

performance (e.g., Smith et al., 1994; Keck, 1997; Boone et al., 2004). That research has also

bene largely cross sectional, thus precluding insights into shifts in diversity. Such shifts

should be revelatory in their effect on performance, representing a quasi experimental

manipulation. 1 It is worth noticing that the setting in which we carry out our test is marked by relevant routine based problem solving (see Maister, 1993).

8

An important yet unanswered question concerns, therefore, the “variations in the

demographic composition of groups” (Williams and O’Reilly III, 1998: 120). Answering this

question would help clarify the nature of the relation between demographic heterogeneity

and performance – whether at the group or the firm level. Although the investigation of the

diversity-performance relation has involved levels (i.e., stock) of heterogeneity, the theoretical

rationale for its harmful effects pertains to the costs of recruitment into the firm (i.e., inflows)

(see Pfeffer, 1983). We agree with Carroll and Harrison (1998: 658) that “a major source of

this diversity comes from the disruption by newly entering individuals.” As heterogeneity

variations produce differences in social integration, communication, and strategic consensus,

the recruitment of new members winds up affecting (increasing or reducing) such

differences. Such recruitment effects occur because of the correspondence between

demographic and cognitive heterogeneity. A recruit’s propensity to embrace the firm’s values

and norms is proportional to the demographic fit with the recipient organization. Unless

newcomers overlap demographically with the roster of people they are joining, their

collective ability to function effectively becomes compromised.2

Theoretical and empirical research in organizational demography of top management

teams has already pointed out that the demographic diversity engenders a downward spiral

of failure (Hambrick and D’Aveni, 1992; Stewman, 1988). Departing from the above

evidence that management team heterogeneity reduces short-term performance,

compositional deterioration will follow due to the inability of attracting new talents.

Retaining valuable executives becomes increasingly difficult because stars are likely to exit

when organizations perform poorly (Groysberg and Nanda, 2002). Increases in demographic

diversity severely threaten the long-term effectiveness of teams. Consider, for instance, that

2 A vast literature initiated by Schneider (1987) examines how top management teams – through ‘cycles’ of attraction, selection and attrition – become increasingly homogeneous. That is because social groups, such as top management teams, have the tendency to reproduce themselves by the selective recruitment of similar people and by facilitating the turnover of dissimilar ones, a process labeled “homosocial reproduction” by Kanter (1977). Adopting this perspective, Boone et al., 2004 for instance have shed light on the short run negative consequences of avoiding this social norm.

9

within-group cooperation is sustainable when people trust each other and deviance from

reciprocity can easily be monitored and sanctioned (Campbell, 1994). Group heterogeneity

however impedes trust and reciprocal altruism in groups (Ruef et al., 2003; Boone, Wezel and

van Witteloostuijn, 2006). Last but not least, a “team of divergent composition may be seen

as incompetent or ill-suited […] causing stakeholders to withdraw or restrict their support

for the organization” (Hambrick and D’Aveni, 1992: 1463). All in all, positive variations of

demographic diversity may trigger a cascade of unforeseeable events increase the risk of

organizational failure. This line of reasoning remains consistent with the insights of

Hambrick and Mason (1984 see Figure 2: 198; see also the recent review of Carpenter et al.,

2004: 751). Because socialization processes, communication and trust depend on the

matching between new entries and the existing team, then new member inflows that

augment group heterogeneity increase the risk organizational failure; conversely, inflows that

further homogenize such a group are beneficial for survival. Accordingly, we hypothesize:

H1a: When key member inflows raise demographic heterogeneity, the hazard of

organizational failure increases.

H1b: When key member inflows decrease demographic heterogeneity, the hazard of

organizational failure declines.

Competitive Implications

Demographic variables however inform us on the mental models of decision makers

that do play a key role in shaping competitive relationships. Mental models have been

demonstrated to influence the way managers perceive their environment (e.g., Porac et al.,

1989; Reger and Huff, 1993) and the strategic actions they take (Thomas, Clark and Gioia,

1993). Mental models can be inferred from demographic variables because managerial

knowledge structures are formed on the basis of differences in origins – e.g., experiences or

10

demograp[hic background (Walsh, 1995). Pfeffer (1983) for instance suggests that

individuals from similar age cohorts exhibit similar values and beliefs. Demographic

characteristics are also critical in influencing the strategic positioning of firms (Finkelstein

and Hambrick, 1990); their degree of diversification (Michel and Hambrick, 1992), and

internationalization (Carpenter and Fredrickson, 2001; Sanders and Carpenter, 1998); and

the propensity, speed, and scope of their competitive response (Hambrick, Cho and Chen,

1996). Furthermore, long-tenured managerial teams follow more persistent strategies and

select strategies committed to the status quo (Hambrick et al., 1993; Finkelstein and Hambrick,

1990). As the profiles of senior organizational members influence the selection of the firms’

prevailing patterns of actions, their demographic characteristics “provide vital information

on a firm’s preference for environmental niches to compete and on the likelihood of success

in the chosen market niches” (Pegels et al., 2000: 914).

A theory concerned with the impact of inter-firm mobility on demographic diversity

and firm performance should then take into account that firms are embedded in a set of

competitively interdependent peers. While it is widely understood that firms do not operate

in a social vacuum, “organizational demographers attribute no causal or mediating force to

the demographic characteristics of other organizations” (Sørensen, 1999b: 714). As a result,

the transfer of people inevitably affects the degree of demographic similarity among firms

and investigations should be extended beyond the focal firm and the evolving demography

of its team of top decision makers.3

Numerous studies have pointed to the inter-firm implications of demographic

heterogeneity (Boone et al., 2004; Keck and Tushman, 1993; Lawrence, 1997). Managers

with equivalent demographic characteristics typically exhibit ‘attentional homogeneity’ – i.e.,

3 Inter-firm mobility, of course, is not the only mechanism stimulating the emergence of shared perceptions of the competitive environment. As Reger and Huff (1993: 106) note, however, hiring from the same labor pool is an important avenue through which company managers interact with each other – besides participations to industry associations and other gatherings, access to similar sources of information such as trade publications, or employment of the same conultants.

11

similarity in the mental models regarding their sector and its incumbents (Abrahmson and

Hambrick, 1997). Convergent and analogous views of the competitive environment in turn

affect resource allocation and other strategic decisions (Sørensen, 1999b). Demographically

akin managers among peer firms express higher levels of mutual awareness, and tacitly or

explicitly mimic their strategic choices (see Pegels et al., 2000; Skaggs and Youndt, 2004).

Firms with demographically similar team members exhibit higher levels of strategic similarity

and hence competitive interdependence. That is because “a cohort is the societal exeriences

that have been imprinted on its members and have helped to shape their values and

perceptions” (Hambrick and Mason, 1984: 202), and cohorts of managers exposed to similar

historical environments share similar mental models (Kiesler and Sproull, 1982).

The argument regarding organizational demography and performance, therefore, takes

on a different twist here because also inter-firm demographic differences matter. Consider

for instance research on organizational ecology and niche overlap (McPherson, 1983;

Hannan and Freeman, 1989). A vast literature in organizational ecology demonstrated how

competitively interdependent firms act locally on some specific dimensions, with intensity

being greater if overlap occurs more locally. McPherson (1983) argued that the competition

between two voluntary organizations for new recruits is proportional to their socio-

demographic similarities. Baum and Singh (1994) showed that overlap between segments of

the child-care population produced higher failure levels.. Similar findings have been found

regarding technological (Podolny et al., 1996) and geographical (Baum and Mezias, 1992)

overlap. As firms with demographically similar members share similar strategic orientations

(for a review see Carpenter et al., 2004), we expect them to vie for the same resources and to

face stiffer competition. In contrast, dissimilarity produces divergent mental dispositions

among them and results in a strategic pursuit of diverse niches or segments (Pegels et al.,

2000). Thus, increases in demographic similarity intensify competition by generating

similarity in decision-making and in resource requirements (Sørensen, 1999b). Conversely,

12

demographic distance among firms creates more room for acting unilaterally. As before, a

strict test of the theory would suggests that increases in inter-firm demographic similarity

due to personnel inflows raises the risk of organizational failure, whereas the opposite holds

true for decreases in similarity:

H2a: When key member inflows raise demographic similarity among firms, the hazard of

organizational failure increases.

H2b: When key member inflows decrease demographic similarity among firms, the hazard of

organizational failure declines.

Both sets of hypotheses will thus be tested by linking diversity with survival. Diversity will

be treated here as an ecologically relevant concept by asserting that a firm’s leadership team

will be locked in strategically, for example by entrenchment into a collective mindset and

widely shared routines. While short term implications of diversity have dominated the

research on top management team research, we position this study very much in terms of

how diversity drives teams to be locked into a certain mold, depending on degree of

heterogeneity of its membership.

3. EMPIRICAL SETTING

Our empirical setting is the Dutch accounting sector during the period 1880-1986.

These firms are either single proprietorships or partnerships. Typically, partnerships

comprise partners or owner-managers and employees, often called associates. As owner-

managers, the partnership is more or less heterogeneous and shares many characteristics

with a top management team of small or medium sized corporations that have been the

focus of TMT (top management team) research tradition. . Several of our partnership

measures were constructed at the sub-population level of analysis– i.e., at the provincial level

13

and not at the national one. Previous research has shown how significant institutional

differences and the localized nature of competition in this industry make the province the

relevant level of analysis (for discussion, see e.g., Cattani, Pennings and Wezel, 2003). Each

province represents a distinct selection environment and becomes a key aspect for

measuring the localness of tenure. Within this sector, firms are embedded in geographic

entities with clear political, physical and administrative boundaries that contain institutional

and socio-economic properties, and are endowed with regional resources.

Provinces are also distinct for historical reasons, dating back to the 16th century. For

example some provinces participated in the Reformation and are largely protestant, while

others are more roman-catholic. Separated by rivers seas and other physical dividers, they are

linguistically set apart with one province, Frisia, even possessing its own official language,

Frysk. The provinces are not merely administrative units but are also historically, culturally,

institutionally and economically distinct. Although their significance is declining, they have

been critical in determining the current administrative structure of the Netherlands (Centraal

Bureau Statistiek, 2002). This suggests that geography matters for organizational activity such

as recruitment and service delivery, because firm embeddedness is largely local and, as an

institutional-competitive environment, provinces still hold sway (de Pree, 1997).

Figure 1 shows that the Dutch accounting industry remained fairly fragmented

during the window of our study. Due to their small size, many organizations compete at the

local (province) level and their critical resources (e.g., talented professionals and new clients)

are local as well. Some larger firms have over time expanded their geographic scope beyond

the provincial boundaries, yet the province is clearly the relevant environment for most of

these professional service firms. Given the central relevance of the local environment, we

chose local industry tenure as the dimension along which we measured demographic

heterogeneity and restricted the analysis to inter-firm mobility between firms located within

the same province – for a detailed explanation, see below. Our choice was premised on the

14

consideration that local experience, more than industry or firm experience, is a more

plausible proxy for managers’ mental models and strategic decision-making.4

---------------------------------

Insert Figure 1 about here

--------------------------------

Data

Data consist of information about individual professional accountants and their

organizations and were collected from the membership lists (or directories) of accountant

associations with one- to five-year gaps or intervals. Pennings, Lee and van Witteloostuijn

(1998) provide a comprehensive description of the sources. The percentages of our temporal

gaps are: 1year: 24%, 2years: 60%, 3years: 6%, 4years: 8%, 5years: 2%. This is clearly an issue

if we wish to map properly upticks and downticks of demographic diversity both within and

between firms. As we explain in the model section (see below), however, we controlled for

the different length in the intervals by creating a variable accounting for diverse time spans.

The membership lists provided information on the name, address, background education

and status (partner or associate) of each professional accountant within the association. We

reconstructed the histories of individual organizations by aggregating individual level data to

that of the firm. The data cover the entire population of Dutch accounting firms during the

period 1880-1986. In building the dataset, we considered the year in which the organization

appeared for the first time on the Register of Accountants as the founding year, whereas the

last year of appearance as the year of dissolution. We coded our dichotomous dependent

variable as 0 if the firm was still in existence in a given year and 1 once it exited the risk set.

4 Although seemingly arbitrary, the choice of using a measure of industry tenure is coherennt with a vast body of research in organizational demography (for an overview see Carpenter et al., 2004). Moreover, Finkelstein and Hambrick (1990: 492) have shown how this measure highly correlates with firm-level measures such as tenure in position and top-management team tenure. In the concluding section we further discuss this issue.

15

This notion is consistent with that proposed by Boone, Bröcheler and Carroll (2000) and by

Phillips (2002) who defined it as exit from the market, without distinguishing between

bankruptcy and merger or acquisition.5

The concern here is with the survival implications of hiring new members as partners

rather than associates. With their dual stratification, accounting firms have a dual stratification

with the top echelon endowed with superior human and social capital. The effects of

member entry on strategic conduct should be higher when the recruitment of new

organizational members involves partners. Partners are the owners of the firm. As such they

have a greater incentive to use their human capital for the growth of an organization than do

associates. Partners are more critical for a firm’s performance and survival as they have

greater influence on organizational outcomes (Galanter and Palay, 1991). Partners are

responsible for the overall management of accounting firms. Their decision-making power

extends to the task of building/changing routines, such as those dealing with hiring and

firing policies, procuring work and deploying junior professionals, differentiation (i.e., to

hedge against market shrinkage), investment, personal financial planning, and liability

insurance premium decisions (for more details see Maister 1993). The partners’ social capital

is thus more germane to the organization’s profit potential than that of the associates (see

Pennings et al. 1998). Thus, the recruitment of new partners often entails a shift in the

competitive landscape within which firms compete.

Independent Variables

5 Since “failure, in the sense of bankruptcy, cannot be observed in the audit industry and, therefore, cannot be distinguished from other types of exit” (Boone, Bröcheler and Carroll, 2000: 368), organizational dissolution encompasses different types of exit, ranging from the case where a single proprietorship vanishes as its owner is no longer listed in the C.P.A. (i.e., certified professional accountant) directories, to the case of dissolution by acquisition (but the professional accountants of the acquired organization kept working under the acquiring organization), to the case of dissolution by merger between two or more organizations – but see results section for an alternative specification. It should be noted that survival is a stochastic proxy that measures a firm’s risk of dissolution and is more encompassing than more immediate and concrete expressions of organizational effectiveness (e.g., Meyer and Gupta, 1994). The probability of dissolution entails an ongoing development of the firm’s health or viability, rather than the achievement of some other performance targets. In this respect, survival may be considered as context and time neutral, whereas targets as arbitrary and period idiosyncratic.

16

The inflow of key members (here, partners) is likely to affect the level of

demographic heterogeneity of the recipient firm. Due to the critical role of localized

networks and knowledge in our industry (see Smigel, 1969), we measured our construct

in terms of local experience. As we highlighted before, the province is the main

competitive arena for these firms. That is why we chose the length of service or tenure

within the focal ‘province’ as a proxy for managers’ demographic characteristics. It is

reasonable to believe that this measure captures the processes of interest here. Our

reasoning implies that two partners or managers with similar organizational and

industrial tenure will exhibit similar mental models and knowledge. Under these

circumstances, “[M]easures based on team tenure therefore potentially underestimate the

extent to which managers are similar” (Sørensen, 1999b: 727). In other words, whenever

partners join another firm the variable would take on the value 0, regardless of their level

of prior experience. This means treating all partners as if they were the same – thereby

ignoring their experience differences – every time they start working for a different firm.

Our choice of local tenure as the unique dimension along which we measure

demographic heterogeneity remains, however, grounded in theoretical reasoning. As

“managers operate on mental representations of the world and those representations are

likely to be of historical environments rather than current ones” (Kiesler and Sproull, 1982:

557), cohorts of managers exposed to similar historical environments share similar mental

models. Tenure implies socialization and mental programming, and the longer its duration,

the more paramount its mental and attitudinal effects. On the premise that ‘localness’ is a

valid attribution of strategic grouping, local tenure is specific and idiosyncratic to the

common fate of firms associated with its strategic groups. Local tenure, and by implication

indigenous or “native” experience, makes the firms’ members more knowledgeable about

current market conditions and future trends, and also instils competitive awareness among

them. When staffing involves such local occupational seniority and equivalence in experience

17

and socialization, the stronger the mutual awareness, the higher the intensity of rivalry

between firms. Moreover, an important strength of this measure is its focus on survival

consequences of demographic equivalence among peer firms. Long-tenured managerial

teams in fact have been shown to engage in persistent strategies and to select policies more

committed to the status quo (see Hambrick et al., 1993; Finkelstein and Hambrick, 1990).

Organizational demographers measure the level of tenure heterogeneity employing

the coefficient of variation – i.e., the ratio of the standard deviation of tenure to its mean. By

aggregating this measure, however, two main problems emerge. First, while this

standardization improves comparability across organizations, it comes “at a minimum price

of interpretive ambiguity” (Sørensen, 2002: 478). Models that use the coefficient of variation

alone risk confounding the differential effect that the mean and the standard deviation might

have on social processes. Second, when averages are computed using lower-level (e.g.,

managers) data, researchers implicitly adopt an “additive” model – i.e., the aggregate or

higher-level measure is just a summation of the lower-level units regardless of the variance

among these units (see Chan, 1998). By contrast, the standard deviation is a more

appropriate measure of heterogeneity because it overcomes the aggregation bias by

computing the distance between individuals, irrespective of the mean (Barsade et al., 2000).6

To test Hypotheses 1a and 1b, the variations in the value of the standard deviation

for local tenure was adopted. For each firm we first calculated the standard deviation of the

number of years the partners spent within the focal province. In particular, we captured the

evolving nature in local tenure heterogeneity by creating a time-varying variable – Tenure

Heterogeneity – which measures the change in the value of the standard deviation in tenure

that can be attributed to the entry of new partners only. We excluded endogenous diversity

6 In analyses not reported here we controlled for the stock of mean tenure due to new entries. It turned out that mean increases were positively related to organizational dissolution. This finding is in line with recent findings (e.g., Sørensen, 1999b) and may be interpreted as an indicator of the risks associated to strategic persistence (Finkelstein and Hambrick, 1990). The qualitative conclusions of the paper remained unaffected by the inclusion of this additional control.

18

shifts – for example changes due to partner death, retirement or simple exit. We calculated

the variable net of the natural increase (i.e., due to the mere passage of time) in tenure

between two consecutive time periods as well as any other variation due to the departure or

death of partners over the same period.

To test Hypotheses 2a and 2b, we created the variable Demographic Overlap that

measures the degree of overlap change in the average stock of provincial experience between

the recipient and its competitors located within the same geographical province. We chose

an average measure rather than a standard deviation based on existing literature. Both

quantitative (Finkelstein and Hambrick, 1990; Sørensen, 1999a, 1999b) and qualitative

research (Tripsas and Gavetti, 2000) seem to agree that average team tenure profoundly

affect strategies and performance. Consider for instance that Finkelstein and Hambrick

(1990: 488), studying the differences across firms in strategic behaviors, concluded that in

teams with longer tenure “perceptions become very restricted and risk taking is avoided. The

lowest risk thing is to follow the general tendency of mainstream competitors.”

The overlap measure varies over time as a result of recruiting new partners, again net

of the natural increase in tenure between time t and time t-1 as well as any other variation

due to the departure or death of partners over the same period. First, for each firm we

computed the maximum and minimum values for the provincial tenure range by adding and

subtracting 10% to (from) the mean value of firm tenure.7 We quantified the degree of niche

overlap by counting the number of firms falling within the same tenure range in a given year.

Since the inflow of key members can affect the degree of demographic similarity between

the recipient and its competitors by increasing or decreasing their mean tenure overlap, the

number of firms falling within the new range is likely to vary accordingly. To measure

changes in demographic overlap among firms, we calculated the difference between the

7 Other studies define an organizational niche using a two standard deviation window centered on the mean (e.g., Sørensen, 1999a, 1999b). To reduce the bias of the 10% band, we also applied a 20% and a 30% band and the results, though not reported here, were not significantly different from those presented in the paper.

19

numbers of firms in two consecutive time periods, respectively. To make sure this effect was

solely due to new entries, we restricted the variation of this variable to the cases in which (at

least) one new partner entry in a given year is observed.

A strict test of the theory sheds light on the counteracting forces behind within and

between firm differences in.. From earlier research it has been observed that both

heterogeneity and overlap ate harmful to the firm. With the variables, Tenure Heterogeneity

(TH) and Demographic Overlap (DO), we model their effect through as a “spline” function. In a

spline specification a variable is allowed to have slopes that differ depending on whether its

values lie above or below a given cut-off point (Greene, 2000: 322-324; see also Greve, 2003:

chapter 5). More specifically, we fitted a model that estimates the risk of dissolution of the

organization i at time t as follows:

[a]

where separate variables are entered to model the positive and negative effects. More

precisely, α1 is the coefficient that refers to tenure heterogeneity decreases (H1b); α2 is the

coefficient that refers to tenure heterogeneity increases (H1a); β1 is the coefficient that refers

to demographic overlap decreases (H2b); and β2 is the coefficient for demographic overlap

increases (H2a). Furthermore, δ is a vector of coefficients for the controls and D and E are

variables that take on the value 1 if the subscript condition is true, 0 otherwise. To facilitate

an intuitive interpretation of the opposite direction of the coefficients, diversity decrements

are measured in absolute values. Thus, while we expect α1 (H1b) and β1 (H2b) to be negative,

α2 (H1a) and β2 (H2a) should be positively related to organizational failure. Such a model

specification allows us to appreciate the different slopes exhibited by the decline and the

increase coefficients as conditional on the aforementioned change due to members entering

the partnership team. Their entry serves as a demography altering event having negative or

),)()()()(( 1012011012011 1111 −>−≤−>−≤− Χ++++=−−−− tDOtDOtTHtTHtit tttt

EDOEDODTHDTHFExit δββαα

20

positive long term performance consequences. We have no reason to expect the two effects

to be symmetrical.

Control Variables

Besides our variables of theoretical interest, we included in the final model several

control variables at the historical, provincial and organizational levels to rule out a number of

competing hypotheses, and to improve comparability with existing research on this industry

(Pennings, Lee and van Witteloostuijn, 1998; Cattani, Pennings and Wezel, 2002).

Historical Controls. The history of the Dutch accounting industry has been marked by

many important events that might well affect organizational survival chances in specific

years. In particular, we created two dummy variables for the governmental regulations

dealing with World War I (1914-1918) conditions and the occurrence of World War II (1941-

1946). Since the 1960s, the Dutch accounting industry has witnessed several fundamental

regulatory changes. More stringent requirements – e.g., the need for higher levels of

education and experience, and the examination to become C.P.A. – have over time restricted

the entry of potential competitors. In particular, four major regulatory changes have

encompassed both the supply and the demand of professional accounting services. In 1966,

with the Law on Registered Accountants, one professional organization or NIvRA

(Nederlands Instituut van Register Accountants) was created. Since then, every professional

accountant in public practice has become one of its members. We then created a variable –

Single Association – which takes on the value 1 if year > 1966, 0 otherwise. In 1970 the Act on

Annual Accounts of Companies (which took effect in 1971) expanded the number of firms

required by law of disclosing audited annual accounts by including large private firms and

cooperative societies in addition to public companies.

Finally, in 1983 the number was further enlarged with the Title 8 of Book 2 of the

Civil Code: every company, public or private, and every cooperative society was forced to

21

disclose audited annual accounts. After the promulgation of definitive guidelines in 1984, the

obligation remained less compulsory for small and medium-sized firms that were “allowed to

submit abridged annual accounts” (Boone, Bröcheler and Carroll, 2000: 366). We captured

the effect of the regulatory changes enforced in 1971 and 1984, which significantly

heightened the demand for audit services, with two dummy variables – i.e., Regulation of 1971

(1 if year > 1971) and Regulation of 1984 (1 if year > 1984), respectively.

Provincial Controls. We included the linear and quadratic effects of density

measures at the provincial level – i.e., Focal Province Density and Focal Province Density2 – to

estimate the extent to which more general ecological phenomena affect organizational

survival. In the presence of high levels of concentration, just a few organizations control

most of the available resources. We thus controlled for concentration of the industry – C4 –

given by the total market share of the top 4 firms.

The risk of organizational failure also depends on how many firms were founded or

disappeared each year – which reflects both the degree of munificence of the environment

and the extent to which ecological conditions affect inter-organizational mobility by

creating/destroying new job opportunities. Accordingly, we created two variables – Firms

Entering Province and Firms Exiting Province – to account for the number of firms founded and

dissolved during the previous year within a given province. We proxied the variations in

carrying capacity (e.g., number of potential clients) over time with the number of inhabitants

– Provincial Inhabitants – in each province for each year. We finally accounted for other

systematic geographical differences by introducing fixed effects at the province level.

Organizational Controls. Controlling for size and age,– Size – as the logarithm of

the annual number of professional accountants (partners and associates) while age was

proxied by number of years since inception. We also controlled for level of diversity: and

overlap at t-1 – i.e., Demographic Heterogeneity Stock and Demographic Overlap Stock, respectively.

Thus, our modeling design provides a stricter test of the ‘acceleration’ in the rate of

22

organizational failure due solely to further increases above the existing level of diversity.8 We

expect both these variables to be positively related to organizational failure. Additionally, we

controlled for the leverage of the firm aborrbing new partners – Recipient Leverage – i.e, the

number of associates per partner for each year. Firms with low leverage enjoy higher

survival chances. Many firms in this sector are single proprietorships and because of such

preponderance, we created a dummy variable – Single Proprietorship – taking on the value 1 if

size is equal to 1, otherwise 0. Although it may seem strange for a study on teams to include

individual firms in the sample, we opted for a control variable, instead of deleting

observations, to avoid adding any bias due to censoring. In the Appendix however we

present the estimates obtained by eliminating individual firms from the risk set. We aim at

capturing part of endogeneity, by controlling for a few other firm characteristics. For each

year, we also computed a ratio where the numerator is the sum of years of provincial

experience of all organizational members before the inflow of new members, and the

denominator is the mean value of provincial experience – i.e., Relative Position. The rationale

is that competitively stronger organizations are less exposed to the hazard of failure. Such an

advantage guarantees the selection of inflows from a larger pool of applicants. Second,

failing to control for team size is likely to bias the estimation of team heterogeneity effects

(Carpenter et al., 2004). We avoid this bias by adding a variable that counts the number of

partners that compose the team – team size. Third, we control for the degree of demographic

turbulence of the firm by counting the number of members (associates and partners)

entering and exiting the organization – Member Entries and Member Exits.

Finally, to ensure exogeneity with respect to the dependent variable, we lagged all our

covariates by one period. In Tables 1 and 2 we report the descriptive statistics and the

correlation values for the variables used in the analysis.

---------------------------------------- 8 That becomes clear when considering that, algebraically, our model resembles a restricted distributed lag regression without the lagged dependent variable as covariate.

23

Insert Tables 1 and 2 about here

----------------------------------------

4. MODEL AND METHOD

Following a standard procedure, the life of each organization was partitioned into

organization-years (Tuma and Hannan, 1984). The final dataset includes the life of 1920

firms over the 106-year study period for a total of 17,491 year-segments. Since our data were

collected at irregular intervals, the use of continuous event history analysis would bias our

estimates (for a discussion see Allison, 1982). Following previous studies using similar data

(e.g., Pennings, Lee and van Witteloostuijn, 1998), we employed discrete event history

analysis. A discrete time hazard rate is defined as:

itP = )t,T | (Pr itii XtT ≥=

[b]

where T is the discrete random variable measuring the uncensored date of failure and Pit is

the probability that an event will occur to firm i at time t given that the firm did not

experience such an event in any previous intervals. The complementary log-log specification

provides consistent estimates of the continuous time irrespective of the interval lengths

and/or the size of the survival rate (Allison, 1982). The model can be expressed as follows:

Pit = 1 – exp[- exp(γt + δ΄xit)] or [c]

or, after taking the logarithm of both sides:

log[- log (1 - Pit)] = γt + δ΄xit

[d]

where γt is an unspecified function of time, xit is the vector that includes both firm level

characteristics and environmental variables measured at different levels of analysis, and δ is

24

the vector of coefficients. The log transformation of the left-hand side is called the

complementary log-log function. In estimating the model, we specified:

γt = γ0 + γ1(organization age) + γ2(organization age2)

[e]

As mentioned before, one potential concern involves the endogeneity problem due

to self-selection among firms and their incoming partners. In other words, the entry of

partners observed may be systematically biased – i.e., not random. Literature in labor

economics indicates however that matching is difficult and a great deal of uncertainty

accompanies the hiring of a new member (see Jovanovic, 1979). Usually, high levels of short-

term turnover suggest matching uncertainty (e.g., Belzil, 2001). Therefore, we included the

number of years elapsed in between individual movements. The findings obtained (not

reported here) suggest that in this service sector , partners on average stayed put for eight

years. We interpret this high mobility frequency as further confirming the existence of

substantive uncertainty in the job searching process. That led us to consider self-selection on

recruitment not to be a serious problem in this study. We controlled for the different length

of the intervals by creating a variable accounting for them and using it into the ‘offset’

option. All the estimates were obtained with Maximum Likelihood Estimation method using

version 8 of STATA.

5. RESULTS

Table 3 presents the maximum likelihood estimates for the complementary log-log

models of organizational failure. Model 1 includes all the control variables. In Model 2, we

introduced the variable Tenure Heterogeneity Decrease and Tenure Heterogeneity Increase to test

Hypotheses 1a and 1b. Finally, in Model 3 we tested Hypotheses 2a and 2b by entering the

variable Demographic Overlap Decrease and Demographic Overlap Increase. As the impact of the

inflows may vary according to the level of diversity exhibited by the focal partnership, we

25

test these hypotheses while controlling for the stock of heterogeneity, i.e., both for the level

of heterogeneity within the firm and level of overlap between firms – i.e., Demographic

Heterogeneity Stock and Demographic Overlap Stock.

------------------------------------------

Insert Table 3 about here

------------------------------------------

The baseline model (Model 1) with all the control variables shows that the creation

of the single association (NIvRA) increased the dissolution rates of accounting firms,

whereas the 1971 and the 1984 regulations reduced it. All variables accounting for provincial

effects – i.e., births, deaths and densities – are in the expected direction and statistically

significant, supporting our choice of the unit of analysis.9

At the organizational level, the controls point to the existence of a curvilinear age

effect: accounting firms dissolve more likely when very young or old (the minimum risk of

dissolution is set at 28 years). A stock of human capital above the provincial average (Relative

Position) significantly reduces the risk of dissolution. As expected, the larger the number of

departures , the greater the hazard of failure, whereas the bigger the partnership, the greater

its survival chances. The coefficient related to the entry of new members does not reach

statistical significance. perhaps because influx of new members is complex and depends on

the new members and team fit Lastly, it is worth noticing that the single proprietorship

dummy is not statistically significant.suggesting that no systematic bias ensues in our results

due to the inclusion of single proprietorships in the risk set. The coefficient estimates of the

controls are by and large consistent with those of previous studies using these data

(Pennings, Lee and van Witteloostuijn, 1998; Cattani, Pennings and Wezel, 2002).

9 Analyses not reported here confirmed that the linear and quadratic effects of local density are prevalent on that of national density. This result is consistent with that obtained by Cattani, Pennings and Wezel (2003) for entry rates.

26

Tenure heterogeneity negatively affects organizational performance. - see the positive

value of the coefficient of the Demographic Heterogeneity Stock variable in Model 2. While

formulating Hypothesis 1, however, we argued that the impact of key member inflows on

the risk of organizational dissolution depends on the shift they produce in tenure

heterogeneity. We treat these changes in composition as a proxy of the costs of the

socialization process. Model 2 also presents the coefficient estimates for change in tenure

heterogeneity due to new inflows. The risk of early exit grows when partner recruitment

produces more heterogeneity. Unexpectedly, a drop in heterogeneity does not produce

higher mortality levels. The implication is that the impact of a decline in diversity (flows)

varies according to its level (stock). Given the increment in average tenure heterogeneity , at

the mean level of heterogeneity increase due to new entries, the risk of organizational

dissolution rises by about 15%. At a one-standard-deviation increase above the mean this

values becomes 35%. The inclusion of these variables improves significantly the fit of our

model, so providing strong support for Hypothesis 1a, but not to Hypothesis 1b.

As for the competitive implications of inter-firm mobility, we expected analogous

results. Shifts in demographic overlap due to a senior professional migrating across firms

produce dissolution in proportion to change in demographic similarity among them. Model

3 presents the estimate of the overlap shifts associated with new entries, while controlling for

the existing level of overlap. Adding these variables significantly improves our model’s

goodness of fit. The positive and significant estimate of the Demographic Overlap Stock variable

confirms its competitive. An increase in demographic overlap due to inter-fim mobility has

the hypothesized effect on organizational dissolution – as the coefficient of the Demographic

Overlap Increase variable indicates. Conversely, recruitment that result in diminished overlap

(Demographic Overlap Decrease) enhance organizational survival. The coefficient measuring the

impact of within-firm diversity loses statistical significance. This result supports the main

thrust of our paper:: the consequences of intra-firm diversity cannot be grasped unless we

27

consider the demographic make-up of competitors. Given the mean value of demographic

overlap increase in our sample, at the mean level of overlap increase due to new entries, the

risk of organizational dissolution is augmented by about 25% At one-standard-deviation

increase above the mean this risk becomes no less than 70%. Interestingly enough,

controlling for inter-firm overlap weakens the Demographic Heterogeneity Stock effect on survival.

The implication is that what matters most for the long term outlook is the change in

diversity, not the actual degree of heterogeneity in tenure among partners. We view these

findings as providing support to Hypothesis 2a and 2b, and, by and large, to our reasoning.

Robustness Tests.

We tested the robustness of the results to alternative model specifications. First,

although we did our best to control for potential endogeneity problems, this is still an issue

that may affect our findings. To control for further unobserved firm-level effects, we re-

estimated our models adopting a two-stage specification (Heckman, 1979). We first

estimated a probit model measuring the organizational probability of experiencing one new

entry (including all the controls presented in Model 1). Then, we used the estimates obtained

from this model as a control in the dissolution-rate specification. Because of our binary

dependent variable in the second step, we created a probit specification with self-selection,

following the approach of Van de Ven and Van Praag (1981). The results, which are

available from the authors upon request, confirm that endogeneity is not a serious concern.

Second, while the way we operationalize organizational dissolution does not coincide

with “failure,” it is nevertheless consistent with previous research (e.g., Boone, Bröcheler

and Carroll, 2000; Phillips, 2002) – which defines failure as exit from the market, without

distinguishing between bankruptcy and merger or acquisition. But mergers or acquisitions

are quite distinct from exit due to failure or extremely poor performance. For this reason,

since the M&A activity in the Dutch accounting industry was mainly, if not exclusively,

concentrated in more recent years, especially after 1966, we double-checked the robustness

28

of our results by running the analysis on a sub-sample that includes firms from 1880 to 1966.

Although the statistical significance of the coefficient estimates of the within-firm diversity

measures appears to be weaker, the results are consistent with those presented here.

Finally, although our models control for single proprietorships, our results might be

driven by the preponderance of these cases. Although the dummy variable flagging

individual firms never reaches statistical significant in the results presented in Table 3, we

further checked for this problem by fitting a model that excluded sole proprietorships from

the risk set. The Appendix reports the results for the models estimated, empirical findings

being qualitatively similar to those reported in Table 3. A significant discrepancy involves

local density-dependence: both its linear and the curvilinear effect disappear, suggesting that

small firms (those excluded from the risk set) are disproportionately exposed to such

negative selection.. Altogether these results led us to conclude that inclusion of single

proprietorships does not alter the main results.

6. DISCUSSION AND CONCLUSIONS

Drawing from organizational demography theories and the class of studies involving

upper echelons, in this paper we sought to create new grounds on diversity and firm long-

term performance. Considering that our sample coincides with the entire population of the

firms belonging to a professional services sector, and that we longitudinally traced the

movements of professionals between firms across geographical space, we are in the position

to observe the long-term performance consequences of variations in demographic diversity.

The inflow of members alters the demographic make-up of firms, both internally and

externally. Mobility changes internal diversity but alters the degree of similarity between

firms as well. We were able to distinguish between mobility events that augment or reduce

variations in heterogeneity, thereby implementing a more robust test of the theory.

29

Our findings suggest that uncovering such relationships is very much dependent on a

joint consideration of firm and inter-firm levels of analysis. By delving into the specific

conditions under which the entry of key individuals produces beneficial or detrimental

consequences, the paper has shown that personnel inflow should be framed as a multi-level

phenomenon. Instead of investigating each level in isolation, our study is among the firsts to

analyze their dynamic interactions in a longitudinal setting (for an exception see Sørensen,

1999b). While the degree of fit between new recruits and the recipient firm conditions the

performance consequences of hiring new members, those consequences are not confined to

the recipient firm. As demographic characteristics are also critical in influencing the strategic

positioning of firms (see Carpenter et al., 2004 for a review), the new members affect degree

of similarity between donor and recipient firms. Given this evidence we believe that the

present paper shows how personnel inflow constitutes an important force of strategy

convergence/divergence within industries. The performance implications of recruiting new

members, therefore, depend critically on changes in competitive overlap among firms.

The present study adds significantly to the body of literature on upper echelons.

First, by modeling both the increases and decreases in tenure heterogeneity, the

methodology adopted here permits us to decompose the positive and negative changes in

demographic diversity both within and between firms. The lack of significance of reductions

in tenure heterogeneity due to member inflow calls for further investigations of the

presumed positive effects of homogenous teams on organizational performance. Our

findings demonstrate that the existing stock of demographic heterogeneity of a team may

influence the positive impact of a diversity reduction on survival. We have followed the

putative assumption regarding demography and cognition, arguing that demographic

heterogeneity is predictive of mental and attitudinal diversity. Mental diversity shapes the

threshold for reaching consensus in beliefs and attitudes and, if that threshold rises, the risk

of disbanding grows, while lowering the threshold should have had opposite consequences.

30

We can speculate the existence of an interaction between the levels of diversity and its

reduction. More specifically, the diversity stock moderates the survival benefit due to

heterogeneity reduction: above certain levels of the stock, a reduction of heterogeneity does

not have beneficial effects because these are offset by the costs of absorbing inflows (see

Chatman, 1991). Second, we created significant additions to the recent cross-sectional

findings concerning inter-firm dynamics of tenure heterogeneity (Pegels et al., 2000). The

evidence presented here shows that changes in demographic overlap significantly affect

organizational performance. We consider especially intriguing the finding obtained by Model

3 in Table 3: after adding the overlap related variables, the coefficient measuring the impact

of the level of within-firm diversity loses its statistical significance. In this respect, our results

provide further support to the work of Sørensen (1999b) and suggest that the consequences

of intra-firm diversity cannot be fully grasped without considering the demographic

characteristics of competitors. The results of this study provide a first hint moreover at the

strategic convergence or divergence between firms due to executive migration and the

subsequent effects on firm survival.

A persistent concern in the literature on organizational demography is the ambiguity

regarding demographic characteristics matching cognitive ones (e.g., Finkelstein and

Hambrick, 1996). The research to date assumes that tenure diversity produces divergent

mental frames without substantiating this claim – it is just assumed that this is what it might

be. By no means is our study immune to such criticism. We likewise assumed that managers

with comparable demographic traits should exhibit similarity in their mental models and, by

implication, convergence in strategic decisions and behaviors. Unfortunately, our data do not

allow us to establish an empirical link between demographic characteristics and mental

proclivities.

On the other hand, we should emphasize some important advantages in the present

study of diversity and survival. Among all possible demographic attributes, tenure, and

31

especially local tenure is defensible as precursor to cognitive dispositions. By contrast,

attributes such as ethnicity, age and gender are much more tenuous when assuming a

demography-cognition connection, especially when such attributes are invoked to impute

mental dispositions around strategic and market conditions. In knowledge intensive

industries such as the accounting sector, with close ties between the professional and their

clients, local and lasting exposure to local conditions appears more germane to the

acquisition of common perceptions and attitudes. The implied socialization within

geographically defined markets is bound to instil commonality in beliefs and values

compared to settings where geography matters less and propinquity is not paramount.

The accounting sector is relatively static and persistent in its compliance with widely

accepted standards. Individual firms posses distinctive qualities and reveal varying degrees of

rents. (The Accounting Reporter, 2004). Demographically homogenous partnerships are

prone towards higher degrees of compliance and strongly ingrained norms and values but

when jolted by misfitting and divergent members joining, their firm becomes gravely

exposed to bankruptcy or industry exit. Particularly in this type of sector, the heterogeneity-

performance relationship comes sharply into focus when recruitment draws a firm into a

more disjointed and divergent membership and preservation of well ingrained routines and

practices is jeopardized.. Recall that the diversity–performance relationship takes on a

negative or positive relationship depending on the complexity of task, and the uncertainty

and ambiguity of markets. The accounting sector strives towards reliability and replicability,

both for individual firms and their industry. Innovation and adaptation occur slowly and are

subject to stringent regulation and standard setting. This industry would therefore be

expected to witness strong harmful effects of diversity and equally strong benefits of

homogeneity since these conditions impede or facilitate the compliance with accounting

rules and conventions, and even more so today as the corporate world is fraught with

numerous accounting scandals which necessitate further elimination of ambiguity in

32

corporate disclosure. We should press for additional research in high technology and other

emerging or novel industries where the findings on diversity might be contrary to what we

have discovered in the present setting.

In the end, we stress that any investigation of the performance implications of inter-

firm mobility should embrace the effect of member exit. Further research is then needed to

decompose the simultaneous positive and negative externalities of member entry and those

of member exit. The results on diversity and performance as decomposed here should

motivate researchers to link demographic with cognitive variables, both at the individual,

team or organizational level so that the researcher can peek more successfully into the black

box of decision making that is present among a set of decision makers. Similarly the

cognitive framing of markets and strategic groups and firm performance would benefit from

data that more tightly connect demographics with mental proclivities.

33

REFERENCES

Abrahamson E, Hambrick D. 1997. Attentional Homogeneity in Industries: the Effect of Discretion.

Journal of Organizational Behavior 18: 513-532.

Adner R, Helfat CE. 2003. Corporate effects and dynamic managerial capabilities Strategic Management

Journal 24(10): 1011-1025.

Allison PD. 1982. Discrete-time methods for the analysis of event histories. In S Leinhardt (ed.),

Sociological Methodology. San Francisco: Jossey-Bass, 61-98.

Ancona DG, Caldwell DF. 1992. Bridging the boundary: External activity and performance in

organizational teams. Administrative Science Quarterly 37: 634-665.

Bantel K, Jackson S. 1989. Top management and innovation in banking: does the composition of the

top team make a difference? Strategic Management Journal 10: 107-124.

Barsade SG, Ward AJ, Turner JDF, Sonnenfeld JA. 2000. To your hearts content: A model of

affective diversity in top management teams. Administrative Science Quarterly 45: 802-836.

Baum, J. A. C., S. Mezias. 1992. Localized competition and organizational failure in the Manhattan

hotel industry, 1898-1990. Administrative Science Quarterly, 37 580-604.

Baum JAC, Singh JV. 1994. Evolutionary Dynamics of Organizations. New York: Oxford University

Press.

Belzil, C. 2001. Unemployment insurance and subsequent job duration: Job matching versus

unobserved heterogeneity. Journal of Applied Econometrics 16(5): 619-636.

Boone C, Bröcheler V, Carroll GR. 2000. Custom service: Application and tests of resource-

partitioning theory among Dutch auditing firms from 1896 to 1992. Organization Studies 21(2):

355-381.

Boone C, de Brabander B, van Olffen W, van Witteloostuijn A. 2004. The Genesis of Top

management Team Diversity. Academy of Management Journal (forthcoming).

Bröcheler V, Maijoor S, van Witteloostuijn A. 2004. Auditor Human Capital and Audit Firm

Survival. The Dutch Audit Industry in 1930-1992. Accounting Organizations and Society

(forthcoming).

Boone, C., Wezel, F.C. and A. van Witteloostuijn 2006. Top Management Team Composition and

Organizational Ecology, in Baum, J.A.C. Dobrev, S. & a. van Witteloostuijn (Eds.), Ecology and

Strategy, Forthcoming, Advances in Strategic Management. Greenwich CT: Jai Press.

Carpenter MA, Fredrickson JW. 2001. Top management teams global strategies posture and the

moderating role of uncertainty. Academy of Management Journal 44 (3): 533-545.

Carpenter, M.A., W.G. Sanders, & M.A. Geletkanycz. 2004. The upper echelons revisited: The

antecedents, elements, and consequences of TEAM composition. Journal of Management. 30: 749-

778.

Carroll GR, Harrison JR. 1998. Organizational demography and culture: Insights from a formal

model and simulation. Administrative Science Quarterly 43(3): 637-667.

34

Cattani G, Pennings JM, Wezel FC. 2002. Organizational turnover as endogenous precursor of

industry dynamics and organizational survival. Working paper #02G39 University of Groningen,

The Netherlands.

Cattani G, Pennings JM, Wezel FC. 2003. Spatial and temporal heterogeneity in founding patterns.

Organization Science 14(6): 670-685.

Centraal Bureau Statistiek 2002. Ministerie van Binnelandse Zaken, Bevolking der Gemeenten van

Nederland (Population of Communities of the Netherlands).

Chatman JA. 1991. Matching people and organizations: Selection and socialization in public

accounting firms. Administrative Science Quarterly 36(3): 459-484.

Chan D. 1998. Functional relations among constructs in the same content domain at different levels of

analysis: a typology of composition models. Journal of Applied Psychology 83: 234-246.

Cyert R, March JG. 1963. A Behavioral Theory of the Firm. Englewood Cliffs, NJ: Prentice Hall.

De Pree JCI. 1997. Grenzen aan Verandering: Reorganisatie en structuurprincipes van het

binnenlands bestuur. (“Borders [or limits] to change: The relationship between reorganization

and structural attributes of interior governance”). The Hague: Wetenschappelijke Raad voor het

Regeringsbeleid.

Eisenhardt K, Schoonhoven K. 1990. Organizational growth: Linking founding team strategy

environment and growth among U.S. semiconductor ventures 1978-1988. Administrative Science

Quarterly 35: 504-529.

Finkelstein S. and D. Hambrick 1990. Top-management-team tenure and organizational outcomes:

The moderating role of managerial discretion. Administrative Science Quarterly 35: 484-503

Finkelstein S, Hambrick D. 1996. Strategic Leadership. St. Paul: West.

Galanter M, Palay T. 1991. Tournament of Lawyers. The Transformation of the Big Law Firm. Chicago: The

University of Chicago Press.

Greene WH. 2000. Econometric Analysis (4th edition). Upper Saddle River N.J.: Prentice Hall.

Greve HR. 2003. Organizational Learning from Performance Feedback: A Behavioral Perspective on Innovation

and Change. Cambridge MA: Cambridge University Press.

Groysberg, B. 2001. Can they take it with them? The portability of star knowledge workers

performance: Myth or reality? Unpublished doctoral dissertation, Harvard Business School.

Groysberg, B., A. Nanda. 2002. Does stardom affect job mobility? Evidence from analyst turnover in

investment banks. Boston, MA: Harvard Business School Division of Research Working paper

series 02-029.

Guzzo RA, Dickson MW. 1996. Teams in organizations: Recent research on performance and

effectiveness. Annual Review of Psychology 47: 307-338

Hambrick DC, Cho T, Chen M. 1996. The influence of top management team heterogeneity on

firms competitive moves. Administrative Science Quarterly, 41: 659-684.

35

Hambrick, D.C., M.A. Geletkancyz, and J.W. Fredrickson, 1993. Top Executive Commitment to the

Status Quo. Strategic Management Journal, 14: 401:418.Hambrick DC, D’Aveni RA. 1992. Top team

deterioration as part of the downward spiral of large corporate bankruptcies Management Science

38(10): 1445-1466.

Hambrick DC, Mason PA. 1984. Upper echelons: The organization as a reflection of its top

management. Academy of Management Journal 9(2): 193-206.

Hannan MT, Freeman J. 1989. Organizational Ecology. Cambridge, MA: Harvard University Press.

Heckman J. 1979. Sample selection bias as a specification error. Econometrica 47: 153-161.

Harris D, Helfat C. 1997. Specificity of CEO human capital and compensation. Strategic

Management Journal, 18: 895-920.

Hitt MA, Bierman L, Shimizu K, Kochhar R. 2001. Direct and moderating effects of human capital

on the strategy and performance in professional service firms: A resource-based perspective.

Academy of Management Journal 44: 13-28.

Jehn KA, Northcraft GB, Neale MA. 1999. Why differences make a difference: a field study of

diversity conflict and performance in workgroups. Administrative Science Quarterly 44: 741-763.

Jovanovic B. 1979. Job matching and the theory of turnover. Journal of Political Economy 87(5): 972-

990.

Keck SL, Tushman ML. 1993. Environmental and organizational context and executive team

structure. Academy of Management Journal 36: 1314-1344.

Keck SL. 1997. Top management team structure: Differential effects by environmental context.

Organization Science 8: 143-156.

Kiesler S. and L. Sproull. 1982. Managerial response to changing environments: Perspectives on

problem sensing from social cognition. Administrative Science Quarterly, 27: 548-570.

Kilduff M, Angelmar R, Merha A. 2000. Top management-team diversity and firm performance:

Examining the role of cognitions. Organization Science 11(1): 21-34.

Knight, D., Pearce, C. L., Smith, K. G., Olian, J. D., Sims, H. P., Smith, K. A., & Flood, P. (1999).

Top management team diversity, group process, and strategic consensus. Strategic Management

Journal, 20: 445-465.

Kristof-Brown, A, Jansen, KJ, Colbert, AE. 2002. A policy-capturing study of the simultaneous effect

of fit with jobs, groups, and organizations. Journal of Applied Psychology, 87 (5), 985-993.

Lawrence B. 1997. The black box of organizational demography. Organization Science 8: 1-22.

Maister D. H. 1993. Managing the Professional Service Firm. New York: Free Press.

McPherson, JM. 1983. An ecology of affiliation. American Sociological Review 48: 519-535.

Meyer MW, Gupta V. 1994. The performance paradox. In LL Cummings, BM Staw (eds.), Research in

Organizational Behavior 16. Greenwich CT: JAI Press, 309–369.

Michel J, Hambrick DC. 1992. Diversification posture and the characteristics of the top management

team. Academy of Management Journal 35: 9-37.

36

O’Reilly CA, Caldwell D, Barnett W. 1989. Work group demography social integration and turnover.

Administrative Science Quarterly 34: 21-37.

Pegels CC, Song YI, Yang B. 2000. Management heterogeneity competitive interaction groups and

firm performance. Strategic Management Journal 21(9): 911-923.

Pennings JM, Lee K, van Witteloostuijn A. 1998. Human capital social capital and firm survival.

Academy of Management Journal 41(4): 425-440.

Pfeffer J. 1983. Organizational demography. In LL Cummings, BM Staw (eds.), Research in

Organizational Behavior 5. Greenwich CT: JAI Press, 299–357.

Pfeffer J. 1994. Competitive Advantage Through People. Boston MA: Harvard Business School Press.

Phillips, DJ. 2002. A Genealogical Approach to Organizational Life Chances: The Parent-Progeny

Transfer and Silicon Valley Law Firms, 1946-1996. Administrative Science Quarterly, 47: 474-5.

Podolny, J.M., T. Stuart & M.T. Hannan. 1996. Networks, knowledge, and niches: competition in the

worldwide semiconductor industry, 1984-1991. American Journal of Sociology 102: 659-689.

Porac JF, Thomas H, Baden-Fuller C. 1989. Rivalry and the industry model of the Scottish knitwear

producers. Journal of Management Studies 26(4): 397-416.

Rao H, Drazin R. 2002. Overcoming resource constraints on product innovation by recruiting talent

from rivals: A study of the mutual fund industry 1986-1994. Academy of Management Journal 45(3):

491-507.

Reger RK, Huff AS. 1993. Strategic groups: A cognitive perspective. Strategic Management Journal 14(2):

103-123.

Rotolo, T. and J. M. McPherson, 2001. The System of Occupations: Modeling the Competition of

Occupations in Sociodemographic Dimensions. Social Forces 79: 1095-1130.

Sanders WMG, Carpenter MA. 1998. Internationalization and firm governance: The roles of CEO

compensation top tram composition and board structure. Academy of Management Journal 41(2):

158- 178.

Schneider, B. 1987. The people make the place. Personnel Psychology, 40: 437-453.

Skaggs BC, Youndt M. 2004. Strategic positioning human capital and performance in service

organizations: A customer interaction approach. Strategic Management Journal 25(1): 85-88.

Smigel EO. 1969. The Wall Street Lawyer: Professional Organization Man? Bloomington: Indiana

University Press.

Smith KG, Smith KA, Ollian JD, Sims HP, O’Bannon DP, Scully JA. 1994. Top management team

demography and process: The role of social integration and communication. Administrative Science

Quarterly 39: 412-438.

Sørensen JB. 1999a. Executive migration and interorganizational competition. Social Science Research

28: 289-315.

Sørensen JB. 1999b. The ecology of organizational demography: Managerial tenure distributions and

organizational competition. Industrial and Corporate Change 8: 713-744.

37

Sørensen JB. 2002. The Use and Misuse of the Coefficient of Variation in Organizational

Demography Research Sociological Methods and Research 30: 475-491.

Stewman, S. 1988. Organizational demography. Annual Review of Sociology, 14: 173-202.

Thomas, H., Clark, S. M., & Gioia, D. A. 1993. Strategic sense making and organizational

performance: Linkages among scanning, interpretation, action, and outcomes. Academy

of Management Journal, 36: 239-270.

Tripsas M, Gavetti G. 2000. Capabilities cognition and inertia: evidence from digital imaging. Strategic

Management Journal 21(10-11): 1147-1161.

Tuma N, Hannan MT. 1984. Social Dynamics: Models and Methods. Orlando: Academic Press.

Tushman M, Smith W. 2002. Organizational Technology. In JAC Baum (ed.), Companion to

Organizations. Oxford: Blackwell, 386-314.

Van de Ven W, Van Praag BMS. 1981. The demand for deductibles in private health Insurance: A

probit model with sample selection. The Journal of Econometrics 17: 229-25.

Wernerfelt B. 1984. A resource-based view of the firm. Strategic Management Journal, 5: 272-280.

Wiersema MA, Bird A. 1993. Organizational demography in Japanese firms: Group heterogeneity

individual dissimilarity and top management team turnover. Academy of Management Journal 36:

996-1025.

Williams KY, O’Reilly III CA. 1998. Demography and diversity in organizations: A review of 40

years of research. In LL Cummings, BM Staw (eds.), Research in Organizational Behavior 20.

Greenwich CT: JAI Press, 77-140.

Zenger TR, Lawrence BS. 1989. Organizational demography: The differential effects of age and

tenure distributions on technical communication. Academy of Management Journal 32(2): 353-376.

38

Figure 1

Concentration Index of the Four Biggest Firms

0

0.2

0.4

0.6

0.8

1

1.2

1880

1884

1888

1892

1896

1899

1903

1908

1913

1916

1920

1923

1928

1932

1937

1941

1947

1950

1953

1956

1960

1964

1968

1971

1973

1978

1986

39

Table 1

Descriptive Statistics (17491 Observations)

Variables

Mean

Std. Dev.

Min

Max

Age World War I World War II Single Association Regulation 1971 Regulation 1984 Provincial Inhabitants C4 Firms Entering Province Firms Exiting Province Focal Province Density Focal Province Density2 Recipient Leverage TEAM size Member Entries Member Exits Size (log) Single Proprietorship Relative Position Demographic Heterogeneity Stock Tenure Heterogeneity Decrease (absolute value)Tenure Heterogeneity Increase Demographic Overlap Stock Demographic Overlap Decrease (absolute value)Demographic Overlap Increase

12.772 0.034 0.077 0.239 0.176 0.031

1723102 0.231 13.531 12.894 67.420 5978.7 0.148 2.41

0.2168 0.1384 0.384 0.691 0.908 1.757 0.0531 3.607 7.749 1.234

17.0001

12.223 0.180 0.266 0.427 0.381 0.172

659750 0.116 11.442 12.430 37.860 4778.3 0.476 7.85 2.47 1.528 0.768 0.462 0.841 8.829 0.5526 4.248 4.234 5.717 23.95

0.5 0 0 0 0 0

221603 0.109557

0 0 0 0 0 0 0 0 0 0 0 0

21.213 0.0018

0 0 1

74 1 1 1 1 1

31214711 65 57 126

15876 8

159 147 100

5.9053621

6.629145149.481

0 28.9913

161 125 158

40

Table 2 Pairwise Correlation Values

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

1 Failure 1.00

2 Age -0.08 1.00

3 World War I -0.03 -0.05 1.00

4 World War II -0.03 0.22 -0.05 1.00

5 Single Association 0.10 -0.21 -0.10 -0.16 1.00

6 Regulation 1971 0.08 -0.16 -0.09 -0.13 0.82 1.00

7 Regulation 1984 0.09 -0.07 -0.03 -0.05 0.32 0.38 1.00

8 Provincial Inhabitants 0.04 0.05 -0.13 -0.01 0.21 0.16 0.05 1.00

9 C4 0.10 -0.21 -0.08 -0.18 0.86 0.85 0.30 0.15 1.00

10 Firms Entering Province 0.03 0.08 -0.06 0.09 0.14 0.08 0.05 0.39 0.04 1.00

11 Firms Exiting Province 0.17 0.04 -0.13 -0.05 0.14 0.16 0.14 0.50 0.14 0.50 1.00

12 Focal Province Density -0.03 0.20 -0.09 0.08 -0.37 -0.34 -0.14 0.60 -0.42 0.28 0.40 1.00

13 Focal Province Density2 -0.03 0.21 -0.13 0.07 -0.41 -0.38 -0.15 0.50 -0.44 0.18 0.36 0.98 1.00

14 Single Proprietorship 0.00 -0.10 0.02 -0.01 -0.04 -0.04 -0.02 -0.09 -0.04 -0.05 -0.07 -0.05 -0.05 1.00

15 Size (log) 0.00 0.13 -0.04 0.00 0.09 0.09 0.04 0.12 0.10 0.09 0.11 0.06 0.05 -0.78 1.00

16 Relative Position -0.09 0.36 0.00 0.03 -0.04 -0.02 -0.03 0.00 -0.05 0.00 -0.01 0.03 0.03 -0.21 0.26 1.00

17 Recipient Leverage 0.00 0.06 -0.05 0.02 0.03 0.05 0.04 0.10 0.04 0.03 0.06 0.05 0.05 -0.46 0.55 0.11 1.00

18 TEAM size 0.01 0.02 -0.02 -0.02 0.15 0.15 0.05 0.09 0.14 0.09 0.11 0.01 -0.01 -0.27 0.66 0.15 0.20 1.00

19 Member Exits 0.04 0.06 -0.01 -0.02 0.04 0.05 0.03 0.03 0.04 0.01 0.03 0.02 0.02 -0.08 0.21 0.07 0.09 0.31 1.00

20 Member Entries 0.03 0.06 -0.01 -0.02 0.05 0.04 0.05 0.04 0.05 0.01 0.04 0.02 0.02 -0.11 0.28 0.10 0.12 0.41 0.50 1.00

21 Demographic Heterogeneity Stock

0.05 0.05 -0.02 -0.01 0.07 0.05 0.05 0.08 0.07 0.02 0.08 0.04 0.04 -0.28 0.39 0.17 0.16 0.36 0.35 0.47 1.00

22 Tenure Heterogeneity Decrease -0.01 -0.01 0.01 0.00 0.02 0.02 0.00 0.01 0.02 0.00 0.01 -0.01 -0.01 0.12 -0.07 -0.03 -0.03 -0.01 -0.02 -0.03 -0.05 1.00

23 Tenure Heterogeneity Increase 0.03 0.03 -0.01 0.03 -0.01 -0.02 0.00 0.03 -0.02 0.02 0.03 0.05 0.05 -0.17 0.11 0.07 0.04 0.02 0.02 0.04 0.25 -0.01 1.00

24 Demographic Overlap Stock 0.08 -0.13 -0.07 -0.09 0.09 0.09 0.07 0.16 0.11 0.03 0.12 0.11 0.13 0.10 -0.02 -0.07 0.01 0.03 0.01 0.01 0.00 0.04 -0.04 1.00

25 Demographic Overlap Decrease -0.06 -0.01 0.02 -0.02 -0.02 0.01 -0.01 -0.01 0.00 0.02 -0.01 -0.02 -0.03 0.09 -0.10 -0.03 -0.03 -0.05 -0.06 -0.09 -0.16 0.12 -0.14 0.04 1

26 Demographic Overlap Increase 0.08 -0.05 -0.02 -0.05 0.01 -0.03 0.04 0.01 -0.02 -0.06 -0.02 0.01 0.02 0.01 -0.02 -0.05 0.00 -0.02 0.03 0.02 0.03 -0.11 0.04 0.02 -0.06 1

41

Table 3 - Complementary Log-Log Models for the Failure Rate of Dutch Accounting Firms 1880-1986 (1661 events)

Model 1 Model 2 Model 3 Coefficient Std Error Coefficient Std Error Coefficient Std Error

Constant -2.982** .249 -3.010** .250 -3.328** .252 Age -.087** .008 -.086** .008 -.079** .008 Age2 .002** .0002 .002** .0002 .001** .0002 WWI .550** .198 .555** .198 .570** .198 WWII -1.269** .148 -1.287** .148 -1.286** .150 Single Association 1.101** .164 1.089** .164 1.115** .167 Regulation 1971 -1.325** .165 -1.321** .165 -1.272** .167 Regulation 1984 -1.139** .137 -1.146** .138 -1.194** .138 Provincial Inhabitants -.0000001 .0000001 -.0000001 .0000001 -.0000001 .0000001 C4 -.381 .801 -.361 .802 -.595 .812 Firms Entering Province -.036** .004 -.035** .004 -.035** .004 Firms Exiting Province .067** .003 .067** .003 .068** .003 Focal Province Density -.015** .007 -.016** .007 -.015** .007 Focal Province Density2

.000130**

.000046

.000130**

.000046

.000115**

.000046

Single Proprietorship -.059 .104 -.028 .106 -.090 .106 Size (log) .068 .086 .064 .086 .032 .085 Relative Position -.233** .043 -.248** .043 -.259** .044 Leverage -.071 .068 -.060 .069 -.063 .070 TEAM size -.011* .007 -.012* .007 -.010 .007 Member Exits .050** .022 .050** .022 .047** .019 Member Entries .005 .016 .003 .016 .003 .014 Demographic Heterogeneity Stock .002** .001 .001 .001 Tenure Heterogeneity Decrease (absolute value)

-.024

.040

-.003

.043

Tenure Heterogeneity Increase .037** .017 .032* .018 Demographic Overlap Stock .045** .006 Demographic Overlap Decrease (absolute value)

-.013**

.003

Demographic Overlap Increase .005** .001 Provincial fixed effects Included Included Included Offset (for different time spans) Included Included Included Log-likelihood -4643.39 -4639.64 -4606.71 Observations 17491 17491 17491

** p< 0.05; * p < 0.10. Two-tailed tests for all variables

42

Appendix - Complementary Log-Log Models for the Failure Rate of Dutch Accounting Firms 1880-1986 (excluding individual firms, 518 events)

Model 1 Model 2 Model 3 Coefficient Std Error Coefficient Std Error Coefficient Std Error

Constant -4.487** .459 -4.564** .465 -4.866** .464 Age -.077** .012 -.073** .012 -.061** .012 Age2 .001** .0002 .001** .0002 .001** .0002 WWI .264 .606 .302 .608 .270 .606 WWII -1.297** .307 -1.353** .307 -1.356** .310 Single Association .674** .271 .617** .271 .774** .276 Regulation 1971 -2.067** .262 -2.064** .263 -2.006** .264 Regulation 1984 -.822** .205 -.827** .205 -.908** .203 Provincial Inhabitants .0000009 .0000002 .0000010 .0000003 .0000008 .0000003 C4 3.682** 1.092 3.817** 1.092 3.228** 1.112 Firms Entering Province -.031** .008 -.030** .008 -.031** .009 Firms Exiting Province .051** .005 .051** .005 .053** .005 Focal Province Density .006 .015 .005 .015 .005 .016 Focal Province Density2

.000039

.000096

.000039

.000096

.000019

.000099

Size (log) .030 .083 .031 .083 -.002 .084 Relative Position -.225** .064 -.256** .065 -.249** .065 Leverage -.110* .069 -.099 .069 -.098 .069 TEAM size -.009 .007 -.009 .007 -.008 .007 Member Exits .029** .014 .028** .013 .026** .013 Member Entries .014 .010 .012 .010 .013 .009 Demographic Heterogeneity Stock .002** .001 .0004 .002 Tenure Heterogeneity Decrease (absolute value)

-.061

.042

-.038

.044

Tenure Heterogeneity Increase .048** .019 .043** .020 Demographic Overlap Stock .052** .011 Demographic Overlap Decrease (absolute value)

-.011**

.004

Demographic Overlap Increase .005* .003 Provincial fixed effects Included Included Included Offset (for different time spans) Included Included Included Log-likelihood -1374.62 -1370.23 -1357.47 Observations 5404 5404 5404

** p< 0.05; * p < 0.10. Two-tailed tests for all variables