Social Capital, Uncertainty, Restructuring 1 Social Capital Activation, Uncertainty, and Organizational Restructuring Sameer B. Srivastava Joint Program in Organizational Behavior & Sociology Harvard University Word Count: 13,479 Running Head: Social Capital, Uncertainty, Restructuring November, 2011 Keywords: social capital; uncertainty; network activation; organizational structure; collective attachments; social exchange; organizational change
51
Embed
Social Capital Activation, Uncertainty, and Organizational ...faculty.chicagobooth.edu/workshops/orgs-markets/past/pdf/srivasta... · Social Capital, Uncertainty, Restructuring 2
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Social Capital, Uncertainty, Restructuring
1
Social Capital Activation, Uncertainty, and Organizational Restructuring
Sameer B. Srivastava
Joint Program in Organizational Behavior & Sociology
Harvard University
Word Count: 13,479
Running Head: Social Capital, Uncertainty, Restructuring
November, 2011
Keywords: social capital; uncertainty; network activation; organizational structure; collective
attachments; social exchange; organizational change
Social Capital, Uncertainty, Restructuring
2
Social Capital Activation, Uncertainty, and Organizational Restructuring
Abstract: How do events transform social structure? This article illuminates the
microsociological mechanisms through which the uncertainty of organizational
restructuring shifts intraorganizational network structure. It helps resolve an
important conceptual puzzle about uncertainty and networks: one perspective
suggests that uncertainty leads actors to decrease network range, while another
implies the opposite effect. The article clarifies how these opposing forces play
out during the uncertainty of restructuring. The author develops theoretical
propositions about how the uncertainty of restructuring alters network range
across formal subunits and cross-cutting work groups. These propositions are
tested using a unique data set that includes the period before, during, and after a
major restructuring in an information services firm. Analyses of 40 weeks of
electronic communications among 114 employees reveal that, during periods of
heightened uncertainty, there was: (1) an increase in network range across formal
subunits; and (2) a decrease in network range across the myriad work groups to
which actors belonged. The study contributes to research on social capital,
collective attachments in social exchange, and the dynamics of organizational
structure during times of change.
November, 2011
Social Capital, Uncertainty, Restructuring
3
Sociological research has long studied the role of events in transforming social structure
(Sewell 2005). Transformative events, such as an organizational restructuring, tend to breed
uncertainty for individual actors – for example, about their structural position or resources. This
uncertainty can, in turn, prompt actors to mobilize social capital – i.e., to seek resources such as
information, influence, and social support that are accessible through social connections (Lin
2001; McDonald and Westphal 2003; Mizruchi and Stearns 2001). As Pescosolido (1992: 1105)
writes, “Events set into motion a specific process of coping with uncertainty….They can be seen
as a ‘shock’ to a network, reverberating through it and altering the overall system of relations.”
Just because valuable resources are available through social relations does not, however,
mean that they will be tapped. Trust-based barriers (Smith 2005), cognitive recall of
relationships (Smith, Menon, and Thompson in press), and interpersonal affect (Casciaro and
Lobo 2008) all can constrain the set of contacts with whom people exchange resources. That is,
during many events, people activate only a subset of the relations to which they have access.
Existing theory poses an important puzzle about uncertainty’s effects on the range of
activated networks – i.e., the diversity of individuals to whom a person initiates contact (Burt
1983; Reagans and McEvily 2003). One theoretical perspective suggests that uncertainty will
lead people to activate ties to socially proximate contacts (e.g., Hurlbert, Haines, and Beggs
2000; McDonald and Westphal 2003), with whom they are more likely to have trust-based
relationships based on a history of past exchange (Cook and Emerson 1978). Another view
suggests that uncertainty will trigger the activation of ties to socially distant contacts, as people
search for novel information (e.g., Burt 2000; Friedkin 1982) and seek to influence or form
coalitions with those on whom they depend (Pfeffer 1989, 1992; Stevenson and Greenburg
Social Capital, Uncertainty, Restructuring
4
2000).1 The former perspective predicts a decrease in the range of activated networks, while the
latter suggests an increase. These choices are important to understand because
intraorganizational network activation can have consequences for individual attainment (Burt
1992; Podolny and Baron 1997; Seibert, Kramer, and Liden 2001) – especially when, as in a
restructuring event, power and resources are in flux (Pfeffer 1989). It can also influence the
level and quality of social support that people obtain to withstand the stresses of uncertain times
(Cohen and Wills 1985; House, Umberson, and Landis 1988; Swanson and Power 2001).
This article helps resolve the conceptual puzzle about uncertainty’s effects on the range
of activated networks. I draw on theories of social capital (Lin 2001), the dynamics of
restructuring (Gulati and Puranam 2009; Huy 2002; Nickerson and Zenger 2002), and
individual-to-collective attachments (Lawler, Thye, and Yoon 2009) to derive propositions about
the effects of uncertainty on two dimensions of intraorganizational network range – across
formal subunits (e.g., departments) and cross-cutting work groups (e.g., project teams consisting
of members from different departments). Next I report on a study that takes advantage of the
quasi-exogenous shock of restructuring to identify the effects of uncertainty on network
activation. The analysis draws on a longitudinal data set – spanning 40 weeks and including the
electronic communication logs of 114 employees, company-wide email distribution lists,
archived employee communications memos, and human resource records – that provides a rare
look into an organization before, during, and after a spell of uncertainty. I also report on semi-
structured interviews with a subset of these employees. Findings from this investigation
contribute to research on social capital, collective attachments in social exchange, and the
dynamics of organizational structure during times of change.
1Following Putnam (2000: 22-23), the former perspective can be thought of as emphasizing “bonding” social capital
– i.e., looking inward and mobilizing solidarity among homogeneous groups – and the latter as focusing on
“bridging” social capital – i.e., looking outward and encompassing people across diverse social cleavages.
Social Capital, Uncertainty, Restructuring
5
THEORY
Restructuring and Uncertainty
Restructuring is defined as the addition, deletion, or recombination of formal subunits (Samina
2006); these changes are often accompanied by downsizing and changes in reporting
relationships. Even when restructuring is well anticipated, it can produce uncertainty for
organizational actors because one set of changes can trigger a cascade of other realignments that
are difficult to predict (Hannan, Polos, and Carroll 2003a, 2003b). In many cases, even the
initial changes of restructuring are poorly anticipated. That is, employees in a restructuring
organization are often left wondering whether they will remain employed, to whom they will
report, and how their job roles might change (for a review of the effects of restructuring on
workers, see Kalleberg [2009]).
Social Capital Activation
The uncertainty of restructuring can lead people to seek resources from their social contacts. At
any given time, many of these ties will be latent – that is, people will have pre-existing
relationships but no current interaction with a set of individuals. In the wake of events, such as
restructuring, individuals convert some latent ties into active ones – that is, they initiate contact
with individuals with whom they have a pre-existing relationship. For example, organizational
actors might seek out supervisors, mentors, and colleagues in other units who can provide
valuable information or influence their outcomes. Following Hurlbert, Haines, and Beggs (2000:
599), who examine resource mobilization through social contacts following a natural disaster, I
define social capital activation as the choice to initiate contact with certain individuals among the
set of actors in one’s pre-existing network (see also Renzulli and Aldrich [2005], Smith [2005]).2
2Social capital activation has also been used to refer to the cognitive recall of contacts in response to a situational
prime (Smith, Menon, and Thompson forthcoming). Because the distinction between recall and the choice to initiate
Social Capital, Uncertainty, Restructuring
6
Social Exchange under Uncertainty
Intraorganizational networks can be conceptualized as exchange relations (Cook and Whitmeyer
1992; Uehara 1990) – i.e., repeated transactions of valued resources, such as information and
influence, among the same actors over time (Emerson 1981). Prior research yields contradictory
expectations about the consequences of uncertainty for the choice of exchange partner.3 On one
hand, uncertainty promotes relational commitment: a tendency toward continued exchange with
longstanding, trusted partners (Cook and Emerson 1978; Kollock 1994; Molm, Peterson, and
Takahashi 2000; Podolny 1994). This inclination can lead to a preference for socially proximate
partners, with whom a person is more likely to have a trust-based relationship based on a history
of prior exchange (Buchan, Croson, and Dawes 2002; Krackhardt 1992; Macy and Skvoretz
1998). As Macy and Skovretz (1998: 651) conclude from a series of simulation experiments,
“The earliest trust rule is based on social distance—trust neighbors but not outsiders.”
Uncertainty, for example, can prompt CEOs to seek advice from contacts with the same
functional background and in the same industry rather than contacts with different functional
backgrounds or in different industries (McDonald and Westphal 2003). The former are more
socially proximate to the CEO than the latter. Similarly, bankers operating in uncertain settings
seek information from close contacts when seeking advice on and support for deals (Mizruchi
and Stearns 2001). Finally, when uncertainty is associated with crisis (Hermann 1963), it can
contact with a recalled contact is not of conceptual relevance to my line of reasoning, I focus on the latter in my
usage of the term. My definition also differs from that used by Smith [2005], who defines activation to include both
an individual’s choice to seek resources from a contact and the contact’s choice to provide instrumental or
expressive aid to the help seeker. Because my arguments pertain only to the choices of the focal actor, I define
activation to include only the first of these steps; i.e., the choice to initiate contact. 3Restructuring creates uncertainty for organizational actors, for example by raising questions about their
employment status or departmental affiliation. These unknowns in turn breed social uncertainty, or the likelihood of
concluding satisfactory exchange with colleagues (Cook and Emerson 1984). Social uncertainty stems from the fact
that people become less sure of the resources, power, and positions they will hold once the restructuring has
concluded. For example, informal negotiations among department heads about resource sharing, personnel
transfers, or collaborative projects can become difficult if the departments are at risk of being dissolved or
undergoing budget cuts. Because restructuring-related uncertainty and social uncertainty tend to co-occur, I use the
term, uncertainty, for both forms unless the distinction is materially relevant.
Social Capital, Uncertainty, Restructuring
7
also lead to increased commitment to one’s formal subunit and reduced cooperation with other
formal subunits (Krackhardt and Stern 1988). Thus, when facing the uncertainty of
restructuring, people may seek contact with organizationally proximate colleagues whom they
are likely to know well and can trust for information, advice, and support.
On the other hand, under conditions of uncertainty, the resources held by organizationally
distant colleagues can become more valuable (Burt 2000; Pfeffer 1989; Pfeffer 1992). For
example, colleagues in other parts of the organization may possess crucial non-redundant
information (Friedkin 1982), such as knowledge of job vacancies in other subunits that may be
created by restructuring. Similarly, past supervisors, potential future supervisors, and mentors in
other parts of the organization may wield important influence over decisions about whether an
actor will remain employed by the organization and, if so, about the job role the person will
assume. Thus, people will be motivated to strengthen political coalitions with actors in other
parts of the organization (Pfeffer 1989; Pfeffer 1992). These factors can promote a preference
for exchange with organizationally distant colleagues. In sum, the former set of arguments
suggests that restructuring-induced uncertainty will lead to a decrease in network range (i.e., the
activation of ties to organizationally proximate colleagues), while the latter implies an increase in
network range (i.e., the activation of ties to organizationally distant colleagues).
Dimensions of Intraorganizational Network Range
To resolve this conceptual puzzle, I suggest the need to distinguish uncertainty’s effects on two
facets of intraorganizational network range. Range can be defined along multiple dimensions,
for example size, complexity, density, and diversity (Campbell, Marsden, and Hurlbert 1986). In
organizational settings, a key dimension of range is diversity – based on affiliations to collective
units, such as departments or project teams (Ancona and Caldwell 1992; Krackhardt and Stern
Social Capital, Uncertainty, Restructuring
8
1988; Reagans and McEvily 2003). For example, a person with a high proportion of ties to
colleagues outside her department has broader network range than a person with ties primarily
within her department. I follow Ibarra (1992: 166) in distinguishing between two kinds of
collective units inside organizations: (1) formal subunits, which are defined by “specified
relationships between superiors and subordinates and among functionally differentiated groups
that must interact to accomplish an organizationally defined task;” and (2) cross-cutting work
groups, such as “committees, task forces, teams, and dotted-line relationships that are formally
sanctioned by the firm” but do not correspond to the organizational chart.4 In most
organizations, people belong to a handful of formal subunits, based on the hierarchical reporting
relationships in which they are embedded, and to many different work groups, based on the
workflows and decision processes in which they participate. An increase in range across formal
subunits can occur through an increase in contact with colleagues in different subunits, a decline
in contact with people in the same subunit, or a combination of both effects. The same is true for
range across work groups.
Network Range across Formal Subunits
For three reasons, I expect that restructuring-induced uncertainty will lead to an increase in
network range across formal subunits. First, colleagues in other departments are more likely to
possess non-redundant information about the restructuring (Friedkin 1982), such as which
individuals and groups are likely to be affected by the organizational change and what job
vacancies are likely to be created. Second, uncertainty can trigger the exercise of power and
influence tactics. Organizational actors are apt to use these tactics during situations “like
4Soda and Zaheer (forthcoming) follow a similar approach, distinguishing between “authority relationships” and
“workflow relationships.” Of course, a formal subunit can also be thought of as a work group, and work groups can
be nested within formal subunits. I focus on those work groups that do not correspond to formal subunits – for
example, project teams consisting of members from different departments. For brevity, I hereafter refer to them
simply as work groups.
Social Capital, Uncertainty, Restructuring
9
reorganizations and budget allocations…and in instances where there is likely to be uncertainty
and disagreement” (Pfeffer, 1992: 37). If the uncertainty of restructuring means that current
reporting relationships and departmental affiliations may not persist, people will instead direct
attention toward colleagues in other organizational subunits who can advocate on their behalf –
for example, keep their names off employee layoff lists and lobby decision makers to help them
secure coveted positions. Finally, in many organizations, periods of restructuring are governed
by strong communication norms. For example, managers are instructed to communicate only
officially sanctioned messages, hew to pre-specified communication timetables, and refrain from
‘leaking’ information to subordinates (for an illustration of prescribed communication protocols
for managers leading organizational change, see Klein [1996]). As a result, the opportunity
structure for exchange within formal subunits can become constrained (Marsden 1983), resulting
in the constriction of communication within formal subunits. Taken together, these arguments
suggest:
Hypothesis 1: An increase in uncertainty will lead to an increase (decrease) in contact
among colleagues in different (the same) formal subunits. That is, an increase in
uncertainty will lead to an increase in range across formal subunits.
Network Range across Work Groups
While the uncertainty of restructuring can be expected to increase network range across formal
subunits, I theorize the opposite effect for range across work groups. Whereas the search for
non-redundant information will lead people to activate ties to colleagues in different
departments, there will be no corresponding preference for interaction with colleagues in
Social Capital, Uncertainty, Restructuring
10
different work groups. Because work groups often consist of people from different formal
subunits, colleagues in the same work group are still likely to possess non-redundant
information. Similarly, because they are often outside the potentially shifting formal subunit
structure, they are likely to wield different sources of power and influence than colleagues in the
same subunit. Finally, whereas normative constraints can limit communication within formal
subunits, there are few such limitations to the exchange of information and gossip within work
groups (Balogun and Johnson 2004; Isabella 1990). Thus, to fill the information void within
formal subunits, people are apt to seek information from trusted colleagues in their work groups.
Further support for the notion that the uncertainty of restructuring will lead people to
activate ties to proximate colleagues in the work group structure comes from the affect theory of
social exchange, which posits that “when people interact with others, they tend to experience
mild, everyday feelings, and under some conditions people associate these feelings with a shared
group affiliation or membership” (Lawler, Thye, and Yoon 2009: 9). These affective
attachments can draw a person into exchange with individuals who belong to the same collective
unit. This exchange need not be limited, however, to expressive resources such as social
support; it can also include instrumental resources such as information or influence (Lin 2001).
The theory suggests that individual-to-collective attachments are most likely to occur when
relations in the collective unit are characterized by productive exchange5 – i.e., they involve joint
activity, mutual interdependence, shared control over outcomes, and collective rewards for group
outcomes (Lawler, Thye, and Yoon 2000; Lawler 2001; Lawler, Thye, and Yoon 2008). Formal
subunits vary in the extent to which their members engage in productive exchange. For example,
a manufacturing department with an integrated production line has more interdependent activity
5Other basic forms of social exchange include negotiated, involving explicit and binding agreements; reciprocal,
involving sequences of unilateral giving between a pair of actors; and generalized, involving unilateral exchange in
which givers and receivers are not matched in pairs (Lawler, Thye, and Yoon 2000: 617).
Social Capital, Uncertainty, Restructuring
11
and shared accountability than does a regional sales office with independent sales agents. Yet
neither of these formal subunits has as much interdependence or shared accountability as a work
group, such as a cross-functional product development team (for a detailed discussion of these
examples, see Lawler, Thye, and Yoon [2009: 62-63]). Thus, even in stable times, individual
attachments to work groups are likely to be stronger than those to formal subunits.
For two reasons, I argue that individual attachments to work groups will remain solid,
and even grow stronger, during the uncertainty of restructuring, even while attachments to
formal subunits erode. First, during times of organizational change, the work group structure
tends to remain more stable than the formal subunit structure. For example, project teams, task
forces, and decision making bodies often remain intact even when departmental lines are
redrawn. Moreover, any changes to the work group structure typically lag behind those to
formal subunits (Gulati and Puranam 2009; Lamont, Williams, and Hoffman 1994; Nickerson
and Zenger 2002). Thus, work groups will tend to remain salient collective entities during times
of uncertainty, and individuals’ relatively strong attachments to work groups will tend to persist.
Second, during periods of radical organizational change, managers have been shown to increase
their emotional commitment to work groups, such as project teams responsible for implementing
change initiatives (Huy 2002). These commitments will further draw individuals into exchange
with colleagues in the same work group. Taken together, these arguments suggest:
Hypothesis 2: An increase in uncertainty will lead to a decrease (increase) in contact
among colleagues in different (the same) work groups. That is, an increase in uncertainty
will lead to a decrease in range across work groups.
Social Capital, Uncertainty, Restructuring
12
METHODS
Research Setting
A major information services company, hereafter referred to as InfoCo, served as the research
site for the study. It employed over 8,000 employees and generated over $3 billion in revenue.
The company’s formal structure consisted of global business divisions, regional marketing and
sales units, a global product development unit, and shared support functions (e.g., Finance). In
addition, there were numerous work groups, such as project teams, task forces, and governance
and decision making bodies.
Declining financial performance led InfoCo’s management team to undertake a major
restructuring. There were three major organizational changes. First, InfoCo created global
“solution lines,” which combined product development and marketing resources from different
regions into newly formed formal subunits that had global responsibility for the profitability of a
set of related products and services. Next, InfoCo consolidated the sales and marketing subunits
and downsized redundant personnel in these two functions. Finally, a new global marketing
subunit was established to set standards and ensure consistent implementation across regions.
Many people they lost their jobs, moved departments, or changed supervisors during this period.
Study Participants
The study included all 114 US-based members of the InfoCo’s senior leadership group.6 They
were mostly male (67.5%), white (84.2%), and geographically concentrated: 38.6% in a
Midwestern city, 19.3% in New York City, and the rest were distributed among smaller sites.
They spanned three salary bands (in ascending rank): 7.5% operational leaders, 80.3% tactical
leaders, and 12.2% executive leaders. I collected archived electronic communications and
6Approximately thirty individuals outside the US were also members of the senior leadership team. It was not
possible to include them in the study because of privacy laws and company policies regarding the use of employee
emails in certain countries.
Social Capital, Uncertainty, Restructuring
13
human resource data on all of these individuals and conducted semi-structured interviews with a
subset (see Appendix A; further details below). Because of concerns about how rank-and-file
employees might react if they somehow became aware of the study (e.g., distraction, distrust of
senior management), it was not possible to include a broader cross-section of employees.
The choice to focus on a relatively senior employee population involves clear tradeoffs.
On the one hand, they all had pre-existing ties to one another through their involvement in the
senior leadership group. Thus, they were an appropriate sample for the study of network
activation (rather than new tie formation). In addition, although they were fairly senior, they had
little ex ante knowledge of the restructuring, which the CEO implemented with limited input
from this group. That is, they experienced uncertainty during restructuring. During the time that
archival data were being collected, they were also unaware of this study.7 Moreover, the
qualitative evidence (see Appendix B) suggests that lower level employees experienced greater
levels of uncertainty than the people included in the study. Thus, the focus on senior employees
probably provides a conservative test of the proposed hypotheses; however, it also raises
questions about the extent to which the findings can be generalized to other employee groups.
Further implications of this choice are explored in the Discussion and Conclusion section below.
Before describing the data collected, I present evidence that the people included in the
study did, in fact, experience a high level of uncertainty during restructuring. First, these
individuals were significantly affected by the restructuring: 43 (37.7%) had a change in
supervisor, 15 (13.6%) moved to a different InfoCo division, and 13 (11.4%) exited the
company.8 Some experienced multiple such changes. Second, the qualitative evidence suggests
7The head of human resources and a handful of his staff knew about the study. Knowledge was kept to this small
group to minimize distraction and keep people from altering their communication patterns in response to the study. 8 I included in the quantitative analysis the thirteen who exited because, during the periods that they were employed
by InfoCo, they were at equal risk of exchanging messages as their colleagues who remained throughout the entire
Social Capital, Uncertainty, Restructuring
14
that they felt uncertain and that the restructuring represented an exogenous shock. As one
marketing director reported, “The announcement happens, and then I get a call from HR and the
guy who was my boss at the time. They say, ‘We’re eliminating your role, and you’re not going
to get job you thought you were going to get.’ Then they offered me another job that I really
didn’t want. I was stunned.” Similarly, a director in charge of a marketing support unit recalled:
There was a peer of mine who lost his job. I literally landed at [City] Airport,
checked my messages, and saw that I was invited that afternoon to a call with the
CEO and a strange list of other people. I sent a note to my boss to find out what
this was all about. He called me to say, “Find a place to sit down.” He then told
me that they had eliminated my peer’s position, and I was assuming responsibility
for his group. Within two hours, I had to get ready for a meeting with the CEO
and [everyone] who would now be reporting to me. I had a ton of questions about
this. Was this part of a broader set of changes that would affect me, or was this
the only shoe to drop? Had we just been acquired by another company?
A product development manager reported: “[L]eaders were given a certain number of
slots to fill. We had to go through a process of assessing and ranking people – for example,
eleven people might be ranked for a job role with ten open slots. The eleventh person was laid
off. If the job role was redefined, we had to tell all incumbents that they were laid off and had to
interview to get their job back. Everyone was feeling insecure.” Similarly, a division general
manager stated, “Our reorganizations tend to be big surprise events when they are unveiled.”
Even some of InfoCo’s well-orchestrated efforts to communicate with people about the
changes and help reduce uncertainty often managed to backfire. A director of product
engineering reported:
Soon after we got the memo [announcing the reorganization], they [i.e., senior
management and HR staff] put a bunch of us in a room and told us about the
‘global solutions journey’ we would soon be embarking on. I still remember this
chart in which they put the pictures of the engineering people at the bottom. I
remember thinking, ‘We are the furthest from Heaven in your chart even though
observation period. I dropped them for the periods that they were no longer employed by InfoCo. There were no
significant differences between those who stayed and those who exited on observable characteristics such as tenure,
rank, gender, or ethnicity.
Social Capital, Uncertainty, Restructuring
15
we are most important to delivering on the strategy.’ What does that say about
how you value us and what these changes will mean for us?
Data Collection
I compiled four kinds of archival data: (1) internal communication memos, which I used to
establish the period of greatest uncertainty stemming from the restructuring; (2) email logs
(spanning a period of 40 weeks) of 114 InfoCo employees9; (3) extracts from InfoCo’s Human
Resource (HR) system – including time-invariant measures (e.g., sex, ethnicity, date of hire) and
time-varying measures (e.g., departmental affiliation, office location); and (4) extracts of
InfoCo’s email distribution lists to identify shared work groups among employees (i.e., based on
list co-membership in a given week). Based on these communications and the interviews, I
established that the period of greatest uncertainty commenced in Week 9, when the first of
several communications providing details of the new organizational structure was released.
Additional memos – announcing the formation of global solution line units, the consolidation of
other organizational units, and the appointment and departure of key personnel – were sent every
couple of weeks until Week 18. By Week 18, all of the changes to the organizational structure
had been made, all key positions had been filled, and all departing employees had exited. Thus,
although there was uncertainty throughout the observation period, Weeks 9 through 18
represented the period of peak uncertainty. I conducted an additional analysis to corroborate the
restructuring timeline, which is depicted in Figure 1. I worked with the Human Resources staff
to identify restructuring-related keywords, such as “organizational announcement,”
“organizational appointment,” “departure,” “open position,” and several internal code names
used for the restructuring. I then identified restructuring-related email communication through a
keyword match of email subject lines. The volume of these restructuring messages rose to a
9Prior research indicates that this time period is appropriate for the study of employee reactions to restructuring
(Brockner, Tyler, and Cooper-Schneider 1992; Shah 2000).
Social Capital, Uncertainty, Restructuring
16
peak in Week 9 and began tapering off after Week 18.10
Having established the timeline, I turn
next to a discussion of the relative merits of two data sources: email logs and distribution lists.
– Figure 1 about here –
Data – Email Logs
Analyses of email communication are becoming increasingly common in organizational research
(Allatta and Singh 2011; Hinds and Kiesler 1995; Kossinets and Watts 2006; Menchik and Tian
2008). Consistent with the ethical standards used in prior studies, I took three steps to protect the
privacy and confidentiality of InfoCo employees (for a discussion of ethical issues in
organizational network analysis, see Borgatti and Molina [2005] and Kadushin [2005]). First, all
identifying information (e.g., names, email addresses) was encrypted using an irreversible
algorithm. Second, email logs did not contain message content – just the trace of who sent a
message to whom, at what date and time, and with what subject line. Finally, I only collected
data on messages exchanged among InfoCo employees – i.e., external messages were excluded.
Electronic communications data of this kind have several advantages over traditional
network surveys. First, they can be collected unobtrusively, which can be useful in observing
network dynamics during a politically sensitive time such as an organizational restructuring.
Next, they provide a window into peripheral ties, which network surveys typically do not seek to
measure. Moreover, they can also yield more reliable indicators of interaction than surveys,
which can suffer from various forms of recall and self-report bias (Marsden 2011). For
longitudinal network analysis, they have the added benefit of allowing for consistent data
collection over time (e.g., by eliminating measurement error created by variation in interviewer
techniques or the context in which surveys are filled out) and limiting the risk that people drop
10
The results reported below were robust to slight (i.e., 1-2 week) variations in the choice of the period of greatest
uncertainty.
Social Capital, Uncertainty, Restructuring
17
out of the study due to survey fatigue. Finally, email logs contain fairly complete records of
electronic communication. At InfoCo, interviewees reported that they routinely used company
email even for personal communication. At the time of the restructuring, it was uncommon for
InfoCo employees to use instant messaging or personal email services at work. In addition,
emails sent from personal digital assistants also went through the company servers.
These benefits are counterbalanced by at least three limitations. First, the trace of email
communication between two colleagues does not always signify purposive interaction between
them. For example, email messages can sometimes be automatically generated – e.g., “Out of
Office” message sent when a recipient is on holiday. In addition, emails are sometimes sent to
pre-determined distribution lists (e.g., all employees in the company or on a project team) or
routinely copied to colleagues who have little interest in or need to know the information. I
addressed these shortcomings by eliminating all emails that included the phrase “Out of Office”
in the subject line and by restricting the analysis to emails sent only to a single recipient – i.e.,
mass emails, “cc” or “bcc” messages, and “reply all” messages were all excluded.
Next, email exchanges reflect only a subset of the interactions among people. Face-to-
face meetings, phone calls, and informal gatherings at the water cooler are not captured in email
logs. In this setting, however, the email system was linked to the electronic calendars that
virtually all employees used to maintain their daily schedules. Email logs therefore included a
record of all electronically scheduled meetings. One interviewee explained:
Of course you wouldn’t talk about very sensitive topics over email. You will
generally want to have the sensitive discussions in a face-to-face meeting. But
you might see this kind of interaction reflected in email scheduling traffic. If I
were feeling anxious about my own situation, I would schedule a one-on-one
meeting with my mentor or some other trusted colleague. The other way this kind
of interaction will be reflected in email is through organization-wide
communications. Those messages tend to get cascaded down the organization.
Social Capital, Uncertainty, Restructuring
18
Especially when you get further down the organization, people will often forward
those messages to someone they know or trust to help them clarify what it means.
Third, email traffic may simply reflect routine communication (e.g., negotiating a
convenient time to meet), rather than meaningful efforts to exchange resources such as
information or influence. I addressed this limitation by taking into account the date and time
stamp of messages. In particular, I separately analyzed messages sent outside business hours
(i.e., early mornings, late evenings, weekends, and holidays). The qualitative evidence suggested
that email traffic outside of normal business hours was more likely to reflect the exchange of
social resources. As a sales executive stated, “The off-hours communication tend to blur the
business and the social a lot more. Your laptop computer can swing from being strictly business
to strictly personal in a matter of minutes.” A vice president with young kids reported: “I’ve got
pretty hefty family obligations, so I try not to do a lot of email over the weekend. But the people
I tend to be in touch with at those times are my tighter-knit group.”
These limitations notwithstanding, it is worth noting that people often feel less inhibited
in email communication than in face-to-face communication (Sproull and Kiesler 1986) and that,
over time, computer-mediated teams can develop levels of trust that are comparable to those in
face-to-face teams (Wilson, Straus, and McEvily 2006). Thus, it is likely that email exchanges
included at least an important subset of sensitive communication about the exchange of social
resources. Still, email communication likely represents a conservative indicator of overall shifts
in network range during politically sensitive periods such as restructuring.
Data – Email Distribution Lists
Just as the choice to use email data involves tradeoffs, so too does the decision to use email
distribution lists to locate individuals in the work group structure. Widely used across
organizations and readily accessible, distribution lists encapsulate the myriad collective units that
Social Capital, Uncertainty, Restructuring
19
exist within an organization. Some lists, of course, trace the formal organizational structure. For
example, a department head might create a list for all department members. Yet, in a typical
organization, email lists also exist for various standing cross-departmental teams, ad hoc task
forces, and professional interest groups that are active. Colleagues with a large proportion of
lists in common are therefore likely to be more proximate in the work group structure than are
colleagues with few shared lists. Furthermore, changes in list membership reflect movement
within the work group structure. Distribution lists are not, however, a perfect data source. Not
every work group in the organization has a corresponding list. Moreover, in the data I collected
at InfoCo, list names were encrypted for confidentiality reasons and could therefore not be
separated into those corresponding to formal subunits, to work groups, and to social groups.
In principle, the work group structure could overlap significantly with the formal
organizational structure, for example if work groups were entirely nested within the formal
subunits. In practice, however, the two measures appeared to reflect non-overlapping
dimensions of network range. There were over 2,300 distribution lists in use during a typical
week in this observation period; the mean number of lists to which an employee belonged was
12.2. The number of distribution lists in use far exceeded the number of departments in which
the sample of 114 employees worked. Moreover, based on a median split of distance in the work
group structure (defined below), only 4.3% of dyads in the sample were both proximate in the
work group structure and in the same department. Thus, at least in this setting, distance in the
work group structure was relatively independent of distance across formal subunits. Interviews
with InfoCo employees further indicated that these lists were used primarily for work groups,
especially project teams that did not correspond to formal subunits. A product development
manager explained, “I use [distribution lists] for very specific project-related activity. People
Social Capital, Uncertainty, Restructuring
20
have gotten so weary of email that we’ve had a push to narrow distribution lists to work-related
projects that are active at a given point in time.” Another interviewee explained, “I maintain
distribution lists for my direct reports, my ‘extended direct reports’ in other departments, and
[the product development team I lead].” A female vice president reported, “The only list I use
that isn’t about project-related work is the Women’s Network. The rest are projects.”
Data – Supplemental Semi-Structured Interviews
To help address the limitations of the archival data sources described above, I conducted
supplemental semi-structured interviews with 23 study participants after the restructuring
concluded. These individuals were selected from stratified sub-samples of people who
experienced high and low levels of uncertainty during the restructuring but remained employed
by InfoCo. Legal concerns kept the company from granting me access to those who had exited.
Interviews lasted between 30 and 45 minutes and were recorded and transcribed. The purpose of
the interviews was to validate the timeline of events, assess the level of uncertainty people
experienced, understand how and why people activated their networks during the restructuring,
and determine how they used electronic communication media in this organizational setting. I
used a software tool – Atlas.ti – to code and analyze the responses. I paid particular attention to
the network activation choices described by respondents, coding the kinds of people who were
contacted (e.g., same department or different department; shared work group), the resources
sought in these interactions (e.g., information, influence, social support), and other factors that
promoted or inhibited communication (e.g., normative constraints, socially awkward situations).
Social Capital, Uncertainty, Restructuring
21
Measures
The response variable was a count of the number of one-to-one email messages sent in a given
week, t, between a dyadic pair, i and j.11
For the reasons noted above, I also separated out
messages sent outside of business hours (i.e., weekday messages sent before 6 am or after 8 pm
and all weekend and holiday messages). Time zones were not materially relevant, given that
over 90% of the sample were located in the eastern United States.
Given the conceptual focus on network range – i.e., the diversity of actors to which one is
connected – explanatory variables were all expressed as differences (e.g., different sex) or
distance between a pair of actors rather than as similarities (e.g., same sex) or proximity. The
time-varying indicator variable of range across the formal subunits was: Different Departmentt
(set to 1 if i and j were in different departments in week t and to 0 otherwise). For range across
work groups, I considered the distance between two actors based on the number of email
distribution lists to which they both belonged. I employed one of the most widely used distance
measures, Jaccard’s distance (Sneath and Sokal 1973)12
:
Distance in Work Group Structuret = 1 – Si,j / (Ni+Nj-Si,j)
Where: Si,j = Shared distribution lists between i and j in week t
Ni = Number of lists to which i belonged in week t
Nj = Number of lists to which j belonged in week t
This measure has a theoretical range from 0, for a pair of actors who belong to distribution lists
that perfectly overlap one another, to 1, for a pair of actors who have no shared lists. Because
11
Comparable results to those reported below were obtained when the response variable was (undirected) messages
exchanged between dyads rather than (directed) messages sent. I also varied the mass email threshold to include up
to four recipients per message. The results reported below were materially unchanged. Finally, an alternative way
to operationalize network range is to consider the conditional log-odds of any contact (i.e., a dichotomous response
variable) between a dyadic pair, rather than the intensity of contact (see, for example, Reagans and McEvily [2003]).
Logit models using a dichotomous indicator produced comparable results to the Poisson models reported below. I
prefer the Poisson framework because it better accounts for the fact that resource exchange likely occurs over the
course of multiple messages. 12
This covariate is centered when included in regression analyses (Aiken and West 1991) but left uncentered in the
descriptive statistics.
Social Capital, Uncertainty, Restructuring
22
the results reported below were robust to the use of alternative distance measures, such as one
based on Dice’s coefficient (Dice 1945)13
, I report only those based on Jaccard’s distance.
To identify the effects of uncertainty on network range, I included two interaction terms:
Uncertainty x Different Departmentt and Uncertainty x Distance in Work Group Structuret.
Positive coefficients for these interaction terms indicate an increase in network range during the
period of uncertainty, while negative coefficients suggest a decrease in range.
Because various individual differences, such as the need for cognitive closure (Webster
and Kruglanski 1994) and the personal need for structure (Neuberg and Newsom 1993), are
known to shape reactions to uncertainty, I included fixed effects for every message sender and
every message receiver to account for such unobserved heterogeneity. To capture temporal
variation in communication exchange (e.g., dips during holiday weeks), I included period (i.e.,
week) fixed effects.14
I also included a number of dyad-level control variables: (a) Different
Locationt; (b) Different Salary Gradet; (c) Different Sex; (d) Different Ethnicity15
; (e) Different
Cohort (i.e., hire dates separated by more than one year); and (f) Different Age (i.e., difference of
more than three years).
Estimation
I constructed a dyad-level panel data set of messages sent between i and j in week, t. Analyses
of such data must contend with the clustering (i.e., non-independence) of observations. The
failure to control for clustering can lead to under-estimated standard errors and over-rejection of
hypothesis tests. I addressed this issue by using a variance estimator that enables cluster-robust
inference when there is two-way or multi-way clustering (Cameron, Gelbach, and Miller 2011).
13
This measure is computed by dividing shared lists by the sum of lists to which each member of a dyadic pair
belongs. As noted above, this measure produced comparable results to those reported below. 14
Week fixed effects subsume the main effects of the period of uncertainty. Including an indicator for uncertainty
period (Weeks 9-18), instead of week fixed effects, did not have any material effect on the results reported below. 15
Because 84% of the population was white, I considered only two categories of ethnicity: white and non-white.
Social Capital, Uncertainty, Restructuring
23
This situation arises when – as in this study – there is clustering at both the cross-sectional and
temporal levels. In the case of two-way clustering, the technique produces three different
variance matrices: for the first dimension, for the second dimension, and for the intersection of
the two. The first two matrices are added together and third subtracted. In the case of three-way
clustering, the analogous technique results in the creation and combination of seven one-way
cluster robust variance matrices.16
This technique is appropriate for the analysis of dyadic
network data, including panel data, and compares favorably in simulation studies to alternative
methods, such as the Quadratic Assignment Procedure (Lindgren 2010). Following this
approach, and because the response variable was a count of messages sent between dyadic pairs
in a given week, I estimated fixed effect Poisson regressions (Cameron and Trivedi 1986) with
three-way clustering of standard errors: by sender, by receiver, and by week.
RESULTS
Quantitative Analysis
Table 1 reports descriptive statistics and a correlation matrix for the main variables of interest.
As expected, there was a negative correlation between messages sent and various measures of
dissimilarity between dyads (e.g., Different Departmentt and Different Locationt).
– Table 1 about here –
Table 2 provides a comparison of aggregate communication patterns between the periods
of uncertainty and relative stability. Although there was a slight increase in communication
16
Each of the first three matrices clusters in one of the dimensions. Because some observation pairs are in the same
cluster in two dimensions, considering only these three matrices would result in double counting. The technique
then clusters on the three combinations of two dimensions and subtracts the resulting matrices. This eliminates
double counting but does not account for pairs that share the same cluster in all three dimensions. So the seventh
matrix, which clusters on pairs sharing the same cluster in all dimensions, is added back (see Cameron, Gelbach, and
Miller [2011: 10-11] for a detailed explanation). This approach also controls for potential over- or under-dispersion
in the data. I implemented it in STATA using the “clus_nway” script (Kleinbaum, Stuart, and Tushman 2011).
Social Capital, Uncertainty, Restructuring
24
volume during the weeks of uncertainty, this change was not statistically significant. The
proportion of messages sent between colleagues in different departments was 0.484 in the period
of uncertainty and 0.432 in the period of relative stability (p<.001). This pattern is consistent
with Hypothesis 1 – that uncertainty will increase network range across the formal structure.
The correlation between Distance in Work Group Structuret and messages sent was -0.148 in the
period of uncertainty and -0.124 in the period of relative stability. This result is consistent with
Hypothesis 2 – that uncertainty will decrease range across the work group structure.
– Table 2 about here –
Table 3 reports the results of the Poisson regression models used to formally test
Hypotheses 1 and 2. Models 1 and 2 show baseline results for all messages sent and for
messages sent outside of business hours, respectively. Different Locationt, Different Sex,
Different Departmentt, and Distance in Work Group Structuret have negative coefficients that are
statistically significant. The negative coefficients for Different Departmentt and Distance in
Work Group Structuret are consistent with prior research indicating a tendency for workplace
networks to hew to the formal organizational structure and work flow (Han 1996; Hinds and
Kiesler 1995; Ibarra 1992). It is also worth noting that Different Salary Gradet has a positive
and significant coefficient. One explanation for the lack of a negative relationship with Different
Salary Gradet is that the study population was relatively homogeneous in organizational rank.
Models 3 and 4 add the interaction terms of interest: Uncertainty x Different Departmentt
and Uncertainty x Distance in Work Group Structuret. The former has a positive coefficient that
is significant at the 95% confidence level in both models, while the latter has a negative
coefficient that is significant at the 95% confidence level in both models. These effects were
Social Capital, Uncertainty, Restructuring
25
more pronounced in the models including only off-business hour communication.17
These
results provide support for both hypotheses. Considering that changes in email communication
probably represent a conservative indicator of overall shifts in network activation, these effects
were sizable: In the period of uncertainty relative to stability, there was a 21% decline in the
predicted number of messages sent between colleagues in the same department and a 7%
increase in the predicted number of messages sent between colleagues in different departments.
For dyads at the median of the Distance in Work Group Structuret measure, there was an 18%
decline in the predicted number of messages sent in the period of uncertainty relative to stability.
– Table 3 about here –
Table 4 reports results that help establish that these effects were in fact driven by the
uncertainty of restructuring. I worked with representatives from InfoCo’s Human Resources
department to identify the job roles that, at the time of the restructuring announcement, seemed
to be at greatest risk of being affected by the changes. For example, a person responsible for
product marketing faced greater uncertainty from the impending creation of global solution lines
than a person responsible for legal compliance (which was largely unaffected by the announced
changes). I identified 27 job roles that involved high levels of uncertainty. Of the 12,882 dyads,
5,742 (45%) included at least one member who held one of these 27 job roles. These dyads
experienced greater uncertainty from restructuring than the 7,140 (55%) dyads in which neither
member held one of these roles. I used seemingly unrelated post-estimation to compare the size
of coefficients in Model 5, which included dyads experiencing low uncertainty, to those in
Model 6, which included dyads experiencing high uncertainty. Uncertainty x Different
17
I used seemingly unrelated post-estimation to compare the size of the coefficients in a model including only off-
business hour communication with those in a model including only business hour communication. The increase in
network range with respect to the formal structure was more pronounced in off-business hour communication
(p<0.05), and the decrease in network range with respect to the work group structure was also more pronounced,
though marginally significant, in off-business hour communication (p<0.10).
Social Capital, Uncertainty, Restructuring
26
Departmentt was significantly more positive (p<.01) and Uncertainty x Distance in Work Group
Structuret was significantly more negative (p<.01) in the model with dyads experiencing high
uncertainty than in the model with dyads experiencing low uncertainty.18
Thus, it appears that
the observed shifts in network range occurred in response to heightened uncertainty.
– Table 4 about here –
Finally, to understand whether communication patterns reverted to their original state
when the uncertainty period ended, I used a three-period model, with interactions for the period
of uncertainty (e.g., Uncertainty x Different Departmentt) and the period after uncertainty (e.g.,
Post-Uncertainty x Different Departmentt). In the three-period model, Uncertainty x Different
Departmentt was positive and marginally significant (beta=.236, p=.086), and the linear
combination of Uncertainty x Different Departmentt – Post-Uncertainty x Different Departmentt
was positive and significant (beta=.358, p=.031). That is, different department communication
appeared to increase during the period of uncertainty but then wane as uncertainty receded. By
contrast, Uncertainty x Distance in Work Group Structuret was negative and significant (beta=-
1.040, p=.023), while the linear combination of Uncertainty x Distance in Work Group Structuret
– Post-Uncertainty x Distance in Work Group Structuret was of the expected (negative) sign but
not significant (beta=-0.626, p=.149). Thus, it appeared that uncertainty had a transient effect on
network range across formal subunits and a more lasting effect on range across work groups.
Supplemental Qualitative Analysis
The semi-structured interviews helped reveal the motivations underlying these communication
shifts. People reported increasing contact with colleagues in other departments to gather
18
Comparable results were obtained when I instead created an indicator variable for the dyads experiencing more
uncertainty and used three-way interaction terms, e.g., Uncertainty Period x Different Departmentt x Experienced
More Uncertainty, and relevant two-way interaction terms to estimate the effect. Similarly, results were robust to
the use of messages exchanged, rather than sent, as the response variable.
Social Capital, Uncertainty, Restructuring
27
intelligence about the restructuring and to position themselves politically for favorable career
outcomes. In choosing whom to contact outside their department, they reported selecting
colleagues with whom they had a trust-based relationship based on a history of working together.
In many cases, they explicitly indicated reaching out to colleagues from project teams with
whom they had built a strong connection on the basis of prior productive exchange – for
example, working together on a difficult assignment. These interactions tended to be driven by a
search for trustworthy information, as well as for social support.19
Finally, the decline in within-
department communication was the result of normative constraints on supervisor-subordinate
communication, as well as socially awkward situations created by restructuring. Table 5
summarizes these points and includes representative quotes.
– Table 5 about here –
ROBUSTNESS CHECKS
I conducted additional robustness checks to address potential alternative explanations for these
results. First, the increase in contact across formal subunits could be explained by anticipated
role transitions – for example, a person reaching out to a likely known future supervisor – or
shifting task interdependencies – such as a person beginning to perform new job responsibilities
prior to a changing roles or completing tasks from a prior role after making the transition. To
account for these transitions, I controlled for lagged and future departmental affiliations and
distance measures and also included a control for the person’s departmental at the end of the
observation period (i.e., week 40).20
These analyses produced comparable results to those
19
People also reported turning to past colleagues with whom they did not necessarily have a current working
relationship, friends within and outside the organization, and family members for social support. 20
That is, I included Different Departmentt-1, Different Departmentt-2, Different Departmentt-3, Distance in Work
Group Structuret-1, Distance in Work Group Structuret-2, Distance in Work Group Structuret-3, Different
Social Capital, Uncertainty, Restructuring
28
presented in Table 3. Second, I analyzed the content of email subject lines to address the
possibility that people somehow suspected their email was being monitored by senior
management and shifted their communication patterns accordingly (i.e., reducing the volume of
frivolous messages sent to supervisors or same-department colleagues). I coded messages as
frivolous based on their subject lines – e.g., those including phrases such as “Beer?” “Play ball!”
“Golf,” and “Gasoline Cartoons.” The proportion of such messages did not vary significantly
across time periods.21
Finally, to account for the role of competition among actors – for
example, a decline in contact if two people were vying for the same job – I used job families, as
defined by the HR system, to construct another control variable: Different Job Familyt.
Individuals in the same job family would presumably be more likely to compete with one
another for the same job. The results reported in Table 3 were materially unchanged with the
inclusion of this control.
DISCUSSION AND CONCLUSION
The goal of this article has been to clarify how the uncertainty of a transformative event – in this
case, restructuring – affects the activation of social capital within organizations. Restructuring-
induced uncertainty exerts two opposing forces on the range, or diversity, of network ties people
activate. On one hand, it leads to a preference for exchange with trusted, socially proximate
partners (e.g., Buchan, Croson, and Dawes 2002). This tendency suggests a contraction in the
Departmentt+1, Different Departmentt+2, Different Departmentt+3, and Distance in Work Group Structuret+1,
Distance in Work Group Structuret+2, and Distance in Work Group Structuret+3 as controls. The interaction terms of
interest, Uncertainty x Different Departmentt and Uncertainty x Distance in Work Group Structuret were significant
and of comparable sign and magnitude to the coefficients reported in Table 3. 21
Because the formal and work group structures are not perfectly orthogonal (e.g., some work groups are nested
within formal subunits), I also tested but found no evidence for a potential three-way interaction; i.e., Uncertainty x
Different Departmentt x Distance in Work Group Structuret was not significant in specifications that included this
term and relevant two-way interactions.
Social Capital, Uncertainty, Restructuring
29
range of activated networks. On the other hand, it can draw people into exchange with distant
partners, who wield resources such as non-redundant information and influence that become
more valuable under uncertainty (e.g., Friedken 1982; Pfeffer 1992). This propensity implies an
expansion in the range of activated networks. I reconcile this conceptual tension by
disentangling uncertainty’s effects on range across formal subunits and cross-cutting work
groups in the organization. Drawing on insights about social capital activation (e.g., Smith
2005), the dynamics of organizational structure in times of change (e.g., Gulati and Puranam
2009), and collective attachments in social exchange (e.g., Lawler, Thye, and Yoon 2008), I
argue that uncertainty will lead to an expansion in range across formal subunits because people
will seek non-redundant information and political influence from colleagues in other departments
(Friedkin 1982; Pfeffer 1992). Moreover, organizational norms will constrain exchange within
formal subunits, leading people to reduce communication with departmental colleagues (Klein
1996; Marsden 1983). Thus, an increase in uncertainty will lead to an increase in network range
across formal subunits.
By contrast, I argue that there will be no corresponding preference for interaction with
colleagues in different work groups; rather, people will tend to activate ties to proximate
colleagues in the work group structure. Because work groups often consist of individuals from
different formal subunits, colleagues in the same work group are likely to have access to non-
redundant information. Because these individuals are outside the potentially changing formal
subunit structure, they will also tend to wield different sources of power and influence than
colleagues in the same subunit. In addition, because there are few normative constraints on
communication within work groups during restructuring (Balogun and Johnson 2004; Isabella
1990), people will fill the information void in their formal subunits by turning to colleagues in
Social Capital, Uncertainty, Restructuring
30
the same cross-cutting work groups. Further support for the prediction that uncertainty will
increase individual attachments to work groups comes from the affect theory of social exchange
(Lawler, Thye, and Yoon 2009). Because the work group structure tends to be more stable
during restructuring than the formal subunit structure and because any changes to work groups
tend to lag behind those to formal subunits (e.g., Gulati and Puranam 2009), individual-to-
collective ties to work groups will remain strong even while those to departments erode (Lawler,
Thye, and Yoon 2009). Finally, during times of organizational change, managers tend to
increase their emotional commitment to work groups (Huy 2002). These factors all suggest that
an increase in uncertainty will lead to a decrease in this network range across work groups. The
empirical evidence supports these propositions about uncertainty’s effects on activated network
range across formal subunits and work groups.
I turn next to two outstanding questions raised by these findings. First, did network
activation choices during the uncertainty of restructuring affect individuals’ subsequent
outcomes? Given the research design, it was not possible to establish a causal link; however, I
examined the association between network activation during uncertainty and the likelihood of
exit from the firm fourteen months after restructuring. Given that the US economy was in a
significant downturn at the time, it is likely – though not known – that many of these exits were
involuntary. These results, which are reported in Appendix C, suggest that those who sent more
messages with colleagues in different departments than predicted by a baseline model had lower
conditional log-odds of exit. This association should, however, be considered preliminary
because other unobserved factors (e.g., external job prospects) may have influenced exit
decisions. Second, the focus on a relatively senior employee population in a single organization
raises the question about the extent to which these findings can be generalized to other settings.
Social Capital, Uncertainty, Restructuring
31
Although the evidence in Appendix B suggests that lower-ranking employees experienced even
greater uncertainty than those involved in this study, further work is needed to establish that
these same patterns hold across different employee populations and other types of organizations.
The findings from this study make three primary contributions. First, they enhance our
understanding of social capital activation. Whereas prior work in this tradition has examined the
job searches through which people gain entry into organizations (Bian 1997; Lai, Lin, and Leung
1988; Lin, Ensel, and Vaughn 1981; Marsden and Hurlbert 1988; Wegener 1991), this study
instead exposes the difficult-to-observe dynamics of intra-organizational network activation. In
addition to the factors – such as trust-based (Smith 2005) and interpersonal affect (Casciaro and
Lobo 2008) – that are known to drive a wedge between the actual and potential resources actors
obtain through networks, this study highlights the role of organizational factors such as norms
that govern supervisor-subordinate relations. In this setting, supervisors sought to maintain
professionalism by not divulging information to just a subset of their subordinates. At the same
time, subordinates felt inhibited in communicating with supervisors whose own career outcomes
were unclear. The net effect was a constriction of contact between subordinates and supervisors,
some of whom could have been potentially useful sources of information, influence, or social
support. At the same time, the study points to the role of work groups as an enabler of network
activation and a potentially valuable conduit for resource exchange during restructuring. In
addition, whereas prior research on network activation has tended to considered ordinary
contexts, this work joins a handful of studies that consider non-routine times. For example, prior
research on the activation choices of people who experienced the effects of Hurricane Andrew
reports that pre-existing core network structure influenced the use of core network ties for
informal support (Hurlbert, Haines, and Beggs 2000). The present study reveals that uncertainty
Social Capital, Uncertainty, Restructuring
32
triggers not only the activation of core networks but also of peripheral ties, such as those that
span formal subunit boundaries. Moreover, it uncovers how people activate networks to obtain
not only social support but also instrumental resources such as information and influence.
Second, this study brings to the literature on attachments in social exchange (Lawler
2001; Lawler, Thye, and Yoon 2008; Lawler and Yoon 1998) insight into the dynamics of
individual-to-collective attachments (see also McPherson and Smith-Lovin [2002]). Previous
research has identified the exchange conditions that are most likely to produce micro social
orders – “the recurring patterns of activity that orient people toward members of a social unit”
(Lawler, Thye, and Yoon 2008: 520). These findings suggest that, under conditions of
uncertainty, micro social orders inside organizations vary in their degree of stability. Those
corresponding to formal subunits tend to be relatively fragile, while those defined by other work
groups tend to endure and may in fact be strengthened by uncertainty. Moreover, while the core
insights about individual-to-collective attachments have been developed in laboratory
experiments, this study demonstrates the theory’s applicability in a field setting.
Third, this study has important implications for research on organizational structure and
performance in turbulent times (Davis, Eisenhardt, and Bingham 2009; Krackhardt and Stern
1988; Lin et al. 2006; Rindova and Kotha 2001). At the organizational level, Krackhardt and
Stern (1988) argue that the structure of internal friendship ties within organizations can influence
their ability to survive crisis situations. In particular, firms with a high ratio of cross- to within-
subunit friendship ties – i.e., a high External-Internal (E-I) Index – were more effective at
surviving crises in a simulation game. Findings from the present study suggest the need to
complicate this account. Whereas the experimentally manipulated organizations created by
Krackhardt and Stern (1988) varied in the structure of intraorganizational ties, this study
Social Capital, Uncertainty, Restructuring
33
suggests the need to also consider network action, in the form of activated networks. In
particular, it may be inadequate to consider a single E-I index, which remains static over time
and determines an organization’s ability to withstand turbulent times. Instead, we must consider
at least two forms of the E-I index – one based on formal subunits and one based on other work
groups. Conditions of uncertainty can cause the E-I index for formal subunits to increase and the
E-I index for work groups to decrease. It remains to be explored how these shifts in E-I index
influence an organization’s ability to survive uncertain crises. At the individual level, Soda and
Zaheer (forthcoming) examine the performance implications of inconsistencies between an
actor’s informal network and her position in the formal authority and workflow structure of the
organization. Based on a cross-sectional network survey and a mapping of individuals within the
formal authority and workflow structures, they find that consistency between the informal
network and formal authority structure supports attainment, while the effects of consistency
between the informal network and workflow structure vary by type of coordination. Although
their study importantly highlights the interplay of formal structures and social networks and
develops a useful method for examining inconsistencies, it also takes a static view of networks.
This study reveals that, during critical junctures in careers – such as restructuring – the
consistency between networks and different organizational structures can endogenously change.
Further work is needed to assess the implications for individual attainment (see, for example,
Burt [1992; 2005]).
Beyond these three core contributions, this work has implications for two other
literatures. For research on events that transform social structure (Clemens 2007; Sewell 1996;
Sewell 2005), this study underscores the importance of the choice of time intervals, which can
shape whether “cases are understood as episodes of transformation or stretches (in time or space)
Social Capital, Uncertainty, Restructuring
34
of coherence and continuity” (Clemens 2007: 543). For example, a recent study of email
communication in a post-merger integration takes four-month communication snapshots and
concludes “worker routines are slow to change even when a transformative event such as an
acquisition occurs” (Allatta and Singh 2011: 1111). Although four-month intervals seem
appropriate to understanding changes in routines, this choice probably also obscured countless
individual reactions to uncertain micro events, such as announcements of key personnel changes.
What appeared as a period of stability in their data may instead have encompassed many spells
of network change. Moreover, if network activation during these micro periods of uncertainty
solidified connections among work groups, then the observed stability in routines at four-month
intervals may even have been reinforced by the cumulative effects of episodic tumult.
Next, this work informs sociological research on the aftermath of organizational