Two of the competencies you must develop in order to become an independent scholar include writing in a scholarly voice and ton

CHANGE AGENTS, NETWORKS, AND INSTITUTIONS: A CONTINGENCY THEORY OF ORGANIZATIONAL CHANGE

JULIE BATTILANA Harvard University

TIZIANA CASCIARO University of Toronto

We develop a contingency theory for how structural closure in a network, defined as terms of the extent to which an actor’s network contacts are connected to one another, affects the initiation and adoption of change in organizations. Using longitudinal survey data supplemented with eight in-depth case studies, we analyze 68 organiza- tional change initiatives undertaken in the United Kingdom’s National Health Service. We show that low levels of structural closure (i.e., “structural holes”) in a change agent’s network aid the initiation and adoption of changes that diverge from the institutional status quo but hinder the adoption of less divergent changes.

Scholars have long recognized the political na- ture of change in organizations (Frost & Egri, 1991; Pettigrew, 1973; Van de Ven & Poole, 1995). To implement planned organizational changes—that is, premeditated interventions intended to modify the functioning of an organization (Lippitt, 1958)— change agents may need to overcome resistance from other members of their organization and en- courage them to adopt new practices (Kanter, 1983; Van de Ven, 1986). Change implementation within an organization can thus be conceptualized as an exercise in social influence, defined as the alteration of an attitude or behavior by one actor in response to another actor’s actions (Marsden & Friedkin, 1993).

Research on organizational change has improved understanding of the challenges inherent in change implementation, but it has not accounted system- atically for how characteristics of a change initia- tive affect its adoption in organizations. Not all

organizational changes are equivalent, however. One important dimension along which they vary is the extent to which they break with existing insti- tutions in a field of activity (Battilana, 2006; Green- wood & Hinings, 2006). Existing institutions are defined as patterns that are so taken-for-granted that actors perceive them as the only possible ways of acting and organizing (Douglas, 1986). Consider the example of medical professionalism, the insti- tutionalized template for organizing in the United Kingdom’s National Health Service (NHS) in the early 2000s. According to this template, physicians are the key decision makers in both the administra- tive and clinical domains. In this context, central- izing information to enable physicians to better control patient discharge decisions would be aligned with the institutionalized template. By con- trast, implementing nurse-led discharge or pread- mission clinics would diverge from the institu- tional status quo by transferring clinical tasks and decision-making authority from physicians to nurses. Organizational changes may thus converge with or diverge from an institutional status quo (Amis, Slack, & Hinings, 2004; D’Aunno, Succi, & Alexander, 2000; Greenwood & Hinings, 1996). Changes that diverge from the status quo, hereafter referred to as divergent organizational changes, are particularly challenging to implement. They re- quire change agents to distance themselves from their existing institutions and persuade other or- ganization members to adopt practices that not only are new, but also break with the norms of their institutional environment (Battilana, Leca, & Box- enbaum, 2009; Greenwood & Hinings, 1996; Kel- logg, 2011).

We would like to thank Wenpin Tsai and three anon- ymous reviewers for their valuable comments on earlier versions of this article. We also wish to acknowledge the helpful comments we received from Jeffrey Alexander, Michel Anteby, Joel Baum, Peter Bracken, Thomas d’Aunno, Stefan Dimitriadis, Martin Gargiulo, Mattia Gilmartin, Ranjay Gulati, Morten Hansen, Herminia Ibarra, Sarah Kaplan, Otto Koppius, Tal Levy, Christo- pher Marquis, Bill McEvily, Jacob Model, Lakshmi Ra- marajan, Metin Sengul, Bill Simpson, Michael Tushman, and seminar participants at INSEAD, McGill, Harvard, Bocconi, and HEC Paris.

Both authors contributed equally to this article. Editor’s note: The manuscript for this article was ac-

cepted for publication during the term of AMJ’s previous editor-in-chief, R. Duane Ireland.

� Academy of Management Journal 2012, Vol. 55, No. 2, 381–398. http://dx.doi.org/10.5465/amj.2009.0891

381

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s express written permission. Users may print, download, or email articles for individual use only.

In this article, we examine the conditions under which change agents are able to influence other organization members to adopt changes with differ- ent degrees of divergence from the institutional status quo. Because informal networks have been identified as key sources of influence in organiza- tions (Brass, 1984; Brass & Burkhardt, 1993; Gar- giulo, 1993; Ibarra, 1993; Ibarra, 1993; Krackhardt, 1990) and policy systems (Laumann, Knoke, & Kim, 1985; Padgett & Ansell, 1993; Stevenson & Green- berg, 2000), we focus on how change agents’ posi- tions in such networks affect their success in initi- ating and implementing organizational change.

Network research has shown that the degree of structural closure in a network, defined as the ex- tent to which an actor’s network contacts are con- nected to one another, has important implications for generating novel ideas and exercising social influence. A high degree of structural closure cre- ates a cohesive network of tightly linked social actors, and a low degree of structural closure cre- ates a network with “structural holes” and broker- age potential (Burt, 2005; Coleman, 1988). The ex- isting evidence suggests that actors with networks rich in structural holes are more likely to generate novel ideas (e.g., Burt, 2004; Fleming, Mingo, & Chen, 2007; Rodan & Galunic, 2004). Studies that have examined the effect of network closure on actors’ ability to implement innovative ideas, how- ever, have yielded contradictory findings, some showing that high levels of network closure facili- tate change adoption (Fleming et al., 2007; Obst- feld, 2005), others showing that low levels of net- work closure do so (Burt, 2005).

In this study, we aim to reconcile these findings by developing a contingency theory of the role of network closure in the initiation and adoption of organizational change. We posit that the informa- tion and control benefits of structural holes (Burt, 1992) take different forms in change initiation than in change adoption, and these benefits are strictly contingent on the degree to which a change di- verges from the institutional status quo in the or- ganization’s field of activity. Accordingly, struc- tural holes in a change agent’s network aid the initiation and adoption of changes that diverge from the institutional status quo but hinder the adoption of less divergent changes.

In developing a contingency theory of the hith- erto underspecified relationship between network closure and organizational change, we draw a the- oretical link between individual-level analyses of network bases for social influence in organizations and field-level analyses of institutional pressures on organizational action. We thus aim to demon- strate the explanatory power that derives from rec-

ognizing the complementary roles of institutional and social network theory in a model of organiza- tional change. To test our theory, we collected data on 68 organizational changes initiated by clinical managers in the United Kingdom’s National Health Service (NHS) from 2004 to 2005 through longitu- dinal surveys and eight in-depth case studies.

NETWORK CLOSURE AND DIVERGENT ORGANIZATIONAL CHANGE

In order to survive, organizations must convince the public of their legitimacy (Meyer & Rowan, 1977) by conforming, at least in appearance, to the prevailing institutions that define how things are done in their environment. This emphasis on legit- imacy constrains change by exerting pressure to adopt particular managerial practices and organiza- tional forms (DiMaggio & Powell, 1983); therefore, organizations embedded in the same environment, and thus subject to the same institutional pres- sures, tend to adopt similar practices.

Organization members are thus motivated to ini- tiate and implement changes that do not affect their organizations’ alignment with existing institutions (for a review, see Heugens and Lander [2009]). Nev- ertheless, not all organizational changes will be convergent with the institutional status quo. In- deed, within the NHS, although many of the changes enacted have been convergent with the institutionalized template of medical professional- ism, a few have diverged from it. The variability in the degree of divergence of organizational changes poses two questions: (1) what accounts for the like- lihood that an organization member will initiate a change that diverges from the institutional status quo and (2) what explains the ability of a change agent to persuade other organization members to adopt such a change.

Research into the enabling role of actors’ social positions in implementing divergent change (Greenwood & Hinings, 2006; Leblebici, Salancik, Copay, & King, 1991; Maguire, Hardy, & Lawrence, 2004; Sherer & Lee, 2002) has tended to focus on the position of the organization within its field of activity, eschewing the intraorganizational level of analysis. The few studies that have accounted for intraorganizational factors have focused on the in- fluence of change agents’ formal position on the initiation of divergent change and largely over- looked the influence of their informal position in organizational networks (Battilana, 2011). This is surprising in light of well-established theory and ev- idence concerning informal networks as sources of influence in organizations (Brass, 1984; Brass & Burkhardt, 1993; Gargiulo, 1993; Ibarra & Andrews,

382 AprilAcademy of Management Journal

1993; Ibarra, 1993; Krackhardt, 1990). To the extent that the ability to implement change hinges on social influence, network position should signifi- cantly affect actors’ ability to initiate divergent changes and persuade other organization members to adopt them.

A network-level structural feature with theoreti- cal relevance to generating new ideas and social influence is the degree of network closure. A con- tinuum of configurations exists: cohesive networks of dense, tightly knit relationships among actors’ contacts are at one end, and networks of contacts separated by structural holes that provide actors with brokerage opportunities are at the other. A number of studies have documented the negative relationship between network closure and the gen- eration of new ideas (Ahuja, 2000; Burt, 2004; Fleming et al., 2007; Lingo & O’Mahony, 2010; Mc- Fadyen, Semadeni, & Cannella, 2009). Two mech- anisms account for this negative association: re- dundancy of information and normative pressures (Ruef, 2002). With regard to the former, occupying a network position rich in structural holes exposes an actor to nonredundant information (Burt, 1992). To the extent that it reflects originality and new- ness, creativity is more likely to be engendered by exposure to nonredundant than to repetitious in- formation. As for normative pressure, network co- hesion not only limits the amount of novel infor- mation that reaches actors, but also pressures them to conform to the modus operandi and norms of the social groups in which they are embedded (Cole- man, 1990; Krackhardt, 1999; Simmel, 1950), which reduces the extent to which available infor- mation can be deployed.

Thus far, no study has directly investigated the relationship between network closure and the char- acteristics of change initiatives in organizations. We propose that the informational and normative mechanisms that underlie the negative association between network cohesion and the generation of new ideas imply that organizational actors embed- ded in networks rich in structural holes are more likely to initiate changes that diverge from the in- stitutional status quo. Bridging structural holes ex- poses change agents to novel information that might suggest opportunities for change not evident to others, and it reduces normative constraints on how agents can use information to initiate changes that do not conform to prevailing institutional pressures.

Hypothesis 1. The richer in structural holes a change agent’s network, the more likely the agent is to initiate a change that diverges from the institutional status quo.

With respect to the probability that a change initiative will actually modify organizational func- tioning, few studies have explored how the degree of closure in change agents’ networks affects the adoption of organizational changes. This dearth of empirical evidence notwithstanding, Burt (2005: 86 – 87) suggested several ways in which brokerage opportunities provided by structural holes in an actor’s network may aid adaptive implementation, which he defined as the ability to carry out projects that take advantage of opportunities—as distinct from the ability to detect opportunities. Structural holes may equip a potential broker with a broad base of referrals and knowledge of how to pitch a project so as to appeal to different constituencies, as well as the ability to anticipate problems and adapt the project to changing circumstances (Burt, 1992).

These potential advantages suggest that struc- tural holes may aid change initiation differently from how they aid change adoption. In change ini- tiation, the information and control benefits of structural holes give a change agent greater expo- sure to opportunities for change, and creative free- dom from taken-for-granted institutional norms. These are, therefore, mainly incoming benefits that flow in the direction of the change agent. By con- trast, in change adoption, the information and con- trol benefits of structural holes are primarily out- going, in that they are directed to the organizational constituencies the change agent is aiming to per- suade. These benefits can be characterized as struc- tural reach and tailoring. Reach concerns a change agent’s social contact with the constituencies that a change project would affect, information about the needs and wants of these constituencies, and infor- mation about how best to communicate how the project will benefit them. Tailoring refers to a change agent’s control over when and how to use available information to persuade diverse audi- ences to mobilize their resources in support of a change project. Being the only connection among otherwise disconnected others, brokers can tailor their use of information and their image in accor- dance with each network contact’s preferences and requirements. Brokers can do this with minimal risk that potential inconsistencies in the presenta- tion of the change will become apparent (Padgett & Ansell, 1993) and possibly delegitimize them.

The argument that structural holes may facilitate change adoption stands in contrast to the argument that network cohesion enhances the adoption of innovation (Fleming et al., 2007; Obstfeld, 2005). Proponents of network cohesion maintain that peo- ple and resources are more readily mobilized in a cohesive network because multiple connections

2012 383Battilana and Casciaro

among members facilitate the sharing of knowledge and meanings and generate normative pressures for collaboration (Coleman, 1988; Gargiulo, Ertug, & Galunic, 2009; Granovetter, 1985; Tortoriello & Krackhardt, 2010). Supporting evidence is pro- vided by Obstfeld (2005), who found cohesive net- work positions to be positively correlated with in- volvement in successful product development, and by Fleming and colleagues (2007), who found col- laborative brokerage to aid in the generation of innovative ideas but maintained that it is network cohesion that facilitates the ideas’ diffusion and use by others.

These seemingly discrepant results are resolved when organizational change is recognized to be a political process that unfolds over time and takes on various forms. The form taken by a change ini- tiative is contingent on the extent to which it di- verges from the institutional status quo. Obstfeld described innovation as

an active political process at the microsocial level. . . . To be successful, the tertius needs to iden- tify the parties to be joined and establish a basis on which each alter would participate in the joining effort. The logic for joining might be presented to both parties simultaneously or might involve ap- peals tailored to each alter before the introduction or on an ongoing basis as the project unfolds. (2005: 188)

Change implementation, according to this ac- count, involves decisions concerning not only which network contacts should be involved, but also the timing and sequencing of appeals directed to different constituencies. A network rich in struc- tural holes affords change agents more freedom in deciding when and how to approach these constit- uencies and facilitate connections among them.

Building on this argument, we predict that in the domain of organizational change the respective ad- vantages of cohesion and structural holes are strictly contingent on whether a change diverges from the institutional status quo, thereby disrupt- ing extant organizational equilibria and creating the potential for significant opposition. Such diver- gent changes are likely to engender greater resis- tance from organization members, who are in turn likely to attempt building coalitions with organiza- tional constituencies to mobilize them against the change initiative. In this case, a high level of net- work cohesion among a change agent’s contacts makes it easy for them to mobilize and form a coalition against the change. By contrast, a network rich in structural holes affords change agents flex- ibility in tailoring arguments to different constitu- encies and deciding when to connect to them, whether separately or jointly, simultaneously or

over time. Less divergent change, because it is less likely to elicit resistance and related attempts at coalition building, renders the tactical flexibility afforded by structural holes unnecessary. Under these circumstances, the advantages of the cooper- ative norms fostered in a cohesive network are more desirable for the change agent. Consequently, we do not posit a main effect for network closure on change implementation, but only predict an inter- action effect, the direction of which depends on a change’s degree of divergence. Figure 1 graphically summarizes the predicted moderation pattern.

Hypothesis 2. The more a change diverges from the institutional status quo, the more closure in a change agent’s network of contacts dimin- ishes the likelihood of change adoption.

METHODS

Site

We tested our model using quantitative and qual- itative data on 68 change initiatives undertaken in the United Kingdom’s National Health Service, a government-funded health care system consisting of more than 600 organizations that fall into three broad categories: administrative units, primary care service providers, and secondary care service pro- viders. In 2004, when the present study was con- ducted, the NHS had a budget of more than £60 bil- lion and employed more than one million people, including health care professionals and managers specializing in the delivery of guaranteed universal health care free at the point of service.

FIGURE 1 Predicted Interactive Effects of Network Closure and Divergence from Institutional Status Quo on

Change Adoption

+

+

–

–

High

High

Low

Low

Network Closure

Change Divergence

384 AprilAcademy of Management Journal

The NHS, being highly institutionalized, was a particularly appropriate context in which to test our hypotheses. Like other health care systems throughout the Western world (e.g., Kitchener, 2002; Scott, Ruef, Mendel, & Caronna, 2000), the NHS is organized according to the model of medi- cal professionalism (Giaimo, 2002), which pre- scribes specific role divisions among professionals and organizations.1 The model of professional groups’ role division is predicated on physicians’ dominance over all other categories of health care professionals. Physicians are the key decision mak- ers, controlling not only the delivery of services, but also, in collaboration with successive govern- ments, the organization of the NHS (for a review, see Harrison, Hunter, Marnoch, and Pollitt [1992]). The model of role division among organizations places hospitals at the heart of the health care sys- tem (Peckham, 2003). Often enjoying a monopoly position as providers of secondary care services in their health communities (Le Grand, 1999), hospi- tals ultimately receive the most resources. The em- phasis on treating acute episodes of disease in a hospital over providing follow-up and preventive care in home or community settings under the re- sponsibility of primary care organizations is char- acteristic of an acute episodic health system.

In 1997, under the leadership of the Labour Gov- ernment, the NHS embarked on a ten-year modern- ization effort aimed at improving the quality, reli- ability, effectiveness, and value of its health care services (Department of Health, 1999). The initia- tive was intended to imbue the NHS with a new model for organizing that challenged the institu- tionalized model of medical professionalism. De- spite the attempt to shift from an acute episodic health care system to one focused on providing con- tinuing care by integrating services and increasing cooperation among professional groups, at the time of the study, a distinct dominance order persisted across NHS organizations, with physicians (Ferlie, Fitzger- ald, Wood, & Hawkins, 2005; Harrison et al., 1992; Richter, West, Van Dick, & Dawson, 2006) and hos- pitals (Peckham, 2003) at the apex. This context— wherein the extant model of medical professionalism continued to define the institutional status quo in these organizations—afforded a unique opportunity to study organizational change in an entrenched

system in which enhancing the capacity for inno- vation and adaptation had potentially vast societal implications.

Sample

The focus of the study being on variability in divergence and adoption of organizational change initiatives, the population germane to our model was that of self-appointed change agents, actors who voluntarily initiate planned organizational changes. Our sample is comprised of 68 clinical managers (i.e., actors with both clinical and mana- gerial responsibilities) responsible for initiating and attempting to implement change initiatives. All had worked in different organizations in the NHS and participated in the Clinical Strategists Programme, a two-week residential learning expe- rience conducted by a European business school. The first week focused on cultivating skills and awareness to improve participants’ effectiveness in their immediate spheres of influence and leader- ship ability within the clinical bureaucracies, the second week on developing participants’ strategic change capabilities at the levels of the organization and the community health system. Applicants were asked to provide a description of a change project they would begin to implement within their organ- ization upon completing the program. Project im- plementation was a required part of the program, which was open to all clinical managers in the NHS and advertised both online and in NHS brochures. There was no mention of divergent organizational change in either the title of the executive program or its presentation. Participation was voluntary. All 95 applicants were selected and chose to attend and complete the program.

The final sample of 68 observations, which cor- responds to 68 change projects, reflects the omis- sion of 27 program participants who did not re- spond to a social network survey administered in the first week of the program. Participants ranged in age from 35 to 56 years (average age, 44). All had clinical backgrounds as well as managerial respon- sibilities. Levels of responsibility varied from mid- to top-level management. The participants also rep- resented a variety of NHS organizations (54 percent primary care organizations, 26 percent hospitals or other secondary care organizations, and 19 percent administrative units) and professions (25 percent physicians and 75 percent nurses and allied health professionals). To control for potential nonre- sponse bias, we compared the full sample for which descriptive data were available with the fi- nal sample. Unpaired t-tests showed no statistically

1 This characterization of the NHS’s dominant tem- plate for organizing is based on a comprehensive review of NHS archival data and the literature on the NHS, as well as on 46 semistructured interviews with NHS pro- fessionals and 3 interviews with academic experts on the NHS analyzed with the methodology developed by Scott and colleagues (2000).

2012 385Battilana and Casciaro

significant differences for individual characteris- tics recorded in both samples.

Procedures and Data

Data on the demographic characteristics, formal positions, professional trajectories, and social net- works of the change agents, together with detailed information on the proposed changes, were col- lected over a period of 12 months. The demo- graphic and professional trajectories data were ob- tained from participants’ curricula vitae, and data on their formal positions were gathered from the NHS’s human resource records. Data on social net- works were collected during the first week of the executive program, during which participants com- pleted an extensive survey detailing their social network ties both in their organizations and in the NHS more broadly.

Participants were assured that data on the con- tent of the change projects, collected at different points during their design and implementation, would remain confidential. They submitted de- scriptions of their intended change projects upon applying to the program and were asked to write a refined project description three months after im- plementation. The two descriptions were very sim- ilar; the latter were generally an expansion of the former. One-on-one (10 –15 minute) telephone in- terviews conducted with the participants and members of their organizations four months after implementation of the change projects enabled us to ascertain whether they had been implemented— all had been—and whether the changes being im- plemented corresponded to those described in the project descriptions, which all did.

During two additional (20 – 40 minute) telephone interviews conducted six and nine months after project implementation, participants were asked to (1) describe the main actions taken in relation to implementing their changes, (2) identify the main obstacles (if any) to implementation, (3) assess their progress, and (4) describe their next steps in imple- menting the changes. We took extensive notes dur- ing the interviews, which were not recorded for reasons of confidentiality. The change agents also gave us access to all organizational documents and NHS official records related to the change initia- tives generated during the first year of implemen- tation. We created longitudinal case studies of each of the 68 change initiatives by aggregating the data collected throughout the year from change agents and other organization members and relevant or- ganizational and NHS documents.

After 12 months of implementation, we con- ducted another telephone survey to collect infor-

mation about the outcomes of the change projects with an emphasis on the degree to which the changes had been adopted. We corroborated the information provided by each change agent by con- ducting telephone interviews with two informants who worked in the same organization. In most cases, one informant was directly involved in the change effort, and the other was either a peer or superior of the agent who knew about, but was not directly involved in, the change effort. These infor- mants’ assessments of the adoption of the change projects were, again for reasons of confidentiality, not recorded; as during the six- and nine-month interviews, we took extensive notes.

At the beginning of the study, we randomly se- lected eight change projects to be the subjects of in-depth case studies. Data on these projects were collected over a year via both telephone and in- person interviews. One year after implementation, at each of the eight organizations, we conducted between 12 and 20 interviews of 45 minutes to two hours in duration. On the basis of these interviews, all of which were transcribed, we wrote eight in- depth case studies about the selected change initia- tives. The qualitative data used to verify the con- sistency of change agents’ reports with the reports made by other organization members provided broad validation for the survey data.

Dependent and Independent Variables

Divergence from the institutional status quo. The institutional status quo for organizing within the NHS is defined by the model of medical …

1