Leadership in Public Sector Interorganizational Networks: A Synthesis of the Literature and Propositions Based on a Multiple Case Study

ABSTRACT

INTRODUCTION

Public organizations increasingly participate in networks in order to create public value (Crosby and Bryson 2010; Sullivan, Williams and Jeffares 2012). Networks are collections of autonomous organizations that collaborate in a joint effort toward a common purpose (Carboni, Saz-Carranza, Raab and Isett, 2019). Networks have become a widespread phenomenon, engaging public, private, and societal stakeholders in areas such as healthcare and crime prevention (Nowell, Hano and Yang 2019; Torfing, Krogh and Ejrnaes 2020). This is due to the ability of networks to pool critical resources across participating actors, enabling the achievement of collaborative advantage—creating (public) value that could not have been created by individual actors alone (Bianchi, Nasi and Rivenbark 2021; Huxham 1996).

The emergence of networks in the public sector generates new opportunities and challenges for leadership. Recent leadership studies have emphasized the context-dependent nature of leadership, demonstrating the relationship between context and the manifestation of different leadership behaviors (Schmidt and Groeneveld 2021; Stoker, Garretsen and Soudis 2019; Van der Hoek, Beerkens and Groeneveld 2021). Leadership is embedded in a social setting at a specific moment in time, and its effectiveness depends in part on the context in which it takes place (Shamir 1999; Shamir and Howell 2018). Taking the specific context of public sector networks into account, research indicates that networks are characterized by horizontal coordination and interaction between organizations, meaning that organizations operate on a theoretically equal footing (Klijn and Skelcher 2007; O’Toole Jr. 1997). In this context, formal incentives and regulatory tools to ensure the commitment of individual network members to the network’s objective appear to be absent (Klijn 2005). Hence, rather than leadership on a formal basis, network leadership would require other forms of enhancing commitment among network members, identifying relevant actors, ensuring the input of all stakeholders, and mobilizing support for a common network objective ( Agranoff and McGuire 2001; Bryson, Crosby and Stone 2006). In the current era of boundaryless organizations, leadership thus appears to play an increasingly vital role (Shamir 1999). This applies especially to the public sector context, where organizations are characterized by hierarchical internal relations and vertical (political) accountability (Powell 1990; Thompson 1991).

Despite the promising role of leadership in public sector networks, as yet no conceptualization of the functioning of leadership within this context has been developed. This lack is mainly due to the current disconnect between various branches of scholarly literature on collaborative governance, network management, and leadership. First, collaborative governance focuses on the (macro-level) design of the collaborative process to explain why and how organizations collaborate toward collective objectives (Emerson, Nabatchi, and Balogh 2012). Yet, this branch of literature does not account for the various behaviors network members use to influence each other to pursue a collective objective. Second, network management literature does describe individual behaviors that foster collaboration toward collective goals (Klijn, Steijn and Edelenbosch 2010). However, its main caveat is that this branch focuses mainly on the efforts of single network managers, and does not take into account the potential role of shared or distributed leadership by various network members. Lastly, leadership literature does offer conceptual tools to study leadership as a process in which various individuals can participate (Gronn 2002; Pearce 2004; Ulhøi and Müller 2014). Aside from a single case study conducted by Kramer et al. (2019), however, these concepts have yet to be applied to the collaborative, interorganizational context of public sector networks.

Consequently, this study aims to answer the following research question: How can leadership in pursuit of collective objectives in public sector interorganizational networks be conceptualized? Through a synthesis of the literature and an empirical investigation of three cases, this study develops a conceptual framework that aims to situate and compare leadership in networks. It encompasses leadership behaviors, the distribution of these behaviors across network members and the various directions in which network members exhibit these behaviors. On the basis of this conceptual model, this study concludes with four theoretical propositions of leadership in networks.

The contribution of this article is threefold. First, the article contributes to the literature on leadership by conceptualizing leadership in a particular context, that is networks in the public sector. Though this context has been given attention in single case studies (see, for instance, Kramer et al. 2019), the scientific contribution of this article consists in its focus on three different networked contexts, enabling the identification of leadership configurations in networks with different characteristics. Second, this article adds to network management literature by acknowledging opportunities for shared or distributed forms of leadership. Lastly, we add to the literature on collaborative governance by examining individual behaviors used to enhance collaboration towards collective goals in public sector networks.

The next section reviews the literature on the role of leadership in public sector networks. Subsequently, based on a qualitative multiple case study in three interorganizational public sector networks in The Netherlands, we analyze the leadership behaviors of network members as well as the characteristics of the networks in which they operate. The article concludes with a synthesis of the conceptual framework, four theoretical propositions on the relationship between leadership and characteristics of network contexts and discusses possible avenues for future research.

LITERATURE REVIEW

The conceptual framework and its theoretical propositions developed in this study draw on three branches of literature. First, collaborative governance literature is consulted to situate leadership in the context of public-sector interorganizational networks. Second, network management literature is used to highlight how networks are steered towards their goals. Lastly, leadership literature is consulted to understand how leadership could be distributed across multiple individuals, and discusses a taxonomy of leadership behaviors that is used to identify leadership behaviors by various members of the networks included in this study.

Collaborative governance literature covers a wide range of scientific contributions that explain the emergence and antecedents of public and cross-sector collaboration (Emerson, Nabatchi and Balogh 2012). Previous studies specify components of a fruitful collaborative process, such as trust (Ansell and Gash 2008), equality, face-to-face dialogue (Ansell and Gash 2008), commitment, and a shared understanding of collective goals (Silvia 2011). Several authors in this field emphasize the role of leadership as a contributing factor in these collaborative processes (see Ansell and Gash 2008; Bryson, Crosby and Stone 2006; Emerson, Nabatchi and Balogh 2012). Morse (2010), for instance, identifies the role of individual boundary spanners who mobilize relevant actors to forge integrative partnerships. Furthermore, Crosby, ‘t Hart, and Torfing (2017) emphasize the role of sponsors, champions, catalysts, and implementers in enhancing collaboration. A caveat of existing research is that these studies tend to reduce leadership to the presence of specific leadership roles or functions in networks. These studies, however, do not explain how these functions are executed in terms of actual leadership behaviors and how these behaviors could be distributed across various network members and shared to achieve a common goal.

In response, network management literature does demonstrate the importance of behaviors, activities, and strategies of network managers in enhancing network effectiveness (Klijn, Steijn and Edelenbosch 2010; McGuire and Silvia 2009). For example, Silvia and McGuire (2010), replicated in a more recent publication by Cepiku and Mastrodascio (2021), compared task- and relations-oriented leadership behaviors in networks with leadership behaviors exhibited by the same managers in their home organizations. These studies demonstrate the distinct nature of leadership in networks, as opposed to single organizations. Particularly, relations-oriented leadership behaviors are more common in networked contexts compared to single-agency structures. Consequently, these studies form a relevant starting point for this study on leadership in networks, as they indicate the relevance of studying leadership from a behavioral point-of-view in the context of networks. One caveat of these studies is their predominant focus on single network managers as opposed to all network members. As a consequence, these studies do not account for the potential of distributed or shared forms of leadership. This study responds to this gap in the literature by studying leadership behaviors of all network members. In so doing, this study extends its scope beyond formal network managers.

A second contribution of network management literature which will be incorporated in this study, is its focus on the structural design of public sector networks and its contingencies with network effectiveness. In terms of structural design, previous studies indicate how the functioning of collaborative processes in networks depends on the congruence between network properties such as form of governance (Provan and Kenis 2008), mandate (Provan and Lemaire 2012), membership diversity or heterogeneity (Baraldi and Strömsten 2009), and function (Milward and Provan 2006). In this study, the structural design of networks will be taken into account as potential contingencies vis-à-vis leadership exhibited in the networks under study.

Leadership research offers three valuable contributions to the above-mentioned branches for building a conceptualization of leadership in networks: first, its attention to the context in which leadership is exhibited; second, its acknowledgement of shared and distributed forms of leadership; and third, its focus on leadership behaviors. Several leadership scholars have emphasized the importance of context in studying leadership. For instance, complexity leadership theory views leadership as a socially constructed, interactive dynamic with emerging outcomes in a complex adaptive system (CAS) characterized by interdependence between multiple actors (Marion and Uhl-Bien 2001). This approach is particularly relevant within the context of inter-organizational networks due to the inherent complexity and interdependence of networked contexts (Koliba and Koppenjan 2023). A second advantage of this approach is that it distinguishes leaders from leadership. Whereas the former focuses on the actions of individuals, the latter examines the process in which multiple actors participate. In so doing, the CAS approach shifts its focus from specific managerial positions as seen in network management literature, to a wider perspective on who participates in the process of leadership (Uhl-Bien, Marion and McKelvey 2007). Previous studies of leadership in collaborative or team contexts have identified forms of leadership that extend beyond individual leaders (Bergman et al. 2012; Carson, Tesluk and Marrone 2007; Gronn 2002; Ospina and Saz-Carranza 2010). Shared leadership involves continuous, mutual influence between multiple individuals, meaning that leadership can be exercised by multiple appointed or emergent leaders simultaneously (Pearce 2004; Pearce and Conger 2003). Distributed leadership involves the delegation of leadership tasks from one leader to other individuals (Ulhøi and Müller 2014). As network management literature has demonstrated, hierarchical relationships are less prevalent in networks. In this context, the acknowledgment of “leadership beyond formal leaders”—as described through the concepts of shared and distributed leadership—is useful in the study of leadership in networks. Therefore, this study selects the network as its unit of analysis and investigates whether and how leadership is shown by all network members, rather than focusing on (predetermined) individual network managers. In so doing, this study responds to earlier calls for the study of leadership by focusing on systems of relationships and differentiating the leader from leadership.

This study uses the behavioral taxonomy developed by Yukl (2012) as its starting point. Yukl (2006, p. 8) defines leadership as “the process if influencing others to understand and agree about what needs to be done and how to do it, and the process of facilitating individual and collective efforts to accomplish shared objectives.” Treating leadership in a processual manner, in which (multiple) individuals demonstrate leadership behavior corresponds with other publications by Ospina et al. (2020) and Uhl-Bien (2006), which emphasize the need to distinguish leaders from leadership. Through this definition, Yukl identifies four distinct categories of behaviors used in the process of leadership: task-oriented, relations-oriented, change-oriented, and externally oriented behaviors aimed at facilitating goal attainment.

Task-oriented behaviors are targeted at ensuring the proper and efficient allocation of resources to accomplish specified objectives. This includes behaviors such as clarifying and planning tasks, monitoring progress, and problem-solving (Yukl 2012). Relations-oriented behaviors involve actions that contribute to interpersonal relationships, commitment to the specified mission, and the enhancement of skills. Examples include supporting by showing positive regard, developing skills and confidence, the use of praise, and empowerment by granting autonomy (Yukl 2012). Change-oriented behaviors encompass actions taken to ensure development, increase collective learning, and enhance the resilience of the organization. This is achieved by advocating and envisioning change, encouraging innovation, and facilitating collective learning (Yukl 2012). External leadership behaviors include activities aimed at drawing knowledge, social networks, and other valuable assets from the external environment. Examples include networking, external monitoring (analyzing relevant changes in the external environment), and representing the organization to lobby for resources (Yukl 2012).

The advantage of focusing on leadership as a process in which actors participate through a repertoire of specific behaviors is that it allows the researcher to not only differentiate between types of behaviors, but also the distribution of these behaviors across various individuals in a specific context (Van der Hoek, Groeneveld & Beerkens 2021). In so doing, this approach adds more depth to previous network leadership studies which tend to focus either on leadership exercised by specific individuals in networks (see Cepiku and Mastrodascio 2021 and; Silvia and McGuire 2010) or use more abstract terms to define leadership in networks as roles or functions (e.g. Crosby, ‘t Hart, and Torfing).

Consequently, the main aim of this study is to explicate which leadership behaviors members of interorganizational networks use to motivate each other to work towards collective (network) goals, recognizing the possibility of shared or distributed leadership rather than focusing on individual network managers alone. Second, this study develops propositions on the basis of the empirical findings, stipulating potential relationships between leadership and particular characteristics (governance, mandate, diversity, and functions) of networks. The theoretical value of this approach is that it confirms whether a behavioral taxonomy is useful in networked contexts, and that it demonstrates whether leadership indeed extends beyond network managers.

The practical value of this approach is that recommendations can be made to members of networks on what type(s) of leadership are prevalent in specific network contexts.

METHODOLOGY

This study employs a qualitative multiple case study, comprised of three collaborative public sector networks located in the Netherlands.

Case Selection

The cases were purposively selected on the basis of capturing various collaborative contexts and data availability. The research project took place within a research program involving several public organizations, including the Municipality of Leiden, the Custodial Institutions Agency, and the National Police of the Netherlands. The authors first contacted these partners individually to verify whether their organization participates in networks. Based on their suggestions, the authors selected one network of each of these organizations that met the following selection criteria: first, the network should still be active; and second, network members should meet frequently (at least once per two months) over a period of more than one year. Both criteria were used to make sure that respondents can recall leadership behaviors.

As highlighted by Lemaire, Mannak, Ospina, and Groenleer (2019) providing detail about the context of networks is considered good research practice when conducting empirical research. In this study, networks were selected on the basis of variety in terms of the networks’ structure, legal basis, function, and diversity. Previous research emphasizes that these features influence both the effectiveness of networks and collaborative processes that take place within them (Nowell and Kenis 2019; Provan and Kenis 2008; Segato and Raab 2018; Milward and Provan 2006; Corsaro, Cantù and Tunisini 2012). Therefore, the cases selected differ on the basis of the aforementioned characteristics. All three cases involve interorganizational collaboration between public organizations with the aim of identifying and tackling complex societal challenges, which require the exchange of resources. The cases are presented in Table 1.

Identifying Collective Environmental Challenges: Spatial Planning Network

This is an informal collaboration between nine municipalities and the regional Water Authority operating in one province in The Netherlands. It was initiated by the largest municipality in the region. The main purpose of this network is to exchange information in order to collectively develop a regional vision of spatial planning. The network targets spatial planning challenges that span across municipal boundaries. Challenges include the municipalities’ approach toward climate change, energy supply, water management, mobility, and housing. The dataset for this case consists of 19 interviews, representing 9 out of 10 organizations.

Customized Care for Juvenile Offenders: Juvenile Detention Network

This is a formally mandated collaboration established to coordinate the accommodation of juvenile prisoners with a low-risk profile. It aims to reduce recidivism rates among juvenile offenders of petty crimes by integrating penal measures (detention) with youth care and support. The collaboration is initiated and funded by the Ministry of Justice and Security and comprises the Custodial Institutions Service, youth care providers, child care and protection board, public prosecutor’s office, and several municipalities in the vicinity of the juvenile facility. The network was initiated by the Ministry as a pilot in 2016, and formally established in 2020. The new low-security, small-scale prison facility was established to replace an existing, large-scale facility with high-level security. In 2021, the facility officially came into use when the first detainee arrived. The dataset for this case consists of nine interviews, representing eight organizations. In addition to these eight organizations, the whole network consists of multiple municipalities. The dataset contains one municipality.

Enhancing Identification and Treatment of Mentally Distressed Citizens: Mental Health and Public Order Network

This is a formally mandated network that coordinates collaboration between organizations operating in the (mental) healthcare domain and the security domain. The societal challenge involves the provision of high-quality (mental) health care for people at risk of becoming mentally ill and causing public order disturbances. The network facilitates primary, secondary, and tertiary prevention and treatment to this group. Participants include municipalities in the area, National Police, the largest mental healthcare provider in the region, a healthcare insurance company, the public prosecutor’s office, and municipal health services. The network was initially established in 2020 as a Network Administrative Organization led by an external consultancy firm. As of 2021, formal network leadership and management is in the hands of the two largest municipalities in the region. This network is part of a larger regional initiative that encourages collaboration between public organizations in the domains of security and healthcare. The dataset for this case consists of 11 interviews, representing nine out of nine organizations.

Although we acknowledge that networks often consist of various layers (political, administrative, street-level/frontline), this study focuses on the administrative layer only, on networks comprising civil servants.

Data Collection

For each of the three cases, interviews were conducted with network members (total N = 39). Respondents were all network participants, but their formal organizational positions could vary. The data includes interviews with managers, policy advisors, senior front-line professionals, and external consultants. For each of the networks, respondents participated in the same “tier” of the network—policy-making. The majority of interviews were conducted on a one-to-one basis. A few interviews were conducted with two network members at the same time. In those instances, the network members represented the same organization in the network. The interviews took place between November 2020 and March 2021. As the coronavirus disease 2019 (COVID-19) pandemic made it impossible to conduct these interviews face-to-face, all interviews were conducted through Microsoft Teams.

The interview protocol was developed on the basis of the taxonomy of leadership behaviors by Yukl (2012). The protocol started with broader questions, such as: “Who demonstrate(s) leadership in this network?” “What do(es) person X do that you identify as leadership?” Follow-up questions were meant to retrieve respondents’ perceptions of task-, relations, change- and externally oriented behaviors in the context of the network. An example of a follow-up question on task-oriented leadership (task division) is: “How did person X divide tasks?” An example of a follow-up question on change-oriented leadership (envisioning change) is: “What did person X do to share their vision of the societal problem and/or the role of the network in tackling it?” Open-ended questions encouraged respondents to also elaborate on types of leadership behavior beyond Yukl’s taxonomy and enabled the researchers to explore various recipients of leadership behaviors (besides just network members). Additionally, document analysis and direct observations of network meetings (if possible) were employed to complement the findings from the interviews.

Coding Process

Interviews were coded using Atlas.ti. The coding process involved a combination of inductive and deductive coding techniques. In the first phase, the behaviors stipulated in Yukl’s taxonomy were used as predefined sets of codes. Behaviors that could not be categorized according to Yukl’s taxonomy were coded inductively as “Leadership:Taxonomy:5:Other:[Name].” In this initial phase, inductive coding was also applied to respondents’ statements about the general nature of leadership in the network, such as: “Who demonstrates leadership?” and: “What does leadership in networks mean to the respondent?” In the second phase of the coding process, segments that were initially labeled as “other” leadership behaviors were reviewed to ascertain whether they indeed formed distinct classes of behavior, or whether they could be included as “additional” types of behavior within the existing coding categories (task-oriented, relations-oriented, change-oriented, and externally oriented leadership). For instance, the behavior “organizing cross-boundary-experiences” was initially labeled as “Leadership:Taxonomy:5:Other,” but was later interpreted as a relations-oriented behavior aimed at enhancing a common identity—and hence, enhancing relations—among network participants. Similarly, “sharing information” was initially labeled as “Leadership:Taxonomy:5:Other,” but later recoded as a task-oriented behavior, as sharing information allowed network members to obtain clarity about the tasks at hand.

A within-case analysis was carried out to explore and understand the types of leadership behavior exercised in each network, after which a cross-case analysis was performed to analyze the types of leadership behavior in distinct network contexts. In this process, the authors also focused on the distribution of leadership (how many participants within the network demonstrated leadership?) and the particular direction in which the behaviors were shown (towards members’ home organizations, among each other, or towards the network’s external environment). This analysis resulted in a conceptual framework and four theoretical propositions regarding leadership in public sector networks, which is discussed in the next section.

A CONCEPTUAL FRAMEWORK OF LEADERSHIP IN PUBLIC SECTOR NETWORKS

This section substantiates our conceptual framework that builds on the four types of leadership behavior presented by Yukl (2012): task-oriented, relations-oriented, change-oriented, and external leadership. It specifies manifestations of these four types of leadership behavior in the context of public sector networks and, in addition, identifies types of leadership which were observed but do not feature in Yukl’s original model. We depart from the original model in two ways, though. First, our framework has a network-centered—rather than organization-centered—orientation. Whereas Yukl’s taxonomy describes ‘externally oriented leadership behaviors’ as behaviors directed towards the external environment of the organization, externally oriented behaviors should be interpreted from the viewpoint of the network. Second, our conceptualization focuses on behaviors aimed at influencing others towards shared network goals, rather than organization-internal goals. Therefore, behaviors geared solely towards the goals of individual organizations are not conceived of as leadership behaviors in public-sector networks.

The data reveal that, besides the types of leadership manifested in the networks, there are differences in the types of recipients to which leadership is directed, as well as the presence of concentrated or shared forms of leadership. These findings culminate in a theoretical synthesis comprising three elements: types of leadership behavior, directions of leadership behavior, and distribution of leadership. For each of these dimensions, quotes are included that represent a consensus across interviewees per case (network). Following the analysis, this section concludes with four theoretical propositions regarding network characteristics as potential contingencies for leadership in networks.

Leadership Behaviors in Networks

Task-oriented behaviors are used to clarify collective public value objectives, and to assign tasks and responsibilities to network members. These behaviors are mostly exercised by network members with a formal leadership role, such as project leaders, or members of the network’s lead organization. Task-oriented leadership in networks aims to facilitate, rather than direct or dictate, the collaboration process.

Relations-oriented behaviors are aimed at building and maintaining network relationships. To achieve this, potential new members are identified and approached to ascertain their attitudes towards collaboration. When building network relationships, potential members are asked about societal challenges they are facing in order to see if they overlap with the other parties. In this process, the common interest—the public value objective—is conceptualized.

To foster network relations, members develop a common identity or set of common values that is expressed through language and visual aids such as logos. This development of a shared identity enables the network to collectively assess the outputs of the network. For example, the Spatial Planning Network would envision the identity of the region as a “beautiful” space. Then, when tackling environmental challenges, such as the transition towards sustainable energy sources, the Network would assess potential policy options (e.g., wind turbines, solar panels) by their impact on the value of “beauty”: for example, how do wind turbines and solar panels impact the beauty of the region?

The networks also displayed other types of relations-oriented leadership behavior that were not identified in previous work on leadership in single organizations. These behaviors were mainly aimed at mitigating barriers to effective collaboration. One such impediment to collaborative public value creation concerns conflicts that emerge between members due to differences in organizational logic and language. For instance, organizations may operate based on distinct underlying principles, values, structures, and methods for organizing their activities and making decisions. As a consequence, tensions may arise between network members due to different ways of understanding and approaching their work, using different communication styles or operating on the basis of different values. This is in line with previous research indicating that organizations may employ different approaches to networking and network management, resulting in tensions between network partners (Herranz Jr. 2008).

To mitigate this, the following behaviors were used: enabling cross-boundary experiences and the use of emotions.

Cross-boundary experiences (Feldman et al. 2006) are used to build bridges between network members. Members would visit each other’s offices to obtain a better understanding of each other’s organizational logic, and show an interest in each other by asking questions. This became especially relevant in understanding the viewpoints of other members when a decision had to be made. Having a better grasp of each others’ professional environment helped members understand why their counterparts might hold different or opposing positions in decision-making processes.

In relation to using emotions, network members would express their joy or discomfort regarding decisions made by other members. For instance, a network member would disclose their relief or disappointment with another members’ decision by expressing how this decision impacted their ability to do their work.

Network members observed two effects of this type of behavior: showing emotions would encourage other members to be more open, and it would also lead to increased empathy and the desire to help one another. Both effects were regarded as positive for the collaborative process.

Change-oriented behaviors were observed in the networks; that is, behaviors that advocate for change by raising awareness of the necessity for collaboration and the creation of opportunities for collective learning. Network members would stimulate awareness of other members about the need for collaboration through the use of statistics, pictures, and maps. In doing so, the public value objective was used to emphasize the importance of collaboration. Network members would often point out how a lack of collaboration could affect the network’s target group. Particularly when discussions ended up in technicalities, members would use change-oriented behaviors to redirect the focus towards the targets of the network, such as vulnerable children or mental health patients. Furthermore, collective learning was stimulated through the use of peer review groups. In these groups, network members representing different organizations would be paired up to make decisions together.

In summary, Yukl’s taxonomy of task-, relations-, and change-oriented behaviors were all found in all three networks. Whereas task-oriented behavior was used to clarify and divide tasks and responsibilities, relationship-oriented behavior was used to prevent or solve tensions between network members. Lastly, change-oriented behaviors redirected network participants’ perspectives on the societal goal they aim to achieve together and helped participants find effective ways of attaining this goal.

Directions for Leadership Behavior

The analysis revealed that network members exercise leadership behaviors in three directions: towards their home organization (and its stakeholders), towards the network (and its stakeholders), and towards the external environment.

First, network members direct leadership toward the organizations they represent. For instance, network members exercise change-oriented behaviors to make their coworkers aware of the need for interorganizational collaboration and the potential changes this will bring to existing organizational procedures and processes. They also motivate and enthuse their colleagues who operate at other levels of the network. A network participant operating at the administrative layer of the network would, for instance, support colleagues collaborating with other network members on the front line. Network members also display behaviors such as “networking” and “representing” to obtain funding or other resources for the benefit of the network. Previous research indicates that in turn, partners in positions of authority often play an important role as “champions” or “sponsors” of networks by offering political support (Crosby and Bryson 2010).

Second, leadership towards other network members is characterized by the need to balance the goals of each individual organization with the goals of the network as a whole. Respondents reported that they were given a mandate to set the agenda and to use network meetings to prioritize tasks vital to their own organization. At the same time, they were aware of the potential for conflict between the objectives of their own organization and the objectives of the network. To prevent and cope with potential conflicts, network members display relations-oriented leadership behaviors. Asking questions, expressing emotions, and discussing commonalities and differences between the organizations enabled network members to identify common values and objectives. At the same time, network members also set boundaries in order to make clear to the other members what they can and cannot expect them to contribute to the network. As well as horizontal relationships with members acting in the same “layer” of the network, there are also vertical relationships between network members and other layers of the network. Task-oriented leadership is required to inform other network layers (such as front-line workers or political decision-makers) of decisions taken.

Third, leadership towards the external environment is mostly displayed by members with a formal leadership position. These individuals represent the network in order to acquire funding and other resources that enable the network to pursue its public value objective. Network members with a formal leadership position would use their personal networks to invite experts to network meetings. In the case of the Spatial Planning Network, these external experts were crucial in providing support for the network’s endeavors.

In sum, the term “external” leadership in Yukl’s framework translates differently in the context of networks, where network members constantly interact with their home organization, fellow network members or the wider external environment. Members interact with their external environment to attract additional resources, use various leadership behaviors to motivate their colleagues to align with network practices and exhibit leadership towards fellow network members to balance organizational and network interests.

Leadership Distribution in Networks

Regarding leadership distribution, the networks showed dynamic patterns in which there was alternation between concentrated and distributed forms of leadership. At one end of the continuum, leadership behaviors may be concentrated in one or a few individuals. In the networks observed in this study, leadership concentration was most often found in individuals who possess legitimizing forces of authority, such as capacity, formal responsibility, and formal position. The first of these concerns an individual’s ability to exert leadership based on the capacity of the organization to support the network. This was observed in the Spatial Planning Network, in which members of the largest municipality exerted a high degree of task-oriented leadership due to the proportion of the network budget contributed by this organization. The second basis concerns the formal responsibility of the member’s organization with regard to the problem at hand. In the Mental Health and Public Order Network, the two largest municipalities exerted more leadership due to the formal obligation of municipalities to take care of the target group (citizens with mental health disorders). The third basis concerns a member’s formal position in a network. In the Juvenile Detention Network and the Mental Health and Public Order Network, external project leader demonstrated more leadership behaviors to steer the network towards its goals than other network members. Concluding, network members possessing (a form of) formal authority—capacity, formal responsibility, and formal position—demonstrated more leadership behaviors in cases of concentrated leadership.

At the other end of the continuum, leadership can be (fully) distributed. Distributed leadership takes place when leadership is exerted by multiple individuals within the network. Distributed leadership improves the allocation of information and expertise across members and enhances widespread support for the network. In the three cases studied, a high degree of technical expertise on a specific subject discussed in the network, and extensive experience with networks or had developed specific collaboration skills appeared to be driving forces of distributed leadership. For example, network members who showed a high degree of technical expertise on a particular subject would be more likely to show task-, relations-, change-, and externally-oriented behaviors when this subject was discussed in network meetings. Network members who had more experience participating in collaborative networks showed more leadership behaviors than those who did not have such experience.

Consistent with earlier studies (e.g. Barry 1991), the distribution of leadership appeared to be dynamic in all three cases. Leadership would often alternate between concentrated and distributed forms. For instance, in all three cases, participants reported more leadership concentration in the startup phase of the network—a phase in which it can be helpful to have a formal leader to gather and mobilize actors and resources. Contrastingly, leadership was more distributed when network participants negotiated a decision or provided input on the aims of the network. In the three cases studied, the (formal) network leader would delegate leadership tasks to other members by assigning them specific network dossiers. It was found that distributed leadership was sometimes used to stimulate shared leadership. Respondents mentioned that delegating leadership tasks to other members was done with the expectation that the particular member would also show more leadership later on in the network process.

In summary, this section demonstrates that leadership concentration and distribution can alternate. This study does indicate that—in the three cases—leadership concentration is associated with the capacity of actors, having a formal responsibility for the societal problem at hand, and/or formally appointing “network leaders.” This finding appears contradictory to the “horizontal nature” networks are associated with in academic literature. However, this section also demonstrates that leadership is also delegated or shared in networked contexts. However, due to the limited scope of the study, it is not possible to offer further explanation of how shared leadership emerges from distributed leadership.

FOUR THEORETICAL PROPOSITIONS ON THE RELATIONSHIP BETWEEN LEADERSHIP AND NETWORK CONTEXT

On the basis of the cross-case analysis, the analysis yields several observations regarding leadership in networks with specific properties. These observations are formulated as propositions for future research. In particular, we identified different manifestations of leadership across the four network characteristics investigated:

First, the type of leadership observed in the networks differed along the networks’ governance forms. NAOs involved formal sources of network leadership in the form of project leaders. These individuals carried out task-oriented and externally oriented-leadership and did so in a directive manner. In the lead organization network, formal sources of leadership were also important: leadership was carried out by representatives of the largest municipality. These individuals displayed task-oriented and externally-oriented leadership behavior, but did so in a more facilitative manner. This difference could be explained by the legitimacy of the project leader role in the NAO; that is, being formally appointed as project leader.

Differences were also observed between the networks with a legal mandate and the network without a legal mandate. The network without a legal mandate demonstrated relations-oriented behaviors. These behaviors were mainly aimed at identifying and convincing potential network members, and developing relations with these actors. In legally mandated networks, these behaviors were observed to a lesser extent. Although future research should indicate whether a causation exists between legal basis and leadership, an explanation could be that non-mandated or serendipitous networks depend more heavily on the commitment of their members, as membership is voluntary.

Third, disparities in leadership were found between networks aimed at coordinating policy and the network aimed at exchanging knowledge. Externally oriented behaviors such as networking, external monitoring, and representing were found more often in the latter than the former. One explanation could be that knowledge exchange networks rely more heavily on gathering external input (knowledge) than networks with other functions. On the other hand, the networks involved in policy coordination showed more behaviors aimed at preventing conflict and mitigating the impact of conflicts. Change-oriented behaviors such as emphasizing the public value objective (“vision”) and creating policy intervention groups (“collective learning”) were also found more often in these networks. This may be attributed to the fact that policy coordination and implementation involves more conflicts, as members work more closely together and task-interdependence is higher.

Lastly, differences were found in terms of the diversity of the networks included. Networks with a moderate to high degree of goal heterogeneity (Baraldi and Strömsten 2009), knowledge/capability heterogeneity (Frenken 2000), and cultural heterogeneity (Chen, Tsou and Ching 2011) showed more relations-oriented and change-oriented behaviors aimed at mitigating conflict. With a high degree of diversity between members, there may be a greater need for “boundary experiences” or activities to bridge differences (Feldman et al. 2006). Networks with a high or medium degree of diversity also showed more distributed leadership on the basis of technical expertise. The Mental Health and Public Order Network showed a high degree of diversity, comprising agencies ranging from the public health domain to the public order domain. In this network, each member would assume the leader role when a subject requiring their specific expertise was discussed. This was exhibited through prioritizing items on the agenda and taking on responsibilities regarding a policy domain (task-oriented leadership), and stepping forward as a spokesperson on a particular subject (externally oriented leadership). More experienced members would often be the first to react to policy proposals and be more vocal in drawing attention to problems, and would often explicate the collective public value objective during network meetings.

Discussion

Leadership as a driving force for collaborative public value in networks is attracting increasing scholarly attention (Sørensen, Bryson and Crosby 2021; Torfing and Ansell 2017). Despite the popularity of the subject, the study of leadership in collaborative contexts is hampered by a lack of conceptual clarity. This study has sought to address this issue by synthesizing the literature, applying it to empirical cases and providing four theoretical propositions about the relationship between network context and leadership. The article thereby demonstrates the conceptual value of a behavioral approach to studying leadership in networks and how this approach can be used to understand both the types of leadership behaviors, their directions and the distribution of leadership in these contexts. This study thereby aids future research seeking to empirically compare leadership behavior across collaborative contexts and assess the contribution of such behavior to collaborative outcomes.

In response to the research question ‘How can leadership in pursuit of collective objectives in public sector interorganizational networks be conceptualized?,’ the findings of this study are synthesized into a conceptual framework and four theoretical propositions regarding leadership in distinct network contexts. The framework adds to existing literature in the following ways.

First, this theoretical synthesis contributes to leadership literature by providing a starting point for understanding leadership in the general context of interorganizational networks in the public sector. In so doing, the framework contributes to leadership theory as it demonstrates that leadership in networks has different manifestations, depending on who exhibits leadership (formal leaders, or both formal and informal leaders), the directions of leadership behavior (the member’s home organization, fellow network members or the external environment) and the specific network context (structure, mandate, function, diversity) in which leadership takes place. Specifically, this study indicates leadership differences between networks with specific properties, prompting theoretical propositions regarding the role of these properties—network structure, legal basis, network function, and level of diversity—in shaping leadership processes in networks. As such, this study confirms earlier calls for the acknowledgement of context when studying leadership (Schmidt and Groeneveld 2021; Stoker, Garretsen and Soudis 2019; Van der Hoek, Beerkens and Groeneveld 2021). These observations provide opportunities for additional research on contingency factors influencing leadership configurations in networks.

Second, the synthesis contributes to network management theory, as this study indicates that multiple network members—as opposed to individual network managers—can exercise behaviors to bring the network towards its goals. The acknowledgement of more concentrated or more distributed forms of leadership contributes to a better understanding of leadership dynamics in the context of networks. Nevertheless, this study also demonstrates that formal sources of leadership can still be present in networks and that these forces may shape leadership processes in networks.

Third, this theoretical synthesis contributes to collaborative governance literature by means of its behavioral lens. The framework allows researchers to understand public sector collaboration as a process in which network members exhibit leadership behaviors to advance networks toward their objectives. The study identifies task-, relations-, change- and externally oriented behaviors network members use to achieve collective goals.

This study also has a few limitations. Its scope is limited to three cases, all of which are based in the Netherlands. Further research is required to study the application of the propositions in other networked contexts. Second, although the design of this study serves the research question, in future research it may be fruitful to make use of other approaches, such as quantitative, experimental, or ethnographic research designs. This echoes recent concerns in leadership studies about the extent to which interviews and surveys of participants in the leadership process actually do measure leadership behaviors (Banks, Woznyj and Mansfield 2021).

CONCLUSION AND IMPLICATIONS

In interorganizational public sector networks, leadership fulfills an important role as a means to coordinate task allocation and establish collaborative processes. This leadership is shaped by the contributions of multiple network members and goes beyond formal leadership positions; moreover, it manifests itself differently depending on the characteristics of the network.

The findings of this study offer the following practical implications. First, leadership behaviors enable network members to conceptualize, interpret, and pursue a common public value objective. Identifying and explicating these behaviors is the first step towards being able to develop these kinds of behaviors in professionals who participate in such networks. Second, as this study indicates, there is a continuum between concentrated and distributed leadership in networks; therefore, strategies to concentrate or distribute leadership can be developed to fit the needs of the network at different points in time. Third, this study has found that every network member has the potential to display leadership, despite not having a formal leadership position. In practice, this finding can empower network members to show more (relations-oriented and change-oriented) behaviors to encourage other members to collaborate in pursuit of a common objective. For the academic study of leadership in networks, this finding implies that a (predetermined) empirical focus on specific “network leaders” does not fully cover leadership at the network level. Rather, this study shows the importance of focusing on the aggregate leadership behaviors of all network members. In so doing, the study reaffirms earlier calls by Ospina and Foldy (2015) Uhl-Bien (2006) to differentiate the leader from leadership, and to focus instead on systems of relationships through which leadership manifests itself. Lastly, this study demonstrates that the contextual properties of networks –structure, mandate, function and level of diversity between members—deserve more academic attention. Future research could examine whether these contextual properties require different forms of leadership.

Conflict of interest statement

No potential conflict of interest was reported by the author(s).

Funding

This research is part of a larger research programme entitled ‘Furthering Public Leadership’ that was established in collaboration with seven Dutch government organizations and Leiden University.

References

Footnotes