Managing successful networks : the case of intermunicipality collaboration in Switzerland

Intermunicipality collaboration is a popular phenomenon in Switzerland. Since the early 1990s, more than 60% of Swiss municipalities have established collaborative networks to jointly provide public services (Steiner 2003). Despite this, questions have frequently been asked in our country about whether intermunicipality networks really work and how they can be managed successfully. Taking their lead from Provan and Milward (1995) and Provan and Sebastian (1998), many scholars have investigated this issue, but with controversial results. As stated by Turrini et al. (2010), we maintain that this is due to the prevalent focus on the «tangible» predictors of network performance (i.e. contextual, structural, functioning and managerial characteristics of public networks) and the lack of studies exploring their simultaneous effect on network success. The aim of our study is to fill this gap. The results cast new light on the «dark side» of public network management. Firstly, they show the existence of a relationship between network structure and network functioning and management, thus emphasizing their joint effect on network performance. Secondly, they highlight the key role, in network success, played by the personal relationships between people in the partner organizations. Our study gives important suggestions about how to manage intermunicipality collaboration successfully: (1) choose managerial instruments and a managerial style that are coherent with the structure of your network; (2) if you are in a well-established and integrated network, ensure that you are flexible and enjoy the benefits of this approach.


Introduction
Intermunicipality collaboration is a popular phenomenon in Switzerland.Since the early 1990s, more than 60% of Swiss municipalities have established collaborative networks to jointly provide public services (Steiner 2003).Despite this, questions have frequently been asked in our country about whether intermunicipality networks really work and how they can be managed successfully.Taking their lead from Provan and Milward (1995) and Provan and Sebastian (1998), many scholars have investigated this issue, but with controversial results.
As stated by Turrini et al. (2010), we maintain that this is due to the prevalent focus on the «tangible» predictors of network performance (i.e.contextual, structural, functioning and managerial characteristics of public networks) and the lack of studies exploring their simultaneous effect on network success.The aim of our study is to fill this gap.
The results cast new light on the «dark side» of public network management.Firstly, they show the existence of a relationship between network structure and network functioning and management, thus emphasizing their joint effect on network performance.Secondly, they highlight the key role, in network success, played by the personal relationships between people in the partner organizations.Our study gives important suggestions about how to manage intermunicipality collaboration successfully: (1) choose managerial instruments and a managerial style that are coherent with the structure of your network; (2) if you are in a well-established and integrated network, ensure that you are flexible and enjoy the benefits of this approach.

Intermunicipality collaboration in Switzerland
Switzerland has approximately 2,500 municipalities, most of which have fewer than 1,000 inhabitants (only eight municipalities have more than 500,000 inhabitants).Swiss municipalities are very different from social, cultural and political standpoints (Steiner 2003), but each is responsible for providing a vast range of public services (i.e.education for all children of compulsory schooling age, social and healthcare services, utilities and waste disposal, construction work, local police, granting of civil rights internal organization of the civil authorities, finances and municipal administration).
During the 1990s, the growing citizen demand for high quality services and the budget constraints of many municipalities (Ladner and Steiner 1998) meant that legislative authorities were left asking themselves how to guarantee effective and efficient fulfilment of municipal tasks.Twenty-one Cantons issued reform laws (Ladner et al. 2000) to promote intermunicipality collaboration and munici-palities began to create networks for the joint provision of public services.Both Cantons and individual municipalities were sure that collaboration would give higher levels of efficiency and effectiveness in the services provided (Steiner 2003).
Gradually the phenomenon gained momentum, and within a few years 63% of Swiss municipalities were involved in networks for the joint provision of public services (Steiner 2003).Public services of all kinds (including information technology, accounting, residential monitoring and human resource management) are now provided through public networks, although the most widespread ones are related to school issues (75% of the Swiss municipalities) and health and social care issues (69% of the Swiss municipalities) (Steiner 2003).
Despite all this success in the spread of intermunicipality collaboration, many scholars and practitioners have started to ask themselves: «Do intermunicipality networks really work?What are the predictors of successful and unsuccessful networks?»

Literature review
Since the early 1990s, public networks have been implemented in many countries to solve «grave» public problems and address such issues as health, social care, local development and education (Provan and Milward 1995;Ferlie and Pettigrew 1996;O'Toole 1997;Agranoff and McGuire 1998;Provan and Sebastian 1998;Provan and Milward 2001;Milward and Provan 2003).Scholars, practitioners and even policy-makers saw public networks as the easiest way to reduce fragmentation and improve coordination among public agencies, thus increasing the efficiency and effectiveness of the services provided (Agranoff and McGuire 2001;Milward and Provan 2003;Ferlie and Andresani 2000).Such enthusiasm for the formation of public networks quickly gave way to doubts about whether networks in the public sector really work, as discussed by Provan and Milward (2001).Many scholars then began assessing network performance and identifying its predictors (Kenis and Provan 2009).Following Turrini et al. (2010), we will categorized these predictors into four groups (Figure 1): contextual (Provan and Milward 1995), structural (Provan and Milward 1995), functioning (Klijn 1996;Brown et al. 1998;Jennings and Ewalt 1998;Mitchell et al. 2002;Agranoff 2003) and managerial characteristics of public networks (Kickert et al. 1997;Hageman et al. 1998;Lasker et al. 2001;Shortell et al. 2002;Bazzoli et al. 2003;Conrad et al. 2003).Table 1 defines single predictors and sheds light on their relations with network success.
Contextual characteristics.Many studies focus on the characteristics of the external environment affecting network performance, and specifically environmental stability/instability (Provan and Milward 1995;O'Toole 1997;Mitchell et al. 2002;Conrad et al. 2003) and external controls (Provan and Milward 1995;O'Toole and Meier 2004).Generally, a curvilinear relationship seems to exist between environmental stability/instability and network performance (Turrini et al. 2010), while a positive relationship links external controls and network success (Provan and Milward 1995;O'Toole and Meier 2004).
Structural characteristics.A large amount of literature focuses on the relationship between the structural characteristics of public networks and their performance (Provan and Milward 1995;Provan and Sebastian 1998).More specifically, extant studies focus on factors including: integration (Provan and Milward 1995), centralization (Provan and Milward 1995), the presence of strictly integrated subgroups (Provan and Sebastian 1998;Echols and Tsai 2005;Voets et al. 2008) and network size and composition (Weiner and Alexander 1998).Whereas network integration and centralization and the presence of subgroups seem to positively affect network performance, network size and heterogeneity seem to be negatively related to it (Mitchell et al. 2002;Hasnain-Wynia et al. 2003;Zacocs and Edwards 2006).
Functioning characteristics.Many scholars have focused on the relationship between network functioning characteristics and network performance.In particular, they look into aspects such as: integration of mechanisms and tools (Provan and Milward 1995;Klijn 1996;Jennings and Ewalt 1998;Provan and Sebastian 1998;Agranoff 2003;Conrad et al. 2003), formalization (Brown et al. 1998;Mitchell et al. 2002;Hasnain-Wynia et al. 2003), accountability (Fawcett et al. 2000;Voets et al. 2008) and internal network stability (Juenke 2005).A positive relationship seems to exist between integration mechanisms and network performance (Provan and Milward 1995;Klijn 1996;Jennings and Ewalt 1998;Provan and Sebastian 1998).In contrast, the relationship between formalization and network performance seems to be controversial, as it varies depending on the level of network integration (Brown et al. 1998;Crozier Kegler et al. 1998;Jennings and Ewalt 1998;Mitchell et al. 2002;Hasnain-Wynia et al. 2003).Accountability and transparency seem to be correlated with higher rates of success in networks (Fawcett et al. 2000).Finally, internal network stability is positively related to network performance.
Network managers.Public management scholars and researchers have only recently focused their attention on the role of network managers in network success, and empirical studies testing their arguments are still lacking (McGuire 2002;Retheneyer 2005).Generally speaking, current studies propose two categories of actions that managers in the network can develop to actualize network success: nurturing the network (Kickert et al. 1997;Shortell et al. 2002;Bazzoli et al. 2003) and steering it (Hageman et al. 1998;Lasker et al. 2001;Shortell et al. 2002;Weiss et al. 2002;Conrad et al. 2003).
Bearing in mind the above considerations, the extant studies provide insights that can help to explain the performance of public networks.Nevertheless, all of the predictors identified by the extant literature deal with the «tangible» and «visible» side of public networks, thus neglecting the «less tangible» and «less visible» factors.In particular, they often neglect the potential for interaction between single factors: the effects of each predictor might be dependent on others.This might affect their overall impact on network performance, limiting it to incremental effects over time or enhancing and facilitating it.Thus, as stated by Turrini et al. (2010), the interaction effects among the predictors of network per-formance also need additional investigation and most causal mechanisms have yet to be explored.This is the objective of our paper.More specifically, it aims to explore the «visible» and «invisible» predictors of successful networks, by considering and investigating their simultaneous effects on network performance.

Method
Given the above considerations, our paper aims to explore the existence of interaction between the context, structure, functioning and management of public networks and look into their joint effect on network success.To this end, we have conducted a multiple case study.As stated by Yin (1984) and Eisenhardt (1989), case study is the most appropriate research strategy when the aim is to explore new avenues for research and the current theories seem inadequate.
The empirical setting for our study was provided by the public networks that Swiss municipalities set up to provide health and social care services in the Canton of Ticino.In particular, we will focus our attention on the intermunicipality networks for the provision of homecare services to the elderly.Originally, the Swiss municipalities set up consortia to provide these services.Following the 1997 Cantonal law on «Homecare services», in 2000 the Cantonal Government identified six geographical areas that could be efficiently and effectively served by single players, and forced the existing consortia to merge into six associations (Servizi di Assistenza e Cura Domiciliare -SACDs).Generally speaking, SACDs involve more than 50 partners on average, with the Cantonal Government and the municipalities on one side, and the organizations providing complementary services for homecare assistance (Red Cross, Pro Senectute, Atte, etc.) on the other.Normally, SACDs are governed by an Assembly, with representatives for each partner municipality, an Executive Committee of Assembly representatives, two directors (one responsible for managing the association and the other responsible for managing professionals such as nurses, psychologists and social workers), and teams of nurses, social workers, psychologists, etc. who are responsible for providing homecare services in different geographical areas.As a rule, the SACDs are contacted by medical doctors requiring care for their patients.The physicians' requests are evaluated by SACD experts, and on the basis of this evaluation a package of homecare services is identified and specific organizations are contacted (e.g.Pro Senectute, Red Cross, etc.).SACDs are funded by the Cantonal Government (which also controls their activities), by municipalities, by donors and by fees paid by patients.In particular, in the event of financial deficits, replenishment is provided for the SACDs by the municipalities.Four years after the establishment of the SACDs, the Cantonal Government assessed their performance in order to evaluate the success of the 2000 law.The results were positive for all six SACDs.
For the purposes of our analysis, we selected two similar cases.Both of them are successfull stories and they involve the biggest SACDs in Ticino in term of size.We will label them Lu-SACD and Lv-SACD in order to preserve their anonymity.
Data were collected through semi-structured interviews (Denzin and Lincoln 1994) with all of the key players in the organizations involved in the two networks.Most of the interviews were conducted by at least two researchers, with one asking questions and the other taking notes.They were also typed and transcribed within the next 24 hours.Table 2 shows the number of interviews for each case.Data were analyzed according to a three step procedure: storing, managing and processing (Miles and Huberman 1984).First of all, we grouped our data into four categories: network context, structure, functioning and management.Secondly, taking our lead from the extant literature, we identified single variables within each category, and then looked for relations between them.In this way, we were able to both identify new predictors of public network performance neglected by extant studies and look for interaction between them.To ensure that the qualitative analysis was reliable (Denzin 1978, Denzin andLincoln 1994), we first triangulated the sources and then used the researchers' triangulation results and arranged meetings to discuss the data with our colleagues and interviewees.

Findings
Table 1 shows the results of our study.Our findings show that the two SACDs analysed present similar contextual characteristics, but different structural characteristics.Furthermore, there are important differences in some aspects concerning functioning mechanisms, and the managerial approaches are in stark contrast.
In terms of contextual factors, Lu-SACD and Lv-SACD are quite similar.In both cases, the stability of the relative environment seems to be high from both a financial and a socio-political point of view.As the General Director of Lu-SACD said: «Financial resources are not a problem.At the end of the day, we find the money we need.»This situation is confirmed by the Director of Lv-SACD «We have never suffered from financial distress.»In any case, the municipalities will bail out the SACDs if they find themselves in financial difficulty.In both cases, the SACDs operate in a context with the high degree of civic culture and political participation that is typical of Swiss society (Ritz 2010).Last but not least, both SACDs are subject to the control of the Cantonal Government.
The network structural characteristics are, on the other hand, significantly different (Figure 2).In both cases it is possible to identify high levels of heterogeneity among the network members (i.e.municipalities, with political interests and competences; managers, who are responsible for the SACDs' work; professionals, who provide the homecare services for the elderly), but there are different num-  As Figure 2 shows, Lu-SACD is centrally integrated, whereas Lv-SACD has a fragmented and horizontal structure.To look at things in more detail, firstly Lu-SACD partners are strictly interdependent, while Lv-SACD has two cohesive subgroups, which are responsible for providing homecare services in two different geographical areas and related by a contractual agreement.Each of them is run by a Director and benefits from a high degree of autonomy and independence.As the Director of one of the subgroups said: «Thanks to the contractual agreement, we can take advantage of the collaboration without losing our independence.»Secondly, Lu-SACD has a General Director's Office that serves as a central core agency and coordinates the network members, as we were told by the Financial Director.There is no such core agency in Lv-SACD, and coordination tasks are performed by the two Directors, each for his own «sub-network».
Regarding functioning characteristics, Lu-SACD and Lv-SACD present both similarities and differences.Both of them have common information and communication systems to support integration and coordination among the network partners.They are especially useful for the team leaders when they are managing and coordinating the daily activities of professionals who are responsible for jointly satisfying the needs of individual patients.In Lu-SACD one of the partner organization responsible for service provision maintained that «Most of the information (concerning the patients' needs) gets to us thanks to the Lu-SACD information and communication system,» although he acknowledged that «a lot also gets through by word of mouth».Similarly, despite the network's fragmented structure, the Lv-SACD Director said: «We are a network, and we decided to share the administrative system».Both SACDs have accountability mechanisms (e.g. an updated website showing data and results, reports, brochures, etc.), although they seem to pay varying degrees of attention to them.The Lu-SACD interviewees appear to consider accountability mechanisms a «well established formality» and do not heed them a great deal, as «our goal is to communicate with our patients: we respond to THEIR questions» declared the Director.On the contrary, most of the Lv-SACD interviewees were very much aware of all of the tools and instruments used for external communication.They quoted: annual reports for the Cantonal Government, yearly reports for municipalities, brochures for citizens and patients, conferences through which future strategies and projects are presented, etc. Lu-SACD and Lv-SACD also seem to be similar if we consider the networks' internal stability.They show great stability in terms of personnel, both on a managerial and a professional level: the Directors have been the same since the creation of the networks, as have many of the team leaders and members.However, some differences can be seen in the degree of formalization of the two SACDs' activities.As with all public sector organizations, the work done by both Lu-SACD and Lv-SACD involves a high degree of formalization (formalized rules, organization of meetings, written agendas, decision-making procedures, etc.); what is different is the way the two SACDs rely on it.In Lu-SACD, formalization does not seem to be the main driver of decision making processes on a strategic and organizational level, and it does not appear to typify meeting and agenda planning (which are instead driven by organizational and managerial needs).As stated by the General Director, «The Executive Committee plays an important role in directing SACD activity, but then it also allows the Management and the team leaders to make very important decisions».In the same way, the relationships between Lu-SACD and the organizations providing voluntary services are not supported by any formalized contracts: «There are no formal agreements… the collaboration is based on a very spontaneous agreement,» said the Director of one of the organizations.«People do not contact each other because of written rules but due to their personal relationships».In contrast, Lv-SACD seems to rely on formalization a great deal.The decision-making process is based on formal rhythms: all interviewees sustain that most of the strategic and operational decisions are made by the Assembly or the Executive Committee «as required by the law and established in the Statute».No informal relations among people are mentioned.Planned meeting and coordination instruments are also seen as the best solution for «clarifying situations or sharing a strategy» (Lv-SACD Director).«Most of the contact with the municipalities goes through the Assembly» declared one of the Directors.«The recommendations and instructions from the municipalities are expressed by their representatives in the Assembly» confirmed the President of the Association of Municipalities.Regarding relations between the network partners, interviewees constantly refer to all of the formal contracts and rules governing them.During the interview, one of the Lv-SACD Directors said the word «contract» fifteen times.Meetings at all levels of the network are scheduled and well-planned: once a week Lv-SACD representatives meet representatives of voluntary organizations.5 to 6 meetings a year take place to coordinate the activities of the team leaders.More relevant is the fact that formalization is the key variable that regulates relations between the two subgroups in Lv-SACD: a contract defines the connections between the two and justifies their managerial autonomy (Figure 2).
Regarding network management, the two SACDs are systematically different.First of all, while it is easy to identify a network manager in Lu-SACD, as the General Director leads the central core agency and acts, at the same time, as mediator, facilitator and leader, it is difficult in Lv-SACD, where the skills necessary to manage the network are shared by different people belonging to different offices.In particular, firstly, the Lu-SACD General Director shows skill in dealing with the various parties in the event of disagreements.For example, he organized meetings with staff members to ease the tension caused by some personnel dismissals, and as a team leader remembered: «When one of the partner Municipalities declared it was unsatisfied by our services, the Director supported us, and helped us to meet the Municipality representative and to find a solution».On the contrary, in Lv-SACD, tensions among partners are usually solved by the governing bodies (the Executive Committee or Assembly) and do not involve the Director's Office.Secondly, the Lu-SACD Director also shows the ability to reconsider the operating rules of the network if necessary.For example, when faced with difficulties in implementing local commissions, which are expected to monitor the needs of patients living in different geographical areas, he relied on informal relations with individuals working in the areas in question: «Originally we implemented the local commissions, as required by law, but they didn't work very well, so in order to be able to satisfy our patients' needs we preferred to rely on trustworthy people working in these areas».On top of this, he has perfect knowledge of the perceptions and language of network partners both at a strategic and an operative level, since he has full awareness of the context and strong relations with all of the network members «In the health and social care sector, people know each other.There are long term relationships.Normally people have been working in the sector for years».In Lv-SACD, the same tasks are performed by the Director, the President and the Vice-President of the Committee all together.Many interviewees recognize this group as the real managerial apex of the network, which is able to manage and change the network operating rules if necessary, and to understand the opinions, interests and specific language of each network partner.Thirdly, in keeping with the existent literature, the Lu-SACD Director shows the ability to build up commitment on critical changes: «We had no particular opposition because we all recognized our responsibilities in the community».The Financial Director argued that «it is thanks to the Director that we have a flexible and dynamic organization».In Lv-SACD, the governing bodies support the creation of commitment, working as an aggregator of interests and balancing power on strategic issues discussed in the Assembly.Fourthly, the Lu-SACD Director clearly shows the ability to develop focused strategies and interventions, by engaging in participative decision-making or in decision-making that facilitates interaction, especially at the Cantonal level.Fifty, he shows the ability to activate the whole network, by «taking emergency measures» to satisfy the citizens' needs.He enjoys the trust and consensus of all network members, and can, as a consequence, intervene to solve problems even if it means activating human and economic resources outside the expected channels.Both of these abilities are shared by the governing bodies in Lv-SACD.Finally, in neither case did we note any radical repositioning of the network.

Conclusion
The results of our study are important for both theory and practice.From a theoretical standpoint, they enrich the existent literature on the predictors of public network performance by shedding light on the «dark side» of network management.Firstly, they show the existence of a correlation between network structure, functioning and management, with the latter depending on the former (Figure 3).Secondly, they highlight the presence of another important factor behind network success: the personal relationships between network members.
In empirical terms, our study provides public managers with important suggestions on how to manage public networks successfully.First, it shows that it is a good idea to choose managerial instruments and a managerial style that are coherent with the network structure.As in the two examined cases, the more integrated and centralized the network is, the less formalized the mechanisms for its functioning will be, and the greater the centralization of power in the role of the network manager will be.In contrast, the more decentralized and fragmented a network is, the more important it will be to rely on formalization on the one side, and power sharing on the other.Secondly, the results of our study seem to suggest that network managers in a well established and integrated network should adopt a more flexible and informal style of governance, by neglecting formalization and relying on the relationships between people.
However, the paper presents the preliminary findings of a wider research project funded by the Swiss National Science Foundation, which aims to test the joint effects of the contextual, structural, functioning and managerial characteristics of public networks on network performance.In this perspective, it is an exploratory study, the results of which need to be further investigated.First of all, the empirical setting could be extended by including networks for the provision of homecare services to the elderly in other Swiss Cantons.Secondly, a quantitative study could be developed through a survey involving all of the Swiss public networks offering homecare services to the elderly, with the aim of simultaneously testing the joint effects of all of the predictors of network performance. 133

Figure 1 :
Figure 1: Conceptual model of public network characteristics

FiguresFigure 1 :
FiguresFigure 1: Conceptual model of public network characteristics Figure 2: Lu-SACD and Lv-SACD -Network structure Figure 3: Definitive model of public network characteristics

Table 2 :
List of intervieweesbers of partners (about 50 in Lu-SACD and around 70 in Lv-SACD).The two networks are also characterized by different degrees of density and centralization.

Table 1 :
Public network characteristics and study results✔ characteristic is present o characteristic is absent

Table 2 :
List of interviewees