The success of service delivery-networks You can maximize your managerial efforts, but Governments set your frontiers

This paper explores the role that Government should play when networks are established for the provision of public services. Should Government leave and give overall autonomy to the networked organizations? Or should it stay and manage the network activity? In order to answer these questions, we employed the theoretical framework developed by Kort and Klijn (2011) for governance networks, and investigated which relations exist between autonomization, management and performance, as far as service delivery networks are concerned. The results show that in order to make service delivery networks succeed, managerial efforts are important, but autonomy from the Government is also paramount.


Introduction
Governments have been involved in reform processes for a long time. Under the umbrella concepts of New Public Management (Hood, 1991;Ferlie et al, 1996), Public Governance (Ferlie et al., 2003;Rhodes, 1996) and New Public Governance (Bevir and Rhodes, 2003;Osborne, 2006), little by little, a new model of administration emerged. It is based on collaborative relationships between public and private players, non-profit organizations and social enterprises involved in the provision of public services (Isett et al., 2011;Klijn, 2008;Provan and Milward, 2001;Provan and Milward, 1995). Within this context of connected and networked organizations, the current economic and social crisis has raised further questions about the role of Government in public service delivery (McGuire and Agranoff, 2011). What role should Government play when networks are established for the provision of public services? Should Government leave and give overall autonomy to networked organizations? Alternatively, should it stay and manage network activities?
In order to answer the aforementioned questions, below we will employ the theoretical framework developed by Kort and Klijn (2011) for governance networks. In particular, we will explore which relations exist between autonomy from the Government (i.e., autonomization), actions implemented by partners to manage network activities (i.e., network management) and network performance, as far as service delivery networks are concerned. Trust among the network partners is also used to control the existing relationships (Klijn, Edelenbos and Steijn, 2010).
Public networks for the provision of homecare assistance for the elderly in Switzerland (Spitex networks) provided the empirical setting for our study. Data were collected through a survey of Spitex directors: 523 directors were contacted and 209 agreed to participate in our study, giving a response rate of about 40%. Exploratory Factor Analysis (EFA), Confirmatory Factor Analysis (CFA) and Structural Equation Modelling (SEM) were then used to analyse the collected data.
The results show that in order to make service delivery networks succeed, managerial efforts are important, but autonomy from the Government is also paramount.
Below, we will firstly review the existing literature on the role of Government in public networks and introduce the research questions behind our study, then we will present the theoretical framework and formulate the study hypotheses. Method and data analyses will follow. The paper ends with a discussion of the study results and some suggestions for future research in the field.

The role of Government in public networks
The role of Government in public networks is one of the longstanding topics in public network literature (Rhodes, 1996;Sorensen and Torfing, 2007). Some scholars focus on whether networks are displacing or replacing Government. Osborne and Gaebler (1992) wrote about eliminating the role of Governments, Milward and Provan (2003) spoke of a hollow State, Keating (1999) argued that Government is growing weaker at the core, and Frederickson (1999) stated that Government is losing its boundaries through disarticulation.
On the other hand, some scholars argue that Government still has a central role in public networks, as it has special power, unique resources, access to media and democratic legitimation (Klijn and Koppenjan, 2000;Kenis and Provan, 2006). According to Klijn and Koppenjan (2000), Government is a special actor in public networks and occupies a special position which cannot be taken by others.
Similarly, Skelcher (2004) showed that Government is the dominant partner in public networks, as it promotes the creation of partnerships, supports their emergence by allocating staff, time and political capital, and brings in other stakeholders around the partnership table. Koontz (2004) showed the critical role played by Government in supporting and encouraging collaboration, offering resources and technical assistance, convening meetings and developing decision making. Agranoff (2005) listed the roles that Government agencies can play when establishing the goals of public networks: exercising executive leadership in mobilization; working as a network promoter; being a network broker; helping network agencies to develop collaboration; investing resources; providing technical Daniela Cristofoli, Laura Maccio': The success of service delivery-networks. You can maximize your managerial efforts, but Governments set your frontiers. Jahrbuch der Schweizerischen Verwaltungswissenschaften 2013: Seiten 175-190 assistance; and assessing the value-adding dimension of collaboration. Edelenbos and Klijn (2007) identified three different Government strategies for networks: (1) attempting to change or influence the composition of the network; (2) influencing the network outcomes; (3) interacting in the network by influencing rules and processes. On top of this, by focusing on Government officials, Eglene et al. (2007) showed that the support of high level Government officials increases network success, and Agranoff (2007) showed the importance of Government leaders, acting as network champions, and top governance agency heads, forming the political core of the network.
Drawing from this evidence, in their recently published article "The limitation of public management networks", McGuire and Agranoff (2011) argued that it is "difficult to conclude that networks in any substantial way are replacing Government agencies, or that networks are controlling Government to any notable degree" (2011: 277): "Government is in the network" (2011: 277) and it is now necessary to focus on and better understand the role of Government within network boundaries.
By following McGuire and Agranoff's (2011) lead, below we will explore the Government's role when networks for the provision of public services are established. More specifically, we are interested in understanding whether it is preferable for the success of service delivery networks for Governments to leave and give overall autonomy to networked organizations or to stay and manage network activity themselves.

Theoretical conceptualization and hypothesis
In order to elaborate on the abovementioned research questions, we will base our study on the theoretical framework developed by Kort and Klijn (2011) and Klijn et al. (2010) for governance networks. By focusing on networks in the field of urban regeneration projects, Kort and Klijn (2011) identified three factors that can affect network performance, as far as the relations between Government and networked organizations are concerned: the autonomy of the networked organizations from the Government (hereinafter "autonomization"); actions implemented by partners to manage network activities (hereinafter "network management"); and trust among networked organizations (hereinafter "trust").

Autonomization
Kort and Klijn (2011: 619) defined "autonomization" (i.e. autonomy from the Government) as the "degree to which the organization has discretionary powers to make independent decisions on various matters, including the use of its financial resources, its organizational structure, and project-related plans". By following the New Public Management arguments (Hood, 1991;Ferlie et al., 1996), they

Daniela Cristofoli, Laura Maccio': The success of service delivery-networks. You can maximize your managerial efforts, but Governments set your frontiers. Jahrbuch der Schweizerischen Verwaltungswissenschaften 2013: Seiten 175-190
argued that when Governments set up separate organizations functioning at arm's length, they should be run in a business-like manner, with performance objectives and the necessary autonomy. Only in this way can established organizations work in a more managerial manner, far from political influence, and achieve better results (Hood, 1991;Osborne and Gaebler, 1992;Pollitt and Bouckaert, 2004;Van Thiel, 2001).

Network management
The existing studies mostly agree that the complex processes typical of public networks are unlikely to generate good performances without an extensive effort to manage partner interaction (Klijn and Koppenjan, 2004;Mandell, 2001;Meier et al., 2007, Kort and Klijn, 2011, Klijn et al., 2010Steijn et al., 2011). The number of actions listed in the literature to manage partner interaction is impressive. Steijn et al. (2011) categorize them into four different groups: connecting actors, exploring content (creating more variety, organizing research, exploring the perceptions of different actors, and so on), arranging the structure of the interaction (securing a temporary organizational arrangement for interaction) and establishing process rules (making temporary agreements and rules to govern interactions).

Trust
Klijn et al. (2010: 42) defined trust as a "stable positive expectation that actor A has (or predicts he has) of the intentions and motives of actor B in refraining from opportunistic behaviour, even if the opportunity arises (Edelenbos and Klijn, 2006). Trust is based on the expectation that actor A will take the interests of actor B into account." The impact of trust on the performance of governance networks has been proved (Klijn et al., 2010): better outcomes are achieved within network settings characterized by goodwill, agreement, absence of opportunistic behaviour and trustworthiness (Edelenbos and Klijn, 2006;Sorensen and Torfing, 2007). In addition to the direct effect on network performance of autonomization, management and trust, the existing literature has also shown that the abovementioned factors have an indirect effect on network performance. Kort and Klijn (2011) showed that greater levels of autonomization favour network management and thus network performance. Klijn et al. (2010) showed that managerial efforts to support connections and interaction within a network have a positive impact on the level of trust and on the performance of the network itself.
With this in mind, the model presented in Figure 1 can be suggested, and the following hypotheses can be formulated.

Daniela Cristofoli, Laura Maccio': The success of service delivery-networks. You can maximize your managerial efforts, but Governments set your frontiers. Jahrbuch der Schweizerischen Verwaltungswissenschaften 2013: Seiten 175-190
Hypothesis 1: In service delivery networks, higher levels of autonomization will lead to better performances. Hypothesis 2: In service delivery networks, higher levels of network management activities will lead to better performances. Hypothesis 3: In service delivery networks, higher levels of trust will lead to better performances. Hypothesis 4: In service delivery networks, the level of network management activities will be higher when autonomization is higher. Hypothesis 5: In service delivery networks, the level of trust will be higher when network management activities are higher.

Method
To evaluate the abovementioned model, we collected data in a survey and analysed them with structural equation modelling (SEM).

Sample
The unit of analysis of our study is the network as a whole (Isett et al., 2011;Isett and Provan, 2005;Kenis and Provan, 2009;Provan et al., 2007) and the empirical setting consists of public networks for the provision of homecare assistance in Switzerland (Spitex networks) .

Daniela Cristofoli, Laura Maccio': The success of service delivery-networks. You can maximize your managerial efforts, but Governments set your frontiers. Jahrbuch der Schweizerischen Verwaltungswissenschaften 2013: Seiten 175-190
Spitex networks provide health and social care services at home for the elderly, single parents, disabled people or other disadvantaged social groups. They are based on collaboration between: the Cantonal Government and municipalities that fund and control the network activities; a central organization (Spitex organization) that provides homecare services and coordinate the network partners; nonprofi t organizations for the provision of complementary services (such as public transport, assistance during the day and at night, and oncological assistance).

Data Collection
We used a survey to collect data. The questionnaire was translated into three Swiss national languages (German, French and Italian) and given to the directors of the 523 Spitex networks operating in Switzerland, in both a paper-based and an internet-based format. We collected 209 responses, with a response rate of about 40%. The characteristics of this sample of 209 networks are reported in Table 1. In accordance with the rules on how to conduct a good survey, the fi rst draft of the questionnaire was tested on three Spitex Directors. They confi rmed the face validity of most of the questionnaire items. However, for some items it was necessary to change the wording in order to cater better to the specifi c nature of the Spitex network activity.

Daniela Cristofoli, Laura Maccio': The success of service delivery-networks. You can maximize your managerial efforts, but Governments set your frontiers. Jahrbuch der Schweizerischen Verwaltungswissenschaften 2013: Seiten 175-190
The data collection process started in September 2012 and it was concluded in November 2012. A reminder was sent out to everyone two weeks before the deadline.

Autonomization
Autonomization was measured with the 6-item scale of autonomization (Cronbach's Alpha = 0.84) developed by Kort & Klijn (2011). On a scale from 1 to 7 (with 1 meaning "low power" and 7 meaning "high power"), we asked how much power the network had to make decisions on its own about the provision of homecare services, the definition of its own mission and vision, the definition of its own long and/or medium term programmes, the definition of its own objectives, the organization of inputs and tasks, and the use of financial resources.

Network management
Network management was evaluated on the basis of two of the four categories identified by Steijn et al. (2011: 5), namely "connecting actors" and "exploring content". "Connecting" refers to "securing contacts between actors, improving relations" (Klijn et al., 2010(Klijn et al., : 1081. "Exploring" denotes the implementation of activities with the aim of "exploring different views of actors and possible new solutions" (Klijn et al., 2010(Klijn et al., : 1081. We did not consider arranging and process agreement activities because they are well-established in all Spitex networks. The networks in place are, in fact, mature and in most of them there are stable, accepted organizational arrangements and rules for interaction. However, the necessity of connecting network members in ways other than formal contracts, improving their relations and promoting the exchange of ideas and practices are managerial activities that vary among Spitex networks and could be the key for better performances. Taking our cue from Kort and Klijn (2011), we developed a 6-item scale (Cronbach's Alpha = 0.87) and asked the Spitex Directors to what extent they agreed with our statements, on a scale from 1 to 7 (with 1 meaning "I totally disagree" and 7 meaning "I totally agree").

Trust
Trust was measured using the 5 items (Cronbach's Alpha = 0.85) developed by Kljin et al. (2010) to investigate the following nuances of trust: goodwill trust, agreement trust, absence of opportunistic behaviour, benefit of the doubt, reliability.

Daniela Cristofoli, Laura Maccio': The success of service delivery-networks. You can maximize your managerial efforts, but Governments set your frontiers. Jahrbuch der Schweizerischen Verwaltungswissenschaften 2013:
Seiten 175-190

Network performance
Referring to the results of our previous paper (Cristofoli and Macciò, 2012), we measured network performance on a community level, taking into consideration Provan and Milward's (2001: 416) definition: "Community-based networks must be judged by the contribution they make to the communities they are trying to serve.
[…] the goal of most public networks is to enhance client services through improved access, utilization, responsiveness, and integration, while maintaining costs." We identified 4 items (Cronbach's Alpha = 0.80) that were in keeping with this definition and fully represented the networks' roles in meeting patients' needs. We asked each network to state to what extent it agreed with our statements, on a scale from 1 to 7 (with 1 meaning "I totally disagree" and 7 meaning "I totally agree").

Control variables
The results of our study were controlled for resource munificence and external controls; this allowed us to consider the effects of the context on network performance (following Provan and Milward, 1995). Resource munificence was measured as the percentage of public funds received by the network. In order to evaluate external controls, a list of 6 main external controllers was drawn up and respondents were asked to select their own external controllers and add anyone that had not been included in the "others" box. The full list of items, sources and scale properties is given in Appendix A.

Data analysis and results
We used structural equation modelling to test our hypothesis. We estimated and assessed the model with SPSS and the Lisrel program. The measurement model is acceptable in terms of reliability and validity analysis (see Appendix A). In particular, the construct reliabilities (CR) were all above the 0.70 cut-off, the completely standardized estimated parameters (l) for each measure were high in value and statistically significant and the average variance extracted (AVE) of the constructs was higher than 0.5. The discriminant validity was also evaluated and confirmed.
The figure below shows the final model that tests the hypothesis among the latent constructs, given our theoretical conceptualization.

Figure 2: Results of Structural Equation Modelling Analysis (completely standardized coeffi cients and t-values in parentheses)
Evaluation of the structural model involved the use of multiple tests to appraise the fi t of the overall model, as there is no consensus among academics about the best test. In particular, we used (Hooper, Coughlan and Mullen, 2008;Hu and Bentler, 1999) the following indices: the (SB) chi-square, which requires a statistically insignifi cant result at a 0.05 threshold level for a good fi t; the relative/ normed chi-square that is chi-square/df, which should range between 1 and 3; the Root Mean Square Error of Approximation (RMSEA), which should range between 0.08 and 0.10 for a mediocre fi t and below 0.08 for a good fi t; the Standardized Root Mean Square Residuals (SRMR), which can go from 0 to 1, with values lower than 0.08 considered acceptable; the Non-Normed Fit Index (NNFI), which ranges between 0 and 1, with a threshold that is usually set at values greater than 0.95; and the Comparative Fit Index (CFI), which ranges from 0 to 1 and recommends that values for a good fi t should be greater than 0.95. We also calculated the Goodness of Fit (GIF), but we did not use it because the current consensus considers it to be affected by sample size (Sharma et al., 2005). The model showed a good fi t. Table 2 presents the overall fi t indices.  As reported in Figure 2, autonomization, network management and trust have positive and significant effects on network performance, thus confirming Hp 1, 2 and 3.

Daniela Cristofoli, Laura Maccio': The success of service delivery-networks. You can maximize your managerial efforts, but Governments set your frontiers. Jahrbuch der Schweizerischen
According to the existing literature (Klijn et al., 2010), network management has a significant and positive effect on trust, thus once again highlighting the importance of promoting trust among network partners with appropriate actions (Hp 5). However, the indirect effect of autonomization on network management is not confirmed by our data (Hp 4). Finally, neither of the control variables affect network performance.

Discussion and Conclusions
Some scholars (Osborne and Gaebler, 1992;Milward and Provan, 2003;Keating, 1999) claim that networks are undermining and taking the helm of Government, while others (Klijn and Koppenjan, 2000;Skelcher, 2004;Kenis and Provan, 2006) see the critical role of Government in controlling, managing and sustaining network activities and collaboration. However, McGuire and Agranoff (2011) argue that it is difficult to solve this debate since Government is an essential part of any public network and a better understanding of it within network boundaries is still necessary. Dealing with this issue, this paper investigated whether the success of service delivery networks could be better fostered by direct Government involvement in network activities, or by leaving more discretionary powers to the networks themselves. We produced a survey investigating the impact of autonomization, network management and trust on network performance, as identified by previous studies. We gave the questionnaire to the Swiss networks for the provision of health and social care services at home and collected 209 responses. The data were analysed with structural equation modelling. The results showed that autonomization, network management and trust have an impact on network performance; moreover, the level of trust was proved to be enhanced by higher levels of network management.

Daniela Cristofoli, Laura Maccio': The success of service delivery-networks. You can maximize your managerial efforts, but Governments set your frontiers. Jahrbuch der Schweizerischen Verwaltungswissenschaften 2013: Seiten 175-190
These results demonstrate that network success is sustained by managerial efforts, but autonomy from the Government is also paramount.
In this perspective, firstly, the result of our study seem to support the New Public Management stance on the marginal role of Government within networking settings: although it is an essential part of any public network, the Government should give networks the power to make autonomous decisions about important issues such as the use of financial resources or the definition of its mission and goals. At the same time, networks need to make substantial managerial efforts in order to succeed: they should be able to connect their members, improve and support their relations, and implement activities that explore their different positions and new opportunities. Moreover, a high level of managerial effort plays a dual role in promoting network performance. As well as having a positive and direct effect on network results, it can also indirectly support them by strengthening the level of trust.
Secondly, these results suggest that Government plays a unique role in the success of public service networks, in contrast with "governance networks". A study by Kort and Klijn (2010) investigated the effect of these variables on the performance of governance networks and showed that only managerial efforts and trust had a significant and positive effect on performance, while autonomy from the Government was not relevant. Therefore, the level of discretionary powers and autonomy is fundamental for networks that join together public and private organizations and deliver public services.
There are some limitations which hinder the generalizability of the results and causal relationships. The research design is cross-sectional and the sample concerns a single country. Some concepts are lacking in previously tested scales and their operationalization is mainly based on the literature. Therefore, a crossnational sample should be examined and our concepts should be tested in future research.