The Inner Functioning of Local Governance Networks in Centralized Countries: A ‘Brave New World’?

: Local governance networks are increasingly seen as the big idea to cope with issues that are complex enough in scope and scale to require a diversity of expertise and resources. While conventional narrative has posited that local networks are optimal for addressing a range of policy problems, and enhancing democratic participation, scarce attention has been devoted to understanding their inner working. A relevant gap in the literature pertains to the impact of central government intervention in igniting such arrangements on the diversity of actors, the intensity of interactions among actors, or their coordination practices. Such assessment is particularly relevant in centralized contexts. This article seeks to map and characterize the inner working of local networks in such a context—Portugal. The empirical analysis highlights the crucial role of central government in igniting local networks and in ensuring higher levels of formal intensity of collaboration to reduce transaction costs. Keeping such arrangements under the radar of central government, however, may curtail the diversity of actors, policy areas, and curb stakeholders’ commitment in local governance arrangements.


Introduction
Defied by political, social, economic, and demographic challenges, contemporary local governments have been forced to operate within a diverse range of formal or informal structures of exchange and production with other relevant stakeholders, namely through local governance networks (LGNs) (Egner et al. 2022;Teles et al. 2021;Teles 2016;Stoker 2004Stoker , 2011. Through these networks, local governments aim at addressing a range of policy problems, securing agreement on solutions, and struggle to maintain negotiated consensus between involved actors over time (Lewis et al. 2017;Nelles 2013;Teles 2016).
LGNs, hence, emerge as arrangements that ignite complex negotiations between members and seek to secure state-society agreement over time, with varying levels of institutionalization (Silva et al. 2022).
The increasing use of these LGNs has been subject to increasing academic interest. However, most of this research has been focused on countries with a significant governance cooperative culture, such as the Netherlands, Switzerland, Denmark, or the United States of America (Lewis et al. 2017;Agranoff and McGuire 2001;Cristofoli et al. 2021), and mostly focused on the variables that impact the performance of such networks, as mentioned by Klijn et al. (2020).
Despite extant research, many questions remain unanswered. From a theoretical standpoint, while conventional narrative has posited that LGNs are optimal for addressing a range of policy problems, and enhancing democratic participation, scarce attention has been devoted to understanding the factors that influence the inner working of LGNs. Specifically, the role of central government intervention in terms of the diversity of actors, the intensity of interactions among actors, or their coordination practices, particularly in Soc. Sci. 2023, 12, 289 2 of 14 centralized contexts, has been less explored.
LGNs may have indeed started to become more common since the 1990s, but there are still huge challenges attached to their use vis-a-vis the traditional use of hierarchies, particularly in centralized countries where the hierarchical mindset prevails, hence our question about the use of networks in such contexts being a "brave new world".
The empirical assessment of the inner workings of LGNs has also been limited. Considerable research has been focused on the strengths and limitations of the repertoire of tools of regional governance (Feiock 2007). As Teles (2016) rightly asserted, existing research has not yet provided enough attention on mapping and assessing the complexity of LGNs, or the choice of the architecture and intensity of collaborative networks. Existing conceptual toolkits to measure the extent to which LGNs are effective arrangements tend to be pursued within the confines of intermunicipal networks (Nelles 2013;Silva et al. 2018), with little systematic attempts to assess the governance capacity of all forms of LGNs, without attempts at clarifying the impact of centralized governmental-administrative procedures in determining the functioning of LGNs.
This article seeks to address these gaps, by mapping LGNs' inner functioning, analyzing their diversity, intensity, and coordination practices, as well as the influence exerted by the central government and territorial contexts in this inner functioning. It does so by considering a wide repertoire of LGNs across several policy domains which were created either by a mandate from the central government (top-down) or created by the initiative of local governments or other local actors (bottom-up). Empirically, this research 1 draws in a unique dataset that captures the diversity in LGNs in Portugal, where governance collaborative efforts have become more common (Mota et al. 2021; Mota and Bittencourt 2019) but where local government is kept tightly under the radar of central government.
Overall, results suggest that in centralized countries, LGNs can demonstrate high levels of vitality both in terms of the number and typology of actors involved. Central government directives emerge as fundamental in igniting LGNs, particularly in smaller and less densely populated municipalities. Additionally, central governments' involvement tends to trigger the intensity of LGNs. Such a role, however, may negatively impact the number of policy areas dealt by LGNs, and the number of local stakeholders involved. Moreover, the tendency to rely on municipalities for the management of such arrangements may discourage other stakeholders' participation.
This article is structured as follows. First, the theoretical framework to analyze LGNs is presented. The second section seeks to describe the variables that can be empirically conceptualized to comparatively map the complexity of LGNs and their functioning. This is followed by a specific account of the relevance of such networks in the Portuguese case, as a typical case study of low levels of decentralization. Section 3 presents the data collection strategies, and the operationalization of variables, and Section 4 presents the main results and discusses these findings. The article ends with a summary of the research findings, theoretical and practical implications, and future research agenda.

Local Governance Networks: A Brave New World?
In a convoluted and multi-layered world of limited resources and multiple demands, governance networks-here defined as "sets of autonomous yet interdependent actors (individuals, groups, organizations) that have developed enduring relationships in governing specific public problems or policy programs" (Klijn and Koppenjan 2014, p. 61)-came to be regarded as "an effective and legitimate way of governing our increasingly complex and fragmented societies and economies" (Torfing et al. 2012, p. 30).
Although not an entirely new phenomenon in Western societies, the interaction between a plethora of public, semi-public, and private actors in policy formulation and implementation networks grew exponentially over the last decades, and especially from the 1990s (Klijn and Koppenjan 2015;Sørensen and Torfing 2007). At some point, an eventual paradigm shift "from government to governance" has been started to be discussed in the literature, but also widely contested with some authors arguing the idea of a loss of power from traditional political and public actors to be clearly exaggerated, claiming that the discussion should be rather about the "changing powers" of these actors (Torfing et al. 2012;Peters et al. 2022). Regardless, it has been posited that no single actor possesses, on its own, enough knowledge, capacity, scrutiny power, or action potential to address the complexities and challenges of public policy implementation, service delivery, and the growing number of "wicked problems" (Klijn and Koppenjan 2015;Torfing et al. 2012).
Despite their increasing importance, governance networks may not be regarded as a governance panacea. Their advantages, such as the promotion of more innovative and flexible policy solutions, are counterbalanced by several potential problems, such as the difficulty to coordinate multiple actors with different interests and logics of action, the blurring of responsibilities, or the capture of participation by more powerful actors (Klijn and Koppenjan 2015;McGuire and Agranoff 2011;Torfing et al. 2012).
The subnational governance sphere has been no stranger to such reformist trends. As Teles (2016, p. 7) rightly asserted, most of the "significant territorial reforms" the European continent has experienced over the last four decades have been accompanied by "interesting experiments on local governance systems". Driven by national governments to address complex issues that can best be met at the local level (Stoker 2004), fostered by the European Union (EU) as instruments of regional development policies (McCann 2015) or self-grown, LGNs are seen as positioned to tackle political, social, economic, and demographic challenges confronting local authorities and their communities today.
The increasing importance of networks in different governance levels has been paralleled by a growing academic interest. After a first generation of studies focused on governance networks' comparative advantages and disadvantages, a second generation of studies has been focused on the identification of the main factors that explain their success and failure (Sørensen and Torfing 2007), while also providing relevant frameworks of analysis (e.g., Ansell and Gash 2008;Skelcher and Sullivan 2008). One of the most famous of these frameworks is the one presented by Ansell and Gash (2008), who explained that the networks' outputs are influenced by the collaborative process itself, which, in turn, is influenced by its starting conditions, the institutional design, and the facilitative leadership.
The existing research on the topic has nevertheless neglected the study of the potential variables that determine the choice of the inner workings of LGNs, such as the diversity of network members, the intensity of their work, or the coordination structures and practices, despite very recent and notable exceptions (Klijn et al. 2020). One of these often-neglected variables pertains to the effect of central government on the inner working of LGNs or the importance of territorial context (high or low-density municipalities), particularly in centralized countries and/or countries with a scarce tradition of cooperation (Mota and Bittencourt 2019) or in countries with huge differences between urban and rural areas. If we consider central governments as potential 'sponsoring' agents or even as network managers or coordinators, their importance to network performance is recognised, particularly when there is a lack of trust among network members or when there is a need to gather a wide diversity of actors (Ansell and Gash 2008). Most of these studies, however, are concentrated within contexts with robust local governments and strong civil societies (e.g., Agranoff and McGuire 2001;Lewis et al. 2017;Milward and Provan 1998). Centralized countries where LGNs are being increasingly used but where there is no collaboration culture have been disregarded, despite the considerable influence of central governments as the main drivers of LGNs with the potential drawback on such networks' autonomy, as illustrated by the Portuguese case (Mota et al. 2021).

Mapping the Inner Functioning of Local Governance Networks
This research seeks to map and characterize the wide spectrum of LGNs, considering unitary and centralized European countries, particularly Portugal. Specifically, this article seeks to chart the diversity, intensity, and coordination of Portuguese LGNs and to assess both their interactions and the effect of central government intervention as a stimulator of LGNs. This section seeks to present the relevance of these three indicators.
The first relevant indicator is related to diversity. Diversity reports to the "inclusion of a diverse range of actors and institutions" in LGNs (Carabine and Wilkinson 2016, p. 64). The concept highlights the fact that networks' actions and outcomes are affected by dyadic (or multiple) relations, as well as by the overall structure of the network (Robins et al. 2011). It refers to the type of actors (public, private, third sector organizations, etc.) involved and the complexity of LGNs: as the diversity of participants increases, the diverse interests, goals, and expectations make the identification of a shared strategy towards common objectives more complex (Eagle et al. 2010). It has also been argued that a wide variety (of types) of actors may curtail the scope of potential policy agreements and the credibility of commitments to fulfil agreements (Schneider et al. 2003). Diversity may also be stimulated or curtailed by the degree of centralization of interactions, i.e., how hierarchical the structure is, and the role central government has in stimulating and sponsoring these networks (Ansell and Gash 2008). On what concerns the territorial context, less diversity may be expected in municipalities of low population density, as the social fabric is less dense.
The second variable is the intensity of interactions. Intensity refers to the frequency of interactions and meetings among members.
LGNs with greater diversity may require a higher degree of intensity to promote group cohesion, reduce enforcement and monitoring costs, and to increase the cooperation within LGNs (Park and Park 2009). In turn, this cohesion promotes various positive benefits, such as higher internal support, higher individual satisfaction, lower team conflict, and lower team member turnover. Thus, members of a cohesive group develop shared values and team loyalty, creating smoother and more effective communication (Sandström and Carlsson 2008;Silva et al. 2022). The relationship between group cohesion and network success has been widely explored, empirically suggesting a strictly positive relationship between group cohesion and network performance (Balkundi and Harrison 2006). Consequently, one may assume that LGNs whose creation was mandated by the central government may require more regular meetings, since the network was not created by the initiative of either of the network members. On what concerns the territorial context, less intensity may be expected in municipalities of low population density, as there are more chances of network members having a history of previous cooperation.
A third relevant dimension pertains to coordination. Coordination refers to the presence of a link (or structure) that supports governance processes and activities, with the purpose of bringing stakeholders to a consensus and engaging each other in a collaborative spirit. As explained by Torfing et al. (2012), coordination or leadership functions in governance networks do not need to be necessarily developed by top political/public decision makers. Instead, other actors may play relevant roles, as long as they are perceived as being legitimate actors, have access to key resources, and have institutional capacity to monitor and coordinate the network. Therefore, one may assume LGNs that are mandated by the central government or created by municipalities to be coordinated by political decision makers. It is expected for this trend to be even higher in municipalities of low population density, as municipalities may be the only network member with organization capacity to coordinate the network.
While diversity and intensity may reduce the efficiency of LGNs-as transaction costs tend to increase (Park and Park 2009)-coordination may emerge as a relevant intervening variable. If a core faction acts as the LGNs gatekeeper (with an effective network manager), it becomes possible to accommodate a larger set of participants and widen the sphere of action, since the involvement of all actors may no longer be required for many (routine) network decisions. Hence, a centralized coordination (with specific external structures) can help circumvent the time-consuming and resource-intensive activities, along with the transaction costs involved (Feiock and Scholz 2009;Park and Park 2009). Arguably, a centralized specific management structure may facilitate the ability of attracting different stakeholders and facilitates (or coerces) collaboration among actors (Ansell and Gash 2008;Emerson et al. 2012).
Centralizing the integration of the network through a central core agent (e.g., network initiator, structure of coordination, and/or leadership) could have effects on the commitment degree within network members (Balkundi and Harrison 2006).

The Portuguese Context
This research is focused on the Portuguese case, as a typical case of a unitary European country, with a low level of decentralization.
Since the transition to democracy, Portugal has gradually established local autonomy as one of the basic principles of local government, by defining new competencies, powers, and resources in the 1976 Constitution. Since 1976, the Portuguese Constitution envisages three types of local authorities: administrative regions, municipalities, and civil parishes. Administrative regions remain, however, as a mere constitutional feature in the mainland, with only two levels of subnational governance: municipalities and civil parishes. The regional level thus only exists for the two autonomous regions-the archipelagos of Madeira and of Azores-which have their own regional governments and legislative assemblies. With a somehow dormant debate of regionalization, one has been witnessing different timid waves of decentralization with municipalities receiving some competences in several policy domains, although mostly related with public management rather than with policy formulation and not always followed by the correspondent transfer of funds (Teles 2021).
Despite the mentioned principle of local autonomy and the increasing number of competences being transferred to local authorities, Portugal still displays a low level of actual decentralization. Taking into consideration the Decentralization Index of the European Committee of the Regions (CoR n.d.), Portugal has a medium-low level of overall decentralization, with local authorities having a medium level of formal competences and political power but a low level of human and financial resources at their disposal.
In the absence of administrative regions, reinforcement of subnational levels of government would occur through the creation of intermunicipal platforms, which would be sufficiently comprehensive at geographic and population levels that could serve as groundwork for new steps in decentralizing the country. The Portuguese Constitution of 1976 recognizes the right to establish associations and federations of municipalities to administer common interests. Since then, many associations were created aiming at specific common purposes in areas such as water supply and wastewater treatment; urban cleaning and waste management; development of software for local authorities; distribution of fuel gas; environmental protection; integrated management of local resources; and culture and regional development.
These features make local government relatively fragile in Portugal, at least compared with most of its European counterparts (Teles 2016), particularly concerning their financial means and the fact that Portuguese local authorities remain tightly under the radar of the national government, despite the gradual increase in local government competences. While working within a relatively fragile scenario, mayors are particularly relevant and powerful vis-à-vis other elected members, given the electoral system and mayors' wide-ranging set of powers (Sousa 2015;Tavares et al. 2018). The pivotal role of both the mayor and the municipality may also weaken the active participation of civil society actors (Tavares et al. 2018). Overall, Portugal has been characterized by low levels of civil society involvement in advocacy functions, focused instead on service provision tasks (Franco 2015).

Data and Methods
This study focuses on LGNs in which Portuguese municipalities develop their political activity. There is a total number of 278 municipalities in mainland Portugal. To allow for an analysis of the different LGNs, a representative sample was selected, through a partially intentional sample of sixteen (16) municipalities, considering the NUT II level of territorial division 2 . As this system of territorial organization includes NUT with different Soc. Sci. 2023, 12, 289 6 of 14 municipality numbers, a stratified sample was built considering municipalities with high and low population density, as reported in Table 1. Criteria for the assessment of population density derive from data available at PORDATA (n.d.) (data regarding 2018), estimating an average value at the national level (111.7 in 2017). Below this value, municipalities were considered to be low density; above this value, they are equated as high-density municipalities. Within this frame, in some NUTS, no cases would have been selected (e.g., high-density municipalities in the Algarve/Alentejo). In these cases, one case was included to maintain national coverage. Source: own production.
After the selection of these 16 municipalities, we performed a systematic scoping analysis of municipalities' websites and Facebook accounts to identify the existing LGNs (our unit of analysis) in each municipality. A total of 167 networks were identified. The identified 167 networks were then analyzed concerning the identified variables through a systematic analysis of networks' statutes, election minutes, and meetings. Formal constitution decrees were also considered. Table 2 provides the operationalization of each dimension used to characterize LGNs, as well as descriptive data. Diversity of networks is assessed through the number of members and the fragmentation of LGNs across different policy areas. Arguably, addressing a wider range of policy domains generates higher predisposition of actors to strategic and cross-cutting problems, which encompass the higher commitment of municipalities (Nelles 2013). Intensity is operationalized through the frequency of interactions within network stakeholders (Hawkins 2010). For analyzing this dimension, the annual number of times that members meet was considered, assuming the formal constitution of the arrangement. When specific regulations were not available, general rules defined by national law (decree) were assumed. Finally, the coordination of LGNs was assessed through the different coordinators: 98 led by municipalities; 24 by other stakeholders; and 44 coordinated by specific management structures. In addition to the variables related to the inner functioning of LGNs, networks were also classified according to their drivers. The vast majority of LGNs were mandated or promoted by national law (107 networks). The remaining networks were created by the initiative of municipalities (23) and other local stakeholders (38).

Data Analysis and Discussion
This article seeks to characterize these networks considering their inner workings (and interconnections), estimating their diversity, intensity, and coordination, while estimating the effects of central government on such dimensions. The descriptive statistics of data gathered are reported in Table 3. Despite being frequently absent from the mainstream of scholarly debates, LGNs structures have made their way within this potentially unfavorable scenario of Portugal, though with different degrees of diversity of actors, intensity of cooperation, and coordination. As it is possible to observe, there is no big difference between municipalities with high and low population density on what concerns the tendency to be involved in LGNs or the average number of involved network members.
As revealed in Figure 1, these networks in low-density municipalities are mostly driven by national governments and local stakeholders.
LGNs in low-density areas tend to be directly stimulated by the central government, as a deliberate strategy to stimulate the emergence of LGNs in a policy field (UCLG 2013). For the central government, these networks constitute a means to stimulate the implementation of certain policy objectives at the local level and facilitate territorialized policies, which are tailored to local needs and contexts. It is also relevant to the fact that bottom-up processes of creation of LGNs are particularly relevant in low-density areas. This somehow surprising result may be related to the emotional attachment local stakeholders have to their municipality or even the pre-existing social capital. On the contrary, municipalities tend to be the main drivers of LGNs in high-density regions, where local stakeholders depict a lower capacity or interest to be involved, at least as promoters of such networks, as portrayed in Figure 1. in high-density regions, where local stakeholders depict a lower capacity or interest to be involved, at least as promoters of such networks, as portrayed in Figure 1. Notwithstanding, there are no significant differences between the number of policy areas tackled by LGNs operating within high-or low-density municipalities (Kruskal-Wallis H test: χ 2 (1) = 0.011, p = 0.915). This does not mean, nevertheless, that LGNs dealing with multiple policy areas tend to include a wider variety and number of local stakeholders. On the contrary, results of the Pearson correlation indicate that there is a significant negative association between the number of policy areas dealt with by the LGNs and the number of stakeholders involved (r(166) = −0.202, p = 0.008), a pattern that holds when we consider both low-and high-density municipalities, as depicted in Figure 2. The negative correlation tends to suggest that dealing with complexity and avoiding the fragmentation of issues between specialized policy areas require greater commitment from local stakeholders, which may be difficult to achieve. In fact, only 35% of the networks develop activities in more than one area (67 out of 167). Here, context matters. Within centralized countries, LGNs tend to be able to operate on a stringent set of policy areas, as most policy issues remain a prerogative of national governments. Portuguese LGNs tend to be focused on social and youth policy (14.7%); environment (8.9%); and Notwithstanding, there are no significant differences between the number of policy areas tackled by LGNs operating within high-or low-density municipalities (Kruskal-Wallis H test: χ 2 (1) = 0.011, p = 0.915). This does not mean, nevertheless, that LGNs dealing with multiple policy areas tend to include a wider variety and number of local stakeholders. On the contrary, results of the Pearson correlation indicate that there is a significant negative association between the number of policy areas dealt with by the LGNs and the number of stakeholders involved (r(166) = −0.202, p = 0.008), a pattern that holds when we consider both low-and high-density municipalities, as depicted in Figure 2. Soc. Sci. 2023, 12, x FOR PEER REVIEW 8 of 14 in high-density regions, where local stakeholders depict a lower capacity or interest to be involved, at least as promoters of such networks, as portrayed in Figure 1. Notwithstanding, there are no significant differences between the number of policy areas tackled by LGNs operating within high-or low-density municipalities (Kruskal-Wallis H test: χ 2 (1) = 0.011, p = 0.915). This does not mean, nevertheless, that LGNs dealing with multiple policy areas tend to include a wider variety and number of local stakeholders. On the contrary, results of the Pearson correlation indicate that there is a significant negative association between the number of policy areas dealt with by the LGNs and the number of stakeholders involved (r(166) = −0.202, p = 0.008), a pattern that holds when we consider both low-and high-density municipalities, as depicted in Figure 2. The negative correlation tends to suggest that dealing with complexity and avoiding the fragmentation of issues between specialized policy areas require greater commitment from local stakeholders, which may be difficult to achieve. In fact, only 35% of the networks develop activities in more than one area (67 out of 167). Here, context matters. Within centralized countries, LGNs tend to be able to operate on a stringent set of policy areas, as most policy issues remain a prerogative of national governments. Portuguese LGNs tend to be focused on social and youth policy (14.7%); environment (8.9%); and The negative correlation tends to suggest that dealing with complexity and avoiding the fragmentation of issues between specialized policy areas require greater commitment from local stakeholders, which may be difficult to achieve. In fact, only 35% of the networks develop activities in more than one area (67 out of 167). Here, context matters. Within centralized countries, LGNs tend to be able to operate on a stringent set of policy areas, as most policy issues remain a prerogative of national governments. Portuguese LGNs tend to be focused on social and youth policy (14.7%); environment (8.9%); and culture (8.1%). Less than 1% of LGNs tend to be involved in research, citizenship, and taxes and fees, policies that tend to remain in the hands of national executives. Most importantly, LGNs ignited by national governments tend to deal with a significantly lower level of policy areas. A Kruskal-Wallis H test showed that there was a statistically significant difference in the average number of policy areas in LGNs (χ 2 (2) = 20.628, p = 0.0001), with networks promoted at the national level dealing with 2.6 policy areas, while those stimulated by local stakeholders tend to deal with 4.2 policy areas.
Additionally, the drivers of LGNs can also impact the diversity of participants. Again, national governments' directives appear to have a negative impact on the involvement of a diversity of actors (Figure 3). While these differences are not statistically significant, the average number of participants in LGNs is higher in locally driven LGNs (mean score of 30.8 participants). On the contrary, those stimulated by national governments depict an average of 26.2 participants per LGN. culture (8.1%). Less than 1% of LGNs tend to be involved in research, citizenship, and taxes and fees, policies that tend to remain in the hands of national executives. Most importantly, LGNs ignited by national governments tend to deal with a significantly lower level of policy areas. A Kruskal-Wallis H test showed that there was a statistically significant difference in the average number of policy areas in LGNs (χ 2 (2) = 20.628, p = 0.0001), with networks promoted at the national level dealing with 2.6 policy areas, while those stimulated by local stakeholders tend to deal with 4.2 policy areas.
Additionally, the drivers of LGNs can also impact the diversity of participants. Again, national governments' directives appear to have a negative impact on the involvement of a diversity of actors (Figure 3). While these differences are not statistically significant, the average number of participants in LGNs is higher in locally driven LGNs (mean score of 30.8 participants). On the contrary, those stimulated by national governments depict an average of 26.2 participants per LGN. In terms of the type of participants involved, results suggest that LGNs tend to be pervaded by public actors (58.9%), as presented in Figure 4. LGNs are also comprised of TSOs and business and industry associations, representing, respectively, 22.4% and 6.9% of the total number of stakeholders. Citizen groups and local personalities are scarcely represented, with only 3% of the total universe of actors (n = 4450). In parallel, higher education institutions represent only 1.3% of the identified stakeholders. Concerning the distribution of types of stakeholders between the two groups (low and high population density), public actors are predominantly present in both, having similar distribution also for the remaining categories of stakeholders, with no significant differences.  In terms of the type of participants involved, results suggest that LGNs tend to be pervaded by public actors (58.9%), as presented in Figure 4. LGNs are also comprised of TSOs and business and industry associations, representing, respectively, 22.4% and 6.9% of the total number of stakeholders. Citizen groups and local personalities are scarcely represented, with only 3% of the total universe of actors (n = 4450). In parallel, higher education institutions represent only 1.3% of the identified stakeholders. Concerning the distribution of types of stakeholders between the two groups (low and high population density), public actors are predominantly present in both, having similar distribution also for the remaining categories of stakeholders, with no significant differences. Soc. Sci. 2023, 12, x FOR PEER REVIEW 9 of 14 culture (8.1%). Less than 1% of LGNs tend to be involved in research, citizenship, and taxes and fees, policies that tend to remain in the hands of national executives. Most importantly, LGNs ignited by national governments tend to deal with a significantly lower level of policy areas. A Kruskal-Wallis H test showed that there was a statistically significant difference in the average number of policy areas in LGNs (χ 2 (2) = 20.628, p = 0.0001), with networks promoted at the national level dealing with 2.6 policy areas, while those stimulated by local stakeholders tend to deal with 4.2 policy areas. Additionally, the drivers of LGNs can also impact the diversity of participants. Again, national governments' directives appear to have a negative impact on the involvement of a diversity of actors (Figure 3). While these differences are not statistically significant, the average number of participants in LGNs is higher in locally driven LGNs (mean score of 30.8 participants). On the contrary, those stimulated by national governments depict an average of 26.2 participants per LGN. In terms of the type of participants involved, results suggest that LGNs tend to be pervaded by public actors (58.9%), as presented in Figure 4. LGNs are also comprised of TSOs and business and industry associations, representing, respectively, 22.4% and 6.9% of the total number of stakeholders. Citizen groups and local personalities are scarcely represented, with only 3% of the total universe of actors (n = 4450). In parallel, higher education institutions represent only 1.3% of the identified stakeholders. Concerning the distribution of types of stakeholders between the two groups (low and high population density), public actors are predominantly present in both, having similar distribution also for the remaining categories of stakeholders, with no significant differences.  As mentioned before, intensity was measured through the frequency of meetings. As depicted in Figure 5, interactions occur more frequently within LGNs stimulated by national governments. A Kruskal-Wallis H test revealed that there is a statistically signif-icant difference in the average number of meetings in LGNs (χ 2 (2) = 35.541, p = 0.0001), considering the initiator of networks. While bottom-up approaches may stimulate the involvement of a higher number of participants, the intensity of their interaction is lower, at least formally. In fact, results of the Pearson correlation indicated that there was a significant negative association between the number of participants and the intensity of interaction within their local governance arrangement (r(112) = −0.60, p = 0.012). This result may be related to the fact that networks created by national governments do not necessarily include network members who are involved willingly or by the fact that these networks deal with more complex issues, hence the need for network members to meet more often. Soc. Sci. 2023, 12, x FOR PEER REVIEW 10 of 14 As mentioned before, intensity was measured through the frequency of meetings. As depicted in Figure 5, interactions occur more frequently within LGNs stimulated by national governments. A Kruskal-Wallis H test revealed that there is a statistically significant difference in the average number of meetings in LGNs (χ 2 (2) = 35.541, p = 0.0001), considering the initiator of networks. While bottom-up approaches may stimulate the involvement of a higher number of participants, the intensity of their interaction is lower, at least formally. In fact, results of the Pearson correlation indicated that there was a significant negative association between the number of participants and the intensity of interaction within their local governance arrangement (r(112) = −0.60, p = 0.012). This result may be related to the fact that networks created by national governments do not necessarily include network members who are involved willingly or by the fact that these networks deal with more complex issues, hence the need for network members to meet more often. Imposed governance structures also generate less positive incentives in terms of coordination. Imposed-driven networks tend to remove the incentives or the possibilities for other relevant actors to be involved in coordination tasks, as Table 4 demonstrates. Overall, concerning the distribution of networks according to the structure of coordination, local government hold the coordination of 58,7% of the total identified networks, while 26,4% are assumed by a specific structure of coordination. There are also 14,4% of networks led by other local actors. Source: own production.
Centrally driven LGNs and local governance structures boosted by municipalities tend to generate coordination structures dominated by local governments. In such scenarios, local governments assume the role of network leaders considering its' dominant position both in controlling relevant flows of information and access to key resources. Imposed governance structures also generate less positive incentives in terms of coordination. Imposed-driven networks tend to remove the incentives or the possibilities for other relevant actors to be involved in coordination tasks, as Table 4 demonstrates. Overall, concerning the distribution of networks according to the structure of coordination, local government hold the coordination of 58.7% of the total identified networks, while 26.4% are assumed by a specific structure of coordination. There are also 14.4% of networks led by other local actors. Source: own production.
Centrally driven LGNs and local governance structures boosted by municipalities tend to generate coordination structures dominated by local governments. In such scenarios, local governments assume the role of network leaders considering its' dominant position both in controlling relevant flows of information and access to key resources. Whereas the national degree level alone would suggest lower commitment of local stakeholders to be involved in such LGNs' coordination and management structures, the existence of municipal regulations and the adaptation of national directives suggest government leading networks are replicated at the local level. This result is not surprising as local stakeholders other than municipalities often do not have the organizational capacity to play coordination tasks. As explained before, mayors and councilors play a pivotal role in local governance (Tavares et al. 2018).
Coordination, however, has a bearing on other relevant dimensions. First, the centralization of management structures in the hands of political and public actors may discourage other stakeholders' participation, a potential further explanation regarding the lower levels of actors' diversity, as aforementioned. On the contrary, voluntary LGNs generate higher predisposition to other coordination strategies, allowing more participated processes, diversity of stakeholders, and, potentially, more flexible approaches to problem-solving, positively impacting motivation and willingness to move forward to new challenges. Local government leading networks tend to generate significantly lower levels of participants (lower diversity). There was a statistically significant difference between groups as determined by one-way ANOVA (F(3.163) = 2.82, p = 0.004). A Tukey post-hoc test revealed that the number of stakeholders involved tends to be statistically significantly higher in the group of LGNs that are coordinated by other local stakeholders, particularly when compared to networks coordinated by municipalities (15.04 ± 5.34 packages, p = 0.02). There are no statistically significant differences in the other coordination strategies and the number of actors involved.
Second, local governments (that is, mayors and councilors) leading networks also tend to be devoted to a lower number of policy areas. On average, LGNs coordinated by specific management structures deal with 6.7 (SD = 4.7), a considerably higher number of policy areas when compared to LGNs coordinated by other stakeholders (M = 2.96; SD = 2.6) or by local governments (M = 1.2; SD = 0.76). Coordination seems to impact the capacity of LGNs to embrace complexity, with local government coordinated networks less likely to address a wide variety of tasks and activities. On the contrary, when coordination is put in the hands of (external) management structures, LGNs tend to be more predisposed to target cross-cutting problems. This is often associated with a higher commitment of actors involved, which may be crucial to increase the efficiency of such networks.

Concluding Remarks
LGNs are increasingly seen as the big idea to cope with issues that are complex enough in scope and scale to require a diversity of expertise and resources (Pope and Lewis 2008). Although the large bulk of research has focused on the benefits to be gained from exploiting LGNs, and hence scale economies, it has been largely confined to specific and institutionalized LGNs. This article departs from the recognition that ignition of such networks can produce relevant differences in terms of the diversity, intensity, and coordination of such local arrangements. This article sough to map the range of LGNs in a centralized country and assess the implication of different drivers in local government arrangements. These can range from national government-imposed rules (top-down arrangements) to collaborative endeavors that emerge to enhance local and regional development (bottom-up).
Overall, this research demonstrated that even in centralized countries, LGNs managed to pave their way, demonstrating high levels of vitality both in terms of the number and typology of actors involved and regarding the policy areas covered. While local governments are frequently regarded as the pivotal actors within these LGNs, central government directives emerge as crucial in igniting LGNs, particularly in smaller and less densely populated municipalities. The role of national governments, however, emerges as a significant burden both in terms of the number of policy areas tackled by LGNs and the number of local stakeholders involved. Moreover, the lower engagement of a wide variety of local stakeholders in terms of the management structures, and instead the tendency to rely on local governments for the management of such arrangements, potentially hinders the commitment of other stakeholders. In the long run, the centralization of management structures in the hands of political and public actors may discourage other stakeholders' participation.
This effect of centrally driven networks is potentially counterbalanced by the positive effect of national governments in triggering the intensity of LGNs. The intensity of interactions of centrally driven networks tend to be higher. Potentially, national governments seem to be aware of the need to enhance mutual trust amongst local stakeholders, reducing transaction costs and promoting a closer involvement in the collaboration (Feiock and Scholz 2009;Hawkins 2010).
Local governance arrangements may be a brave new world, where all relevant stakeholders strive to ensure that the denser institutional fabric is able to effectively work. There is, however, a brave new world for academics as well, both in terms of the theoretical and empirical endeavors. While the centralization of power emerged as the relevant variable for the case selection, it becomes of paramount relevance to map and characterize the inner workings of LGNs in contexts with different settings. Hence, a relevant future avenue of research pertains to the analysis of the capacity of decentralized systems to enable the participation of a set of diverse local political and social structures. Moreover, by focusing on institutional and external aspects of cooperation, through the formation of governance arrangements at a local (municipal) scale, crucial aspects concerning the roles of local leaderships are largely disregarded. Mayors carry a decisive weight in most part of the processes of forging links between LGNs, and promoting effective orientation to local politics, particularly in the case of highly centralized countries.
Funding: This research was funded by the Operational Programme Competitiveness and Internationalisation 2020 (POCI, 2020) and the Portuguese Agency for Science and Technology (Fundação para a Ciência e Tecnologia), grant number POCI-01-0145-FEDER-032502.

Data Availability Statement:
The data are not publicly available.

Conflicts of Interest:
The authors declare no conflict of interest.

1
This research was developed under the frame of the research project "DECIDE: Decentralized Territorial Governance-coordination, capacity and accountability in local governance arrangements in complex regional settings"-http://decide.web.ua.pt/ (accessed on 17 April 2023). 2 As mentioned in the website from the Eurostat (n.d.), NUTS stands for Nomenclature of Territorial Units for Statistics, which ". . . is a hierarchical system for dividing up the economic territory of the EU and the UK". In Portugal, there are three NUTS 1 (Portugal mainland and the Autonomous Regions of Madeira and Azores), 7 NUTS 2 (5 regions in the mainland Portugal-North, Centre, Lisbon Metropolitan Area, Alentejo and Algarve-and the 2 mentioned autonomous regions), 25 NUTS III (clusters of municipalities), and 308 NUTS IV (the existing municipalities).