The potential of trans‐local policy networks for contributing to sustainable food systems—The Dutch City Deal: Food on the Urban Agenda

To foster more sustainable food systems, collaboration between local governments for knowledge exchange and cooperation is essential. Trans‐local food policy networks potentially serve this purpose but their functioning and outcomes remain largely unexplored. We address this gap by analyzing collaboration and its outcomes for one of the first trans‐local food policy networks in the Netherlands: City Deal: Food on the Urban Agenda. We use Ansell and Gash's collaborative governance model as an ideal type to analyze the City Deal drawing on two rounds of semi‐structured interviews with civil servants and politicians in 2016 and 2019 resulting in a total of 37 interviews with 49 unique respondents. The collaborative process was a continuous searching and negotiating for roles, goals, and activities, on the one hand, combined with great eagerness among participants to collaborate and improve local food systems on the other. Although this process led to collective identity building and learning, it resulted in limited collaborative action between participants or tangible results on the ground. The main outcomes were the active network itself, which fostered the strengthening of connections, exchanging knowledge, learning, and agenda setting. Based on our findings, we identify five key points of attention for successful food policy collaboration: ensuring stakeholder commitment, striking a balance between a sectoral and holistic focus, avoiding too abstract ambitions, fostering interdependence, and investing in political commitment.


INTRODUCTION
Local governments around the world are increasingly developing food policies to foster more sustainable food systems and tackle issues like food waste, food insecurity, and obesity through a systemic approach. To this end, they col- Lang et al., 2005;McCartan & Palermo, 2017;Moragues-Faus, 2017;Moragues-Faus & Morgan, 2015;Scherb et al., 2012;Schiff, 2008) or have studied LFPGs (Koopmans et al., 2017;Moragues-Faus & Carroll, 2018;Reed et al., 2018) or networks of multiple LFPGs-trans-local food policy networks-between countries (Santo & Moragues-Faus, 2018). Contrary to the interest in local food policy groups, so far, little attention has been paid to collaborative networks and processes between local governments within trans-local food policy networks for enhancing local food systems. Collaboration between stakeholders is key for exchanging knowledge (Halliday et al., 2019. Moreover, collaboration between local governments is necessary as they are constrained by higherlevel policies (Clancy, 2012(Clancy, , 2014 and therefore need to join forces. This study therefore aims to explore how one such network functions and develops over time and what factors determine if the network leads to genuine collaboration for enhancing local food systems. In this paper, we therefore conduct an in-depth case study of a pioneering trans-local network for enhancing local food systems: the Dutch City Deal: Food on the Urban Agenda (hereafter called City Deal). We aim to contribute to the literature on local food policy by answering the question: To what extent did the City Deal lead to genuine collaboration for enhancing local food systems, and what stimulated and constrained this collaboration?
The City Deal is a Dutch network between 12 local governments, one province, three ministries, 1 and nongovernmental stakeholders intending to contribute to safe, healthy, ecologically sustainable, robust, and accessible food systems in and around cities (Citydeal Voedsel op de Stedelijke Agenda, 2017). For this study, we view the City Deal as a trans-local arrangement intended to foster collaborative governance. A trans-local governance arrangement is an arrangement in which several local or regional governments collaborate. In collaborative governance, multiple stakeholders engage in consensus-oriented decision making (Ansell & Gash, 2008) and focus on achieving policy goals together (Blomgren Bingham, 2011). Its core premise is that higher degrees of collaboration may result in more effective governance of complex societal problems such as unsustainable food systems (van Buuren & Edelenbos, 2007). We used Ansell and Gash's (2008) collaborative governance model to analyze both the City Deal's collaborative governance process and outcomes.
We structured the remainder of the paper as follows. We first review the literature on collaborative governance and specifically elaborate on Ansell and Gash's (2008) model. 1 The municipalities of Almere, Amsterdam, Den Bosch, Den Haag, Ede, Groningen, Helmond, Leeuwarden, Oss, Rotterdam, Utrecht and Venlo, the province of Gelderland, and the ministries of Economic Affairs (later Agriculture, Nature and Food Quality); the Interior and Kingdom Relations and; Health, Welfare and Sport.

Core Ideas
• Collaboration between local governments is key for effective local food policy • The factors that constrain and stimulate trans-local food policy collaboration are unknown • Food policy collaboration is about identity building and learning not collective action • Striking a balance between a sectoral and holistic focus proves key for collaboration • Lack of shared understanding and political commitment can constrain collaboration Second, we explain how we applied this model to analyze collaboration in the City Deal case and sketch the context of the case. Third, we present our findings on the process and outcomes of the City Deal. We subsequently elaborate on factors that seem to stimulate or constrain food policy collaboration and on the potential contributions of trans-local food policy networks for enhancing local food systems.

COLLABORATIVE GOVERNANCE
In this study, we consider the City Deal a trans-local network intended to foster collaborative governance for enhancing local food systems. A trans-local food policy network is a network between local or regional governments, typically within one country, with the aim of facilitating peer-to-peer learning, building capacity, supporting research and evaluation, and potentially enabling collective action (Santo & Moragues-Faus, 2018). Collaborative governance is a governance mode in which multiple stakeholders engage in consensus-oriented decision making (Ansell & Gash, 2008). It distinguishes itself from more traditional modes of governance by its focus on the process of achieving policy goals together (Blomgren Bingham, 2011) through discussions, cooperation, collaboration, and consensus building (Gibson, 2014). The premise of collaborative governance is that it leads to increased legitimacy of public policies, a more diverse range of solutions, more flexible policies that are better suited to changed circumstances, and to the acceleration of the policy process (van Buuren & Edelenbos, 2007). This is because successful collaborative governance stimulates inclusion and participation, from which the policy process benefits in two ways: diverse stakeholders' expertise, resources, and support are included (Emerson & Nabatchi, 2015a) and more agreement, trust, and understanding between stakeholders is realized (Ansell & Gash, 2008). At the same time, many initiatives intended as collaborative governance arrangements often do not seem to develop active collaboration or develop collaboration only to a limited extent (Bryson et al., 2006). They face three main challenges: speed of the process, contested legitimacy, and hesitancy by government to change (Gibson, 2014). Sjöblom and Anderssson (2018) warn, for example, that "collaborative governance can contribute to diversity and fragmentation because of competing values and interests among involved actors." Collaborative governance approaches are typically used for addressing so-called 'wicked problems' (Rittel & Webber, 1973), that is, problems for which existing policy infrastructure is insufficient (Ansell & Gash, 2008) such as natural resource management (e.g. Koontz & Thomas, 2006;Memon & Weber, 2010;Taylor et al., 2013) and conflict resolution (e.g. Costantino & Merchant, 1996). Enhancing local food systems in and around cities is one such wicked problem. Enhancing local food systems requires integrated food policy, which acknowledges the multifaceted and interrelated nature of food challenges and addresses these in a concerted manner (Mendes, 2007;Sibbing et al., 2019) thereby integrating health, environmental, social, and economic dimensions (Lang et al., 2009;Moragues-Faus & Morgan, 2015). Authors have studied several collaborative arrangements that have been developed to achieve improved food systems such as local food policy groups (LFPGs) (Halliday, 2015) including food policy councils (Schiff, 2008;Siddiki et al., 2015;Koski et al., 2018) and trans-local food policy networks (Santo & Moragues-Faus, 2018).
Several analytical frameworks for successful collaborative governance have been developed (e.g. Bryson et al., 2006;Thomson & Perry, 2006;Ansell & Gash, 2008;Provan & Kenis, 2008;Emerson et al., 2012). As the model of Ansell and Gash (2008) has proven the most influential (Batory & Svensson, 2019), we use this model as our theoretical framework and follow Ansell and Gash's (2008) definition of collaborative governance as: "A governing arrangement where one or more public agencies directly engage non-state stakeholders in a collective decision-making process that is formal, consensus-oriented, and deliberative and that aims to make or implement public policy or manage public programs or assets." This definition is suitable for our case, as the food policy network in our case is initiated by public agencies. Ansell and Gash (2008) found that the process of collaborative governance is cyclical rather than linear and can be interpreted as a (simplified) cycle of face-to-face-dialogue, trust building, commitment, shared understanding, and intermediate outcomes. Face-to-face communication between stakeholders is necessary for identifying mutual gain together (Ansell & Gash, 2008). Building trust is considered key, especially when stakeholders start with a lack of it (Ansell & Gash, 2008).
The level of stakeholders' commitment is considered critical for the success of the collaboration. Commitment is considered key for developing mutual recognition of interdependence, ownership of the process, and openness to exploring mutual gains (Ansell & Gash, 2008). Shared understanding then ideally develops when stakeholders determine what they can and want to collectively achieve (Ansell & Gash, 2008). Lastly, intermediate outcomes are considered key for building momentum. They can feedback into the cycle of trust building and commitment, thus encouraging the collaborative process (Ansell & Gash, 2008).
Three factors are considered to be of particular importance to constraining or enhancing the collaborative governance process: the starting conditions, the institutional design, and the facilitative leadership (Ansell & Gash, 2008). The starting conditions comprise three key conditions at the start of the process: the differences in resources such as knowledge and finances that participants possess, the incentives and constraints on participation participants have, and the prehistory of cooperation or conflict participants share (Ansell & Gash, 2008). Second, for a successful collaborative process, the institutional design of the collaborative arrangement needs to be participatory inclusive, be exclusive as a forum, have clear ground rules, and include a transparent process (Ansell & Gash, 2008). Third, strong facilitative leadership is necessary, as it ensures setting and maintaining of ground rules, empowering weaker participants, facilitating trust building and dialogue, and exploring mutual gains.
The collaborative process typically leads to outcomes. Ansell and Gash's (2008) model falls short on explaining what the outcomes of the collaborative governance process entail. A common distinction though is between immediate, intermediate, and long-term outcomes (Bryson et al., 2015) or what Emerson and Nabatchi (2015b) call outputs, outcomes, and adaptation. Also, outcomes can be processual or content based. Processual outcomes are the outcomes that unintendedly result from the collaboration process, while content outcomes were anticipated and motivated the collaboration (Seitanidi, 2010, 121). In this study, we compared respondents' perceived outcomes and related these to the initial goals of the City Deal. To conclude, Figure 1 summarizes Ansell and Gash's (2008) collaborative governance model.

METHODS
Our study is a case study in which we compared the City Deal's collaborative governance process and its outcomes to Ansell and Gash's (2008) collaborative governance model to explore key aspects and influencing factors of collaboration in trans-local food policy networks and the main outcomes of trans-local food policy networks. We based our assessment F I G U R E 1 The collaborative governance model (Ansell & Gash, 2008) on the City Deal participants' reflections, comparing their expectations before the start of the City Deal to their reflections 3 yr later. Participants were the civil servants who represented their administrations in the City Deal. They were predominantly public policy makers and project leaders in the fields of health, economy, sustainability, or general strategy. We selected the City Deal as our case, as it is one of the first trans-local food policy networks in continental Europe and one of the first national spin offs of the Milan Urban Food Policy Pact (MUFPP), a global pact to enhance local food systems currently signed by more than 200 cities (https: //www.milanurbanfoodpolicypact.org). This makes the City Deal a unique and therefore suitable case. To reconstruct the collaborative governance process and its outcomes, we conducted two semi-structured interview rounds and consulted field notes, reports, and press releases. Before the official start of the City Deal, we conducted interviews (n = 18) with participating civil servants about motivations for participation in, goals for, and expectations of the City Deal. Three years later, before the official ending of the City Deal (in 2019), we conducted interviews (n = 15) with the civil servants again about the collaborative process and the outcomes of it. Each interview was held with the one or two involved civil servants from the participating organization. In the majority of cases, these were the same civil servants as interviewed in 2016. In one case, the interview was held with three civil servants. In 2019, we also interviewed three actively involved politicians from participating local governments and the former program manager to gain a broader perspective. This resulted in a total of 37 interviews with 49 respondents. For all interviews, we used an interview guide (see Appendix A), based on Ansell and Gash's (2008) collaborative governance model. Subsequently, interview transcripts were thematically coded for the corresponding collaborative governance variables and compared with the model for analysis. The main author of this paper was part of the City Deal in her role as a policy adviser for the local government of Ede, one of the participating local governments. This provided her with the opportunity to closely follow the City Deal's process. We acknowledge that while the main author's engagement gives us a unique perspective, it also creates the potential for bias. Therefore, we performed continuous discussions about data collection and interpretation between the authors and organized a final member check with the City Deal participants.

An elaborate preparation
In 2015, two Dutch local governments took the initiative to unite interested local governments, provinces, and ministries, to further explore how food policies could improve food systems in the Netherlands. In 2014, the Dutch Scientific Council for Government Policy had published a report that recommended that the Netherlands adopt a national food policy (Wetenschappelijke Raad voor het Regeringsbeleid, 2014) for achieving a more healthy, sustainable, and resilient food system. However, this report inadequately addressed food policy at the local level and the role of subnational administrations in food policy. The local governments of Den Bosch and Ede, therefore, aimed to start a City Deal; an agreement between a group of stakeholders to address a specific urban challenge-in this case improving local food systemsfor a determined period. City Deals are an instrument of the Dutch Ministry of the Interior to stimulate sustainable social transitions in cities (Scherpenisse et al., 2017). The ministry of the Interior agreed to the City Deal, and in 2016, an extensive collective process was started to prepare for the founding of a City Deal on local food policy as a trans-local governance arrangement: an arrangement between local administrations and other partners to collectively establish a food policy platform, exchange knowledge and experiences, and contribute to and raise awareness about local food policy.
Over almost a year, participating administrations investigated priorities, incentives, and constraints and collectively developed goals, organizational structure, and ground rules. The ultimate group of 12 participating local governments turned out to be diverse. They differed in population size, socioeconomic character, available budget and personnel capacity, and in local food priorities, experience with food policy, and level of political support for food policy. The City Deal included, for example, the Dutch capital Amsterdam, a highly urbanized city facing obesity issues as well as the small, rural local government of Oss facing sustainabilityrelated agricultural issues. Also, cities like Rotterdam and Ede already had vast experience with food governance and had even adopted integrated food strategies, while cities like Venlo and Helmond were new to the topic. Incentives to participate were relatively similar among participating local governments: establishing a food policy platform, exchanging knowledge and experiences, and contributing to and raising awareness about local food policy. Another widely shared incentive among local governments was to promote their own cities or regions and to get in touch with and lobby the national government for their own interests. They saw the City Deal as an opportunity to advocate for their needs to the ministries and to gain information about the ministries' plans affecting lower administrative levels. Expectations and ambitions among the local governments were high. Among the most-often indicated desired results were tangible results requiring collaboration such as healthier food in hospitals, a European project, or a project on better distribution of value throughout the food chain. The most common constraints local governments faced to participate were lack of time and lack of organizational and political support.
Eventually, the overarching aim, goals, and structure were formalized in a covenant that came to form the base of the City Deal (Citydeal Voedsel op de Stedelijke Agenda, 2017). The main aim was to contribute to safe, healthy, ecologically sustainable, robust, and accessible food systems in and around cities (Citydeal Voedsel op de Stedelijke Agenda, 2017). Three goals were pursued: 1. Establishing a platform for knowledge exchange and collaboration on food policy. 2. Investigating the changing role of the government toward a more integrated and more interactive governance approach. 3. Identifying best practices on food policy, sharing these internationally, and learning from other countries' examples (Citydeal Voedsel op de Stedelijke Agenda, 2017).
The organizational structure comprised a coordinating team and four working groups. The four working groups include the governance innovation group, which is a higherlevel, cross-cutting group that ensures a holistic approach is adopted, and three thematic working groups: (a) ecological and economic innovation; (b) regional food systems and fair and short supply chains; and (c) food education, health, and social inclusion (Figure 2). Each working group had two coordinating administrations assigned and included all administrations interested in the addressed topic, which resulted in each administration joining one or two working groups. To connect the entire group, all administrations joined the governance innovation group by default. Each local government made a financial contribution between €5,000 and €10,000 to start the City Deal. With this budget, a program manager and secretary would be hired, plenary sessions would be organized, and projects proposed by the working groups would be facilitated.
Despite a shared enthusiasm to exchange experiences, there was a notable difference in expectations and commitments from the start. Only later in the process did it became clear that these differences had resulted in broad and, therefore, unclear goals being set at the start.

An enthusiastic start
The official signing of the covenant in 2017, marked the launch of the City Deal. For practical reasons, the covenant was only signed by the participating governments. Each administration was to subsequently engage its nongovernment stakeholders, such as farmers, retailers, health care professionals, and nongovernmental organizations, to participate. These nongovernmental stakeholders were, however, hardly engaged because administrators lacked time and capacity. The role of nongovernment stakeholders was therefore reduced to participation in the official gatherings, in which they were visited on excursions, invited to share their knowledge and experience, and invited to join workshops. Although they had already met each other several times in the preparation process, for most participating civil servants, both the group of governments and the issue of improving local food systems were new. They were intrinsically motivated though, as they felt the urgency of the issue, and felt the need to be part of this promising intervention to tackle it. Despite participants' differences, the start of the City Deal was therefore characterized by a strong enthusiasm and commitment among all participants, which kickstarted the collaborative process and remained one of the process' main stimuli.
However, in the starting phase of the City Deal, there was also an event that negatively influenced the course of the City Deal's collaborative process. The City Deal was officially signed by all representative elected officials at a national food summit on the future of the Dutch food system. Only elected officials were allowed to join this top-level meeting, while participating civil servants were not. This was a disappointment for the civil servants, while the meeting itself was similarly disappointing for the invited elected officials. At the summit, the elected officials were restricted to merely listening to ministers and CEOs instead of actively sharing their ideas about local food policy. This event set the tone for the role of the elected officials during the collaborative process of the City Deal; they remained passive during the entire process, while the civil servants ran the network.

A challenging middle
When the working groups started to develop their project plans in 2017, it became clear that underneath the collectively developed main aim of the City Deal, ideas about its meaning and the means to achieve it differed greatly. The City Deal covenant had been an attempt at striking a balance between retaining a holistic food system approach and catering for each organization's priorities, such as healthy food environments or short food chains. When the working groups started to develop more concrete action plans, it turned out that shared understanding existed on an abstract level but not on specific goals and means. Key reasons for this were the differences in context, priorities, and expectations between participants. For example, several local governments were predominantly interested in stimulating their food business sector by attracting new food companies, while others were interested in regulations to foster healthy food environments, perhaps excluding certain retailing or food-selling companies. Several local governments, therefore, found out that the City Deal did not match their expectations and priorities adequately, which in one case even led to a local government leaving the City Deal. For this government, the financial contribution was high, while output in terms of results, such as concrete projects, lagged. The city council therefore did not grant the municipality the budget for the yearly City Deal contribution. The government's policy officers subsequently decided that without political back-up and financial resources it would not be feasible to keep participating in the City Deal. In addition, ambitions in the covenant were manifold, both with regards to content-a range of issues covering the entire food system-as well as process-exchanging knowledge, lobbying, operating internationally, mapping best practices, conducting research, and agenda setting. As a respondent explains, This made it difficult to translate objectives into actions, both for the general and the working group goals. It, therefore, became challenging to develop shared understanding, which came to constrain the collaborative process.
At the same time, it also became apparent that participants did not feel strongly dependent on each other to achieve their goals. On an abstract level, participants perceived collaboration between all administrative levels to be necessary for improving local food systems. On an operational level though, achieving mutual recognition of interdependence was challenging. First, improving local food systems was a rather unfamiliar challenge that differed per city and for which no clear recipe existed. Second, the City Deal was a self-imposed, voluntary challenge with no major financial or legally binding consequences. Participants, therefore, did not depend on each other in the way they do when they need to construct a road that crosses various local governments. As a result, cities invested their energy in what would certainly render benefits; they focused on achieving individual objectives and on positioning their city or region instead of exploring how they could join forces. One respondent remarked: "But you just saw that people chose their own thing. And subsequently, that the network was meant to take things on collectively, but that people used the network to strengthen their own position. [. . . ] It would have been nice if we, as City Deal cities, in the collaboration with the ministries, would have collectively made a proposition for the challenges that the ministries discussed in the IBP [network]. That was of course the essence of why you collaborate. And if you would have then said like: 'We define ourselves as cities or regions and distinguish ourselves in a specific way,' instead of going for your own goals." (Interviewee 18, 2019) This lack of mutual recognition of interdependence, therefore, led to a focus on individual priorities, which inhibited collaboration.
At this point, it also became clear that the lack of commitment of elected officials constrained the collaborative process, while the lack of nongovernment stakeholders had a positive influence. Elected officials had come to merely fulfil an accountability role, receiving a yearly update on the City Deal's progress. Civil servants lacked time and capacity to engage and empower their elected officials. Commitment among elected officials therefore decreased during the collaborative process. This was aggravated by three rounds of elections in the City Deal hierarchy over 3 yr, which resulted in changes in the political context. One respondent explains how this happened in his local government: The absence of elected officials reduced the network's impact as actions typical for the role of elected officials, for example, lobbying for changes in laws and regulations was not performed. On the other hand, the lack of involvement of nongovernment stakeholders was not considered a problem among participants. Quite the reverse, the government-only character of the network was key to its exclusiveness. Participants appreciated that, different from the many existing food networks and groups, the City Deal was tailored specifically toward governments and facilitated them meeting their governmental peers to exchange and collaborate. Having a network mainly for civil servants, without elected officials and nongovernment stakeholders, thus had both negative and positive consequences.
After the first year, the City Deal's general leaders changed several times, which led to unstable leadership. This resulted in irregular and unclear communication that weakened the bond between participants. Participants, therefore, became disconnected from the process, and weaker stakeholders were empowered less and represented less. It also resulted in dayto-day organization being challenging and failed to mitigate competition between local governments. However, the City Deal leadership did not depend on one leader. The coordinating team helped to mitigate the constraining effect of the changing leaders as it safeguarded stability. This structure also ensured legitimacy, as responsibilities were shared and leaders were representatives from within the group. The structure of a coordinating team in the institutional design therefore strengthened the City Deal.
Despite the challenges, intermediate outcomes and face-toface dialogue helped to build and maintain an active network. The City Deal succeeded in producing a wide range of intermediate outcomes, which stimulated the collaborative governance process as they fueled enthusiasm. These intermediate outcomes ranged from lectures with experts, a video clip, excursions to good-practice examples, an online recipe book with good practices in food governance, to an opinion piece in a national newspaper. Another type of intermediate outcomes were the City Deal's meetings, in which face-to-face dialogue took place. Plenary meetings contributed to the collaborative process because they were a mix of inspiration, substantive lessons, and the possibility of knowledge exchange and networking and they served to build trust between participants over the years. Participants met twice a year with the entire group in a plenary session and several additional times per year in the working groups. Participants found the plenary sessions well organized and greatly appreciated the opportunity of directly connecting with their peers that these sessions offered. The working group meetings and excursions served a slightly different purpose. They inspired participants as they provided them with in-depth experience on topics such as a healthy food environment, sustainable food procurement, or shortening food chains. Because of the meetings and intermediate outcomes, participants managed to build and maintain an active network despite constraining factors in the collaborative process.
Another key stimulus in the City Deal's collaborative process was its institutional design, which set clear ground rules and made the City Deal an exclusive forum. One of these rules was, for example, the division across four thematic groups. Each city was to choose between two or three thematic groups to participate in, and, for each group, two cities were designated as its daily coordinators. The covenant was key for making the City Deal's ground rules clear, as it defined both rules and players of the game and served to get all players on the same page at the start. The particular strength of the covenant was that almost a year was invested to collectively develop it before starting the network officially, which resulted in the covenant being widely supported. The City Deal was attractive to participants, as it distinguished itself from the wide range of existing Dutch food networks in three ways: by focusing on public governance, uniting all administrative levels, and utilizing a window of opportunity. The focus on public governance and multi-levelness, meant there was finally a platform where government participants interested in food policy-as opposed to platforms for citizens or entrepreneurs-could exchange ideas with their peers and receive information tailored specifically to them. The City Deal's exclusivity was further increased by using a window of opportunity. Not only did the publication of the national report raise awareness on food policy in 2015 but the MUFPP had also been signed, and the City Deal fulfilled the need of Dutch MUFPP signatory local governments to implement the MUFPP.

An end with unexpected outcomes
Toward the end of the City Deal, in 2019, differences between participants started to play out more and two groups formed. The first group can be characterized as an active, close, core group, of which, members generally had more social and economic capital (bigger cities and the higher government levels), while the second group consisted of participants that did not have the time and capacity to participate regularly and therefore 'slipped away,' being mostly smaller local governments. An exception was Ede. Though not a large city, Ede belonged to the active group, as Ede has put considerable effort into food governance, which provided Ede with the capital to play an active role in the City Deal. A key explanation for the split was that leadership in the City Deal became unstable after the first year, which resulted in too little facilitation and empowerment of 'weaker' participants. Still, the majority of participants were content with the City Deal. The majority of participants wanted to continue collaborating in one way or another. Some wanted more of a knowledge network where they would meet once a year, while others wanted an active lobby group that would take on pilot projects together. In the end, discussions were going on about how to continue the City Deal. During the 3 yr of the City Deal, it's collaborative governance process predominantly led to processual outcomes, while it hardly led to tangible outcomes. According to participants, the main achievement of the collaborative process was managing to turn the network into an active platform under challenging circumstances. As one respondent points out: "If I think of how difficult it is to realize transitions and of how difficult it is to get new themes-that might not directly lie in the heart of the administrative-political responsibilityon the agenda within organizations, then I think this story [the City Deal] has gotten pretty far." (Interviewee 15, 2019) In other words, the network served to get the right players to the table and keep them at the table. This active network served to identify motivated local governments, connect all administrative layers, and strengthen connections. It served to identify local food policy best practices and facilitate the exchange of food policy knowledge and experiences both among local governments and between multiple administrative levels. It facilitated the development of a joint vision on the scope and content of municipal food policy and facilitated agenda setting on the need for local food policy both within participating organizations and in the Netherlands in general. However, though the City Deal led to collective learning and exchanging, genuinely laboring together (co-laboring) happened much less. This means the joint undertaking of activities such as lobbying and working on producing concrete, tangible results on the ground failed to materialize. The collaborative process, therefore, produced few tangible content outcomes such as joint pilots, campaigns, or adapted legislation.
A respondent says about the achievement of the City Deal: "It feels like we have done a lot of preparing and that the real 'scoring the goal' still has to happen." (Interviewee 12, 2019) Table 1 summarizes the key developments and enabling and constraining factors in the City Deal's collaborative governance process.

DISCUSSION
In this study, we aimed to explore what collaboration and its outcomes looked like in the City Deal and what stimulated and constrained this collaboration. Three findings stand out as particularly relevant beyond the City Deal case. First, based on our findings, it seems that collaboration in the City Deal was about collective identity building and learning rather than about collective working. The collaborative process was about identifying motivated adminis- trations, strengthening connections between administrations, exchanging knowledge and experience, learning about and developing a vision on local food policy, and agenda setting for local food policy. Collective working, for example, in the form of joint lobbying, and concrete, tangible results, such as joint pilots, campaigns, or adapted legislation or regulations, proved a lot harder to achieve. These findings underpin findings from other authors, who found that both trans-local food policy networks (Santo & Moragues-Faus, 2018) and food policy councils (Schiff, 2008) serve for agenda setting, connecting, and building capital. Agenda setting includes voicing the need for system-wide changes in food governance (Schiff, 2008) and normalizing the integration of food into municipal governments' agendas (Santo & Moragues-Faus, 2018). Connecting includes fulfilling a network role across the spectrum of food system interests (Schiff, 2008), serving as facilitators in the networking and implementation capacity of other organizations (Schiff, 2008), and reducing feelings of isolation among stakeholders. Capital building includes building credibility and capacity within local governments before formulating policies (Schiff, 2008), building capacity to bring diverse voices together to deliberate and identify collective goals (Santo & Moragues-Faus, 2018), and providing legitimacy to stakeholders' efforts. Schiff (2008) summarizes this as food policy councils in general focusing more attention on programmatic and project work rather than policy work. Outcomes of the City Deal, therefore, were accordingly predominantly processual while lacking tangible outcomes on the ground. The nascent nature of the network might be the reason for these processual outcomes, so-called 'collaborative actions' or 'outputs' (Emerson & Nabatchi, 2015b), as age is positively correlated to a network's results (Leach et al., 2002). Food policy is relatively new to local governments and, logically, preparatory work needs to be done first. In the City Deal, most participants did not even know each other, and for many municipal civil servants, it was their first time to formally collaborate. The lack of collaborative actions and tangible outputs might also be a result of the City Deal participants having left this ambition inexplicit instead of explicitly adopting it as one of the goals formulated at the start. A last explanation for the predominantly processual outputs might be the absence of nongovernmental stakeholders in the City Deal. Ansell and Gash (2008) stress the participation of different stakeholders for successful collaborative processes, while hesitancy by governments to change has been found to be an inhibiting factor for collaborative action (Gibson, 2014). The absence of nongovernmental stakeholders may therefore limit a food policy network's potential to achieve change and concrete results, as forces to counteract governments' hesitancy to change are missing in such a network. The absence of nongovernmental stakeholders could even lead to groupthink (Janis, 1972), something that might have happened in the City Deal, considering that participants did not perceive the absence of nongovernmental stakeholders as a problem. At the same time, when the urgency to 'do something' is felt among administrations, for example, rapidly achieving tangible results on the ground, there is a risk of 'taking on too much,' resulting in the paralyzing of participants (Termeer et al., 2018), which leads to fewer outcomes. However, processual results might form a first step that allows for collaboration and tangible level-two outcomes and level-three adaptation (Emerson & Nabatchi, 2015b) in the next phase.
Second, we identified two interconnected factors that may be key for fostering collaboration in trans-local food policy networks: ensuring commitment among participants and striking a balance between a sectoral focus and a holistic foodsystems focus. Commitment among participants is key as it is the fuel of the collaborative process. As long as commitment is high, there is a willingness to find solutions together despite obstacles. As the City Deal case has shown, commitment can be stimulated by face-to-face dialogue in plenary meetings. A lesson from the City Deal case is that smaller coalition meetings, instead of meetings with the entire network, might offer more potential. In this way, cities can meet their peers, work on their topics of interest, and be inspired while not having to engage in too generic meetings that may temper their enthusiasm.
For successful food policy collaboration both within and beyond the Netherlands, the City Deal case shows that a right balance should be found between a sectoral focus by letting every participant work on their individual pressing food issues and a systemic focus by sticking to a holistic approach that addresses the entire food system. Too much working on preferred challenges may lead to high commitment but no holistic food policy, while a too-holistic approach may ask too much of participants, lowering the level of attractiveness of the network. It is also important to keep everyone on board so that the networks can keep on functioning. Losing commitment is a key threat to collaborative governance, but when every city focuses on their priorities, the so-important holistic character is lost and the network turns into a collection of siloed policy efforts again. Striking this balance can be done via the institutional design at the start of a food policy network, such as the covenant of the City Deal and its preparatory process, have shown. A strong institutional design that balances commitment and effectiveness is therefore key for a successful food policy network.
Third, we identified two interconnected factors that proved to be constraining in our case: a lack of shared understanding and a lack of political commitment. Lack of shared understanding about the main aim and goals may constrain a food policy network, as it makes the entire collaborative process more challenging. Achieving shared understanding is a common challenge in trans-local food policy networks (Santo & Moragues-Faus, 2018) especially with diverse membership and constituencies (e.g. Harper et al. (2009) and Gore (2010)).
Our findings show that a lack of shared understanding may be caused by imbalances between participants and abstract food ambitions. Imbalances occur as sustainable food transitions in cities are conditioned by their specific urban socioecological configurations and interests (Moragues-Faus & Morgan, 2015), which can differ greatly. Recognizing the diversity of cities and their relation with their food issues in diverse places is, therefore, the first step for successful food system enhancement (Moragues-Faus & Carroll, 2018). Also, abstract ambitions further challenge achieving shared understanding, as they can lead to unclarity about goals, an often encountered problem in food policy groups (e.g. Coplen & Cuneo, 2015;Santo & Moragues-Faus, 2018).
Lack of commitment among elected officials may constrain a food policy network, as this may make it harder for civil servants to represent their organizations and reduces the impact of the network. Others also found the consistent involvement of elected officials and political will (Mendes, 2007) to be crucial for successful trans-local food policy networks (Yeatman, 2003;Mendes, 2007;Sonnino, 2009;Coplen & Cuneo, 2015;Halliday, 2015). A particular risk factor for consistent political commitment is elections, as they may lead to decreasing commitment (Sonnino, 2009;Halliday, 2015) and to prior commitments of elected officials being abandoned (Yeatman, 2003). For the City Deal, elections were a big challenge but other food policy processes also faced this, such as for the food strategy of Victoria (Caraher et al., 2013) and the Toronto Food Strategy (Mah & Thang, 2013). To create support from elected officials, Halliday and Barling (2018) have several recommendations. One example is identifying ways to institutionalize food policy under a supportive official by taking into account, such as the current official's interest and knowledge of food issues, the degree to which the food agenda corresponds to their priorities, and the likelihood of municipal priorities changing (Halliday & Barling, 2018).
More research on trans-local food policy networks should be conducted in the future to gain more insight into their potential to improve food systems both within and beyond the Netherlands. As our conclusions are based on one case only, more networks should be studied to draw broader conclusions for trans-local food policy networks. In addition, it would be valuable to investigate the perspectives of nongovernmental stakeholders in these networks instead of focusing on governmental actors only. Also, studying collaboration through longitudinal research designs seems a promising future avenue to find out how collaboration develops and what results a food policy network produces in the long term. The key question here is do networks also manage to realize impact over time, instead of mere outcomes and output?
Based on our findings, we encourage local food policy makers to participate in collaborative arrangements with their peers as these can foster identity building and learning. Pol-icy makers should keep in mind though that successful collaboration depends on them focusing on shared interests rather than just their own interests. Forming smaller coalitions within the collaboration on specific issues of shared interest (such as food waste) could be a solution to achieve this.

A C K N O W L E D G M E N T S
We would like to thank all respondents for contributing their time. We also want to thank the municipality of Ede, the Netherlands, for facilitating the main author's PhD research, of which this paper is part.