1.1. Post-NPM public action model and regulation logicsn

6 Post-NPM paradigms have been associated with collaborative network regulation (Torfing & Sørensen, 2014; 2015). Actors are coordinated in a decentralized way and their relationships are more horizontal or flattened. The mechanism of coordination is trust, i.e., the belief in mutual reliability (Perrin, 2013). Collaborative networks are hard to implement since paradigms overlap rather than substitute each other (Thynne & Peters, 2015). They layer on top of existing logics and relationships (Agranoff, 2007; Bevir & Rhodes, 2016). Non-collaborative regulation limits actors’ ability to create collaborative networks (Torfing & Sørensen, 2014; Bryson et al., 2015). Firstly, hierarchical regulation is associated with traditional public administration, the bureaucratic organization being its ideal type (Weber, 1958). In this centralized logic, the public actor’s vertical authority ensures the coordination of actors. The framing of government-nonprofit relations emphasizes the asymmetry of power in favor of the state over the nonprofits which it governs (Ramanath & Ebrahim, 2010). As a result, this paradigm offers a passive image of nonprofits implementing the public decision-maker’s policies. This hierarchical logic is challenged by NPM. Yet it persists because of the financial dependence of nonprofits on public subsidies (Johnston et al., 2011). Secondly, NPM introduces a logic of market regulation. Organization is decentralized, and the price is the coordinating mechanism. Relationships are based on generalized competition and the principal-agent frame. The role of nonprofits in public action is residual, within the State-market dualism (Thynne & Peters, 2015). The contractual relationship between the public sponsor and the nonprofit provider frames nonprofits as instruments of public agencies (Gazley & Guo, 2020). The persistence of this market logic is linked to the survival of NPM despite critiques (Bezes, 2018). Despite the collaborative lens on government-nonprofit relationships (Camus, 2014), nonprofits still face market competition (Robelet, 2017).

7 Regulation logics are articulated in a complex way. Firstly, the expression “shadow of hierarchy” reflects the ambivalent role of hierarchy in collaboration (Whitehead, 2003; Cram, 2011). On the one hand, it constrains collaboration. Actors attempt to flatten their relationships while hierarchy persists and renews itself through collaboration. Indeed, collaborative networks often give public actors a decisive role in promoting and structuring (Jessop, 1997). Public actors can increase their authority via the collaborative network, by controlling the procedures for selecting, monitoring, funding and sanctioning partners (Whitehead, 2003; Johnston et al., 2011; Chabault & Martineau, 2013). Structuring collaborations often involves transforming the instruments public actors use to manage nonprofit activity and a reproduction of power imbalances (Robelet, 2017). The governmental actor “at the intersection of all connections” (Geddes, 2012, quoted by Suquet et al., 2020, p.12) can hinder collaboration (Suquet et al., 2020). The embeddedness of collaborative networks in hierarchy raises the paradox of hierarchically mandated collaboration (Rodriguez et al., 2007). On the other hand, the shadow of hierarchy means that collaboration is protected by hierarchy (Héritier & Lehmkuhl, 2008). Three benefits of hierarchy for collaboration have been identified (Levelt & Metze, 2014): (1) it encourages actors to collaborate through the control of subsidies by public actors; (2) the accountability of actors involved in collaboration is enhanced because of the public actors’ ability to control and evaluate (Acar et al., 2008); (3) hierarchy brings legitimacy to collaboration. The importance of hierarchy in collaboration is not limited to the emergence of the network (Jessop, 1997). The coordination of actors in networks is based on “self-coordination embedded in structures” (Scharpf, 1994, p.36). Collaborative networks and hierarchy are intertwined, often giving a central position to the public actor (Tenbensel et al., 2011; Nederhand et al., 2016). The challenge for governments is to strike a balance between the democratic legitimacy associated with less hierarchical modes of coordination in collaborative networks, and the highly visible efficiency of coordination based on hierarchy as a source of legal-rational legitimacy (Peters, 2011). Secondly, the articulation of collaboration with the market is addressed in research on the effects of market competition on nonprofits and their relationships with public actors (Maier et al., 2016). NPM promotes generalized competition which is in tension with the creation of trust. Yet, market logic proposes an already decentralized way of organizing (Torfing & Sørensen, 2014). Competition can stimulate collaboration on projects deemed strategic and on the defense of sectoral interests (Irvin, 2010). However, the outcomes are mixed on the longer term: the competitive context also increases the perceived risks of free riding and predation between partners.

1.2. Collaboration as institutional change to move beyond NPM

8 Due to the persistence of non-collaborative regulation logics, previous research has primarily analyzed the institutional environment [1] as a constraint for actors willing to develop collaboration (Bryson et al., 2015). The institutional environment has been described as a source of collaborative inertia (Vangen & Huxham, 2010) leading to disappointing collaborative outcomes and a lot of failures. As collaboration is difficult, the actors may not perceive their organizational interest to engage in it if an alternative exists. As a result, the success of collaboration has been linked to a rare environmental situation that is difficult for actors to have control over (Bryson et al., 2015). The environment also frames the perception of the organizational costs and benefits of collaboration. Firstly, for public actors, the perceived benefits are associated with the legitimacy brought by working with civil society. Their costs are linked to developing network management skills (Salamon & Toepler, 2015; O’Leary et al., 2009) and renewing public managers’ positions [2] (Stoker, 2006; O’Flynn, 2007). Secondly, nonprofits mostly perceive a disadvantage in collaborating (Gazley, 2010). The costs are high and visible at short-term, for example when allocating resources to collaboration instead of the organizational project. A closer relationship with public actors is associated with risks for nonprofits, such as mission drift or weaker political and advocacy capacities. By contrast, the benefits are perceived as uncertain and long-term. Because of this perception, nonprofits have strong incentives not to commit to collaboration (Gazley, 2010).

9 Previous research has overlooked actors’ agency in this constraining environment (Gazley, 2017). Indeed, literatures on CSC and on the transformation of public action paradigms have adopted a macro level of analysis (paradigmatic or sectoral), which results in neglecting the role of actors. The focus on power imbalances (Gazley & Guo, 2020) and on an institutional vision of government-nonprofit relations limits the understanding of actors’ agency (Ramanath & Ebrahim, 2010; Gazley & Guo, 2020). We need to explore how collaboration as a process driven by actors committed to develop a network logic is articulated with the emergence of a post-NPM paradigm (Gazley, 2017). Actors are constrained as the institutional environment frames their action and their perception of collaboration. Yet, they also participate in the (re)production and transformation of this environment. Actors can be reintegrated by conceptualizing collaboration as a process of institutional change (DiMaggio & Powell, 1997) led by actors in an environment that constrains them, which contributes to the transformation of public action.

2.1. An inter-organizational ethnography of a collaboration

10 There have been few empirical studies of CSCs (Gazley, 2017) and implementing post-NPM models (Lo, 2018; Suquet et al., 2020). The existing studies have mainly proposed a retrospective analysis of successful collaborations. This method has a success bias and limits the conceptualization of collaboration as an actor-driven process (Gazley, 2017; Gazley & Guo, 2020) with uncertain outcomes. I have chosen a qualitative approach (Dumez, 2016) as the research question seeks to understand the perceptions and organization of collaborative networks by public and nonprofit actors. Responding to the calls to develop research on collaboration engaging with actors over time (Gazley, 2017), I designed an ethnographic study of collaboration. This method enables immersion in the daily activities of communities within their environment (Atkinson & Hammersley, 1998). Ethnography has a unique ability to capture relational dynamics in progress (Berthod et al., 2018). The new organizational ethnographies (Van Maanen, 2006) represent opportunities to study collaboration as an inter-organizational phenomenon, as a process (organizing) rather than a structure (organization) (Rouleau, 2013). Those methods have great potential for studying transformations of contemporary public management (Cappellaro, 2016) and particularly government-nonprofit collaboration.

2.2. Presentation of the case

11 The selected ethnographic case is that of the Equality Group. It is a collaborative structure, set up from 2018 to 2020 by public actors and nonprofits in a French region. I was involved in the Group from its start thanks to my academic and nonprofit identities. The Group is a collaborative structure bringing together around thirty public actors and nonprofits around a core group of eleven organizations involved in its governance. It aimed to transform the organization of regional public policies on gender equality into a collaborative network with extensive participation of local SSE organizations in addition to nonprofits specializing in gender equality. The projects organized by the Group included organizing a series of local meetings for public and SSE actors, developing a participative Mapping of resources and actors for gender equality (the Mapping project), and setting up a network of equality referents in SSE organizations. The two public bodies at the heart of the Group are the Regional Directorate for Women’s Rights and Equality (DRDFE), a public administrative body responsible for specific public policies on gender equality representing the National Ministry at the regional level, and the Regional Directorate for Youth, Sport, and Social Cohesion (DRJCS) supported by its public innovation laboratory. Two main types of nonprofits are taking part in the Group. The nonprofits specializing in equality are major ones and hire paid workers: RFemmes*, Universelle* (professional equality), the departmental federations of Women’s Rights Information Centers (women’s rights) and the departmental federations of Planned Parenthood (reproductive health and rights). Nonprofits who do not specialize in equality are also active. Échange asso* leads the Group and is a large regional nonprofit that brings together nonprofits, SSE organizations and their federations. ForESS* is a nonprofit sharing resources with Échange asso* and which also lobbies in this sector. Other actors non-specialized in equality include the departmental federations of social and community centers and Jeunessetsports*, a federation of sports nonprofits. The DRDFE funds some nonprofits of the group through public subsidies, but the nonprofits’ financial dependence on this public actor is limited because of the modesty of its contribution to their budget both in absolute and relative terms.

12 To study the transformation of cross-sector relationships associated with post-NPM paradigms and consider the contextualized nature of collaboration empirically, researchers must target a particular area of public action (Rigaud, 2012). For this reason, the perimeter of the ethnographic case was defined around a specific public policy with a context in favor of developing collaboration. Public actors specializing in gender equality have historically had less hierarchical relationships with nonprofits because of limited human and financial resources and a common value transcending sectoral identities: gender equality (Dauphin, 2010). The Equality Group involves a small number of organizational and individual actors. It draws on rich interpersonal connections between actors and a common territory and community. I chose to focus on a regional level of public action because of the territorial nature of collaboration (Thiel, 2019) and gender-equality issues [3].

2.3. Data collection and analysis

13 Participant observation plays a central role in ethnography. Field notes are the primary support for collecting and analyzing data (Van Maanen, 2011). The field notes collected for this study comprise six hundred pages of structured notes tracing the investigation from the launch of the Group in 2018 to the end of the data collection in January 2020. The observation covered all types of the Group’s activities. Internally, I observed the governance and project management meetings. I also observed the Group’s public events. Many informal interviews with the Group’s members were conducted during the participant observation and later transcribed. I complemented the participant observation data with extensive document collection aimed at contextualizing the case internally (history of relations between members, activities of the nonprofits) and externally (national and supranational frame for gender equality public policies). Nineteen semi-structured interviews with Group members were conducted, recorded, and transcribed during the exploratory phase. The data collection process was designed in a complementary perspective: participant observation documented collaboration as an ongoing process driven by the actors, document collection improved the understanding of the context, and the semi-structured interviews documented the actors’ perceptions of collaboration.

14 Organizational ethnography is an analytical perspective in its own right (Van Maanen, 2011, p.218): in which the fieldwork, the headwork and the text work are intertwined. The analysis began during the process of data collection. Principles of writing fieldnotes (Emerson et al., 2011) have guided me to connect three types of writing. (1) On-the-spot fieldnotes are the preferred method. Actors knew about the on-going research, and it was perceived as normal to take notes in most observed situations. This enabled immediate and detailed transcriptions. Writing with precision limits the risks of the researcher being judgmental and drawing on overly abstract generalizations (Goffman, 1989, p.131). I transcribed native language, direct quotes, and conversations. Informal interviews were collected at the margins of observed events as part of a trust relationship between the researcher and the actors in the fieldwork. This unstructured data (Flick, 2018) is a strategic source of information. When it was not possible to transcribe on the spot, notes were written in a narrative form right after the event with a focus on the key idiomatic expressions that I recalled from the conversation. (2) Summaries of the observed events constitute an initial condensation of the notes. (3) Regular re-reading of the field notes gives rise to analytical writing, the “memos in progress” (Emerson et al., 2011). These are nurtured by sharing the analyses with the actors of the fieldwork. They represent an intermediary narrative of the fieldwork, bridging the conceptual framework and empirical work. The relationship between theory and the field is considered using the extended case method [4] (Burawoy, 2003), which conceptualizes research as iterative. The narrative is organized around vignettes (Van Maanen, 2011).

3.1. Two accounts of the genesis and structuring of the collaboration

15 There are two distinct narratives of the genesis and the structuring of the collaboration, depending on the actors’ sector. On the one hand, the Group’s nonprofits claim they engaged in the Group due to the DRDFE’s initiative. They describe their motivation to collaborate in terms of obligation and recognition by the public actor:

16

“If we’re invited around the table, we can’t say no. What kind of signal would that send? It’s good that they’re thinking of us. It shows that we’re still a force to be reckoned with.” Margot, RFemmes* employee
“They [the DRDFE] asked us to take part, and it’s an honor because it’s a testament to all the efforts we’ve made over the years.” Joy, Head of Fédésports*

17 The nonprofits credited the DRDFE with selecting the organizations taking part in the Group. The nonprofits specialized in gender-equality highlight the absence of fully volunteer-based nonprofits, nonprofits whose activity center on feminist advocacy and with a multicultural perspective on equality, even though these organizations participate as members of the audience during the Group’s public event. They attribute this to a strategy of delimiting collaboration by the DRDFE. Only Échange asso* explained that it had invited other SSE actors to take part in the Group meetings at the DRDFE’s request.

18 On the other hand, the DRDFE downplayed its role in initiating the collaboration. Its director explained that she only identified an existing need in the region’s SSE sector and had provided support for the structuring of a self-organized collaboration by the nonprofits.

19

“I have no power over nonprofits. They have their freedom of association and their history. We try to work together and see how we can build something together. Once again, I see the role of the DRDFE more as ’sowing seeds’: we try to facilitate local initiatives, but we don’t control what grows.” Gaby, DRDFE

20 There are also two distinct narratives to explain the central position in the Group of the nonprofit Échange asso* as the “network leader” or “facilitator”. The nonprofits refer to it as a “designation”, “appointment” or “choice” made by the DRDFE. Léo, an employee of Échange asso*, describes his organization’s involvement as a wish of the DRDFE, which “came looking for us” to structure the Equality Group “that she [Gaby] wanted to create”. Other nonprofits never expressed their concerns to the public actors and to Échange asso* regarding the lack of transparency of the leader’s designation. Yet, the nonprofits specialized in gender equality see themselves as the “natural” leaders of the Group, because of their expertise, and do not understand why they have not been chosen as facilitators.

21

“Choosing a generalist [non-specialized in gender equality] organization… This made me doubt and at one point I said to myself “is she [Gaby] angry with us? What have we done wrong?”. And then no, I don’t think so, but it’s not clear this appointment, it’s not clear…” Margot, RFemmes* employee

22 Échange asso*’s commitment, its leadership expertise and Léo’s personal qualities are acknowledged, but the lack of specific skills in gender equality for a collaboration dedicated to this issue is described as an obstacle to collaboration. On the contrary, Gaby on the behalf of the DRDFE describes how Échange asso*’s central position was not a result of their choice but emerged organically.

3.2. Tensions about the funding of collaboration and bricolage of non-competition rules

23 Funding of the collaboration is a source of tension, particularly around the Mapping project. The goals seem unrealistic to the nonprofits without dedicated financial resources, i.e., an exhaustive, up-to-date list of resources with an efficient IT system. The subject is frequently brought into discussion during meetings. The DRDFE suggests that funding should be sought from corporate foundations. This recurring incentive to seek private incomes and the absence of a public budget for the Group’s project is perceived as a way of relieving the public actors from their responsibility, even among nonprofits whose budgets are already partly funded by foundations. This criticism is never expressed directly to the DRDFE.

24 Besides the funding of the collaboration, the link between the funding of the nonprofits’ activities and that of the collaboration is an issue. Nonprofits experience and perceive an ambiguous attitude from the DRDFE towards competition between nonprofits.

25 This perception of an ambivalent position of the public sector fosters mistrust. The nonprofits identify competition between the group’s nonprofits as well as between nonprofits more generally as the primary threat to the development of collaboration. The Group is marked by differences in size and specialization of the nonprofit members. The nonprofits specialized in gender equality are smaller, while the generalist nonprofits have more human and financial resources. This imbalance is a threat to gender equality nonprofits: collaboration is an opportunity to spread the culture of equality, but sharing information could equip partners to become competitors. Nonprofits non-specialized in gender equality are aware of this asymmetry and the mistrust it generates. Because there is no clear commitment from public actors and no formal rules on collaboration addressing the competition and predation issue, generalist nonprofits engage in the bricolage of rules. They frame collaboration as a collective project of resistance of SSE actors to market competition.

3.3. Dissatisfaction with the collaborative governance and loss of momentum

26 The collaboration appears to be running out of steam. The gender equality nonprofits expressed early their skepticism about the expected outcomes and their feeling that the cost of working together was too high. The multiplication of meetings and reporting tasks are seen as time-consuming for disappointing and slow results.

27

“I think they [public and generalist nonprofits] find it hard to realize that in small nonprofits, a worker who sits in a “useless” meeting is someone missing on the field.” Jeanne, FédéC* employee
Pierre, director of Universelles*, is pessimistic about the Mapping project: “When it’s slow like this, it’s easy to get discouraged. I don’t see how we’re going to manage without someone to update the data […] and there’s no one to take responsibility for that on the medium term. We’re investing time in a project that can’t possibly be solid.”

28 Concerns and a collaborative fatigue are fueled by the feeling of a lack of transparency and formalization of the collaboration rules and governance and an ambivalence of public actors on the funding issue. During the process, Échange asso* shows signs of demotivation, questions the motivations of the public actors and reflects on the persistence of mistrust between members despite the bricolage of non-competition rules and friendly interpersonal relationships. The structure of the collaboration is formally designed around a central nonprofit actor, Échange asso*, but the nonprofits perceive it as a hidden proxy governance led by the DRDFE.

29 Relations between group members are affected by this situation.

30

During an informal interview, Léo, an employee at Échange asso*, described a feeling of unease and of things being left unsaid during the meetings. I asked him how he felt about his role as a facilitator. He replied he felt he got on well personally with all the members of the Group, but that something seemed to be “stuck”. In his view, the problem was failing to clarify the responsibilities of the partners beforehand: “There was a lack of dialogue, and we’re paying for that now”. He regrets the lack of collective discussion forums around the governance, consultation or formal choice by vote.

31 The conflicts remain unresolved as nonprofits do not express their concerns, dissatisfaction and uncertainty to the public actors. Additionally, it is difficult for them to communicate their concerns to Échange asso*, who is perceived as a delegate of the DRDFE’s authority. The regular demand for funding for the Mapping project is the most explicit expression of concern to the public actor. The nonprofits never expressed any desire to leave the collaboration, even in informal conversations, and all the organizations stayed involved in it. The outcomes of the collaboration are difficult to assess; projects took a long time to be implemented. The Group’s public activities came to a halt with the Covid crisis and the simultaneous departure of the civil servant who initiated it. However, the Group laid the groundwork for new relationships between public and SSE actors, which have been essential for developing innovative cross-sector collaboration around gender equality policy during the crisis.

4.1. Mandated collaboration versus coopetition

32 Actors share an aspiration to transform their relationships and the organization of public policies towards a collaborative network. However, the study shows a misalignment of the perceptions of the collaborative network and the expectations of the partners. Perceptions differ about the positions within the collaborative network, the sectoral boundaries, the distribution of responsibilities throughout the collaborative process, its structuring, governance and funding, and the design of collaborative rules. These perceptions are sector-specific. While the diversity of the Group’s nonprofits explains the different perceptions of the collaborative advantage (Gazley, 2010), nonprofits have a common perception of collaboration.

33 The Group’s public actors envision collaboration as incompatible with the persistence of hierarchy. This explains their strategies of effacement or even denial of the authority they have on nonprofits and in the collaboration. The DRDFE downplays its responsibility for initiating the Group and selecting its partners and leader. Its discourse rejects a specific position of the public actor in the collaborative networks, conceptualized as polycentric and organized around a nonprofit actor. The staging (Goffman, 1973) of collaborative relationships by the public actor aims at equalizing status and erasing sectoral differences. The use of the gardener-sower metaphor limits the responsibility of the public actor in the collaborative process and in the collaboration’s outcomes. How the collaborative network should be articulated with the market regulation is not perceived as a tension. As a result, it is not at the heart of the public actors’ work to build collaboration. The DRDFE participates indirectly in the reinforcement of competition and market regulation: by rejecting the possibility of using hierarchical authority to formalize and reinforce the rules of non-predation and non-competition created by the Group’s nonprofits or by not reacting when another public actor encourages predation between members. It contributes directly to the embedding of the collaborative network in the market by extending competition between nonprofits through collaborative activities, by proposing the funding of collaboration outside the public sector and by rejecting a specific responsibility in the modes of nonprofit financing. This vision of collaboration as coopetition against hierarchy reflects a conception of going beyond NPM oriented towards a post-bureaucratic paradigm (Clegg et al., 2011).

34 The nonprofits perceive the implementation of the collaborative network as a hierarchically mandated collaboration (Rodriguez et al., 2007), organized as an alternative to the market logic. They enter in the collaboration on the unilateral request from the DRDFE, which supports the benefit of hierarchy for the actors’ engagement in the collaboration (Whitehead, 2003; Levelt & Metze, 2014). Nonprofits support the embedding of collaboration in hierarchy (Nederhand et al., 2016). The position of a governmental actor, the DRDFE, as the central node of the collaborative network (Geddes 2012) is not experienced as a hindrance to collaboration (Chabault & Martineau, 2013; Suquet et al., 2020) by nonprofits. Sectoral differences and complementarity (Salamon & Toepler, 2015) are seen as necessary for collaboration, which is manifested in the metaphor of the public conductor coordinating the nonprofit orchestra. The collaborative process as it is led by the public actor thus disappoints nonprofits, as they see it as an ethical renewal of NPM through the civil society (Walzer, 1991) they represent.

35 These two sectoral visions of collaboration are in opposition. Public actors engage in blurring sectoral boundaries, which prevents actors from acknowledging and discussing the existence of two visions. In this situation, there is no possibility of alignment, negotiation or construction of a common middle way or alternative vision of collaboration. This generates nonprofits’ mistrust towards the public actors and slows down the trust building between the Group’s nonprofits.

4.2. Illegitimacy versus legitimate ways of exercising hierarchy in collaboration

36 The study highlights a paradoxical situation: while nonprofits accept and support the hierarchical authority exercised over them to build collaboration, the public actors holding this authority seek to erase it. The way hierarchy and collaboration are articulated (Gazley & Guo, 2020; Gazley, 2017) has been theorized as a balancing of conflicting forms of legitimacy: the rational legal legitimacy linked to the efficiency of centralized coordination by a hierarchical authority and the democratic legitimacy associated with the decentralization of decision-making in collaborative networks (Peters, 2011). The opposition between hierarchy and collaboration reflects the public actors’ internalization of the illegitimacy of unilateral regulation and hierarchical bureaucracy, which is a core argument of NPM (Pollitt & Bouckaert, 2011). As such, the public actors convey “the contestation of [their] hierarchical leadership in the orientation and coordination of the actors involved in the production of public action” (Bergeron et al., 2011, p.1). The DRDFE refuses to be using hierarchy as a lever for engaging other actors (Levelt & Metze, 2014). However, the use of hierarchy is not seen as illegitimate per se by the nonprofits. In addition, the hierarchical position supports the collaborative network despite the public actor’s formal refusal to use it.

37 It is surprising that nonprofits stay in collaboration despite their perception of a collaborative disadvantage (Gazley, 2017). Indeed, it seems contradictory with the idea that actors have no interest to engage in collaboration if they perceive any alternative to it (Bryson et al., 2015). The use of hierarchy explains this situation. The impetus to collaborate from the public actor transforms the nonprofits’ frame of perception. Collaboration is perceived as an expectation public actors have regarding nonprofits, which conditions their legitimacy in and beyond the specific collaboration. Hierarchy transforms nonprofits’ perceived cost/benefit assessment to enter and remain in cross-sector collaboration: exiting becomes a risk. The presence of the public actors encourages nonprofits to respect their commitment to the Group (Acar et al., 2008).

38 This result refines the work on the benefits of hierarchy for collaboration (Levelt & Metze, 2014) which in this paper is theorized as a collaborative [5] ratchet effect enabled by hierarchy. Hierarchy creates a collaborative obligation. It leads participants to frame collaboration as the only conceivable solution (Bryson et al., 2015) despite the perceived organizational risks. The involvement of hierarchy in collaboration prevents any backtracking and disengagement from the collaboration.

39 Nonprofits do not blame hierarchy for the difficulties they experience in working together. Instead, they identify the issue as hierarchy embedded in a market logic, and the way it is exercised by the public actor is perceived as problematic. Nonprofits do not understand the official narrative of the DRDFE and its refusal to admit any specific role in initiating the Group and structuring its governance. They criticize the lack of transparency in exercising the hierarchical authority and the refusal to accept sectoral differences. The gap between the sectoral visions of the network generates mistrust towards the public sector. Under these conditions, Échange asso* cannot become a boundary organization in the collaboration (Suquet et al., 2020), because it is perceived as an undeclared proxy for the public sector. The implementation of the principle of non-competition by the facilitator strengthens trust between the nonprofits in the Group, but it is hampered by the lack of transparency on the part of the public sector actor. Trust is difficult to build because the public authority’s stance is perceived as ambivalent. Hierarchy is not explicitly mobilized, but it implicitly structures the collaborative network. It is invisible, but it persists and reinforces itself in the collaborative process. This situation leads to uncertainty among the nonprofits regarding the agenda that public actors pursue in collaboration. They fear that collaboration is used as an instrument for disengaging governments from public action, for diluting its responsibility towards nonprofits. Nonprofits worry about the role of public actors in increasing market competition.

40 The findings question the dichotomy between the rational legal legitimacy associated with hierarchy and the democratic legitimacy associated with networks (Peters, 2011; Gazley & Guo, 2020; Gazley, 2017).

41 Hierarchy can be democratically legitimate depending on its condition of exercise. Therefore, the issue for public managers doesn’t lay in balancing conflicting legitimacies as much as in creating legitimate ways to exercise hierarchy in and for collaborative networks. This aligns with the debates on the public service ethos (Stoker, 2006, p.49) that should frame collaborative networks’ practices, i.e., the public actors’ commitment to principles and a code of conduct aimed at transforming their relations with actors of their environment for the sake of managing public interest.

4.3. Practical implications for public and nonprofits organizations

42 DRDFE’s attempts to build more informal spaces and gender equality policies as favorable to flatter relationships with nonprofits (Dauphin, 2010). Highquality interpersonal relationships are a first step in trust building for the collaborative networks, but they are not sufficient. There is no direct translation of the trust in the public actor as a person representing the public organization, to the DRDFE as an organization, and to the public sector it is a part of. The interactions we observe at a micro-level are intersectoral relationships beyond interpersonal and inter-organizational ones. For the nonprofits in the Group, the DRDFE has a particular financial responsibility because it is a public actor, even if the subsidies it controls directly are small. Public actors shall take seriously the nonprofits’ perception of a power imbalance, even if the power of their organization is restricted. Public actors who are not engaged in the collaboration affect the nonprofits’ perception of cross-sector relations and can limit trust building. Collaborative public actors cannot necessarily influence these other actors, but they can make it clear that they do not approve incentives of predation between nonprofits.

43 This case illustrates the difficulty of public actors to position themselves (Stoker, 2006; O’Flynn, 2007) in their new relationship with nonprofits. Public actors would benefit from using their hierarchical authority as a lever for collaboration and deconstructing their vision of the illegitimacy of hierarchy per se. This is a belief associated with NPM that does not correspond to the perception of nonprofits as civil society organizations. Hierarchy is not an obstacle to collaboration. Efforts to erase hierarchy result in its hidden renewal in collaboration, which is perceived as not transparent and not legitimate by the nonprofit partners. Hierarchy should be conceptualized by public actors as a sectoral difference they can mobilize strategically to serve collaboration in a perspective of sectoral complementarity (Salamon & Toepler, 2015.