Designing collaborative governance that is fit for purpose: theorising policy support and voluntary action for road safety in Sweden

Abstract

Collaboration has become a key element of governments’ efforts to address complex policy problems as well as of attempts to revitalise democracy. Depending on what public agencies want to achieve by engaging in collaboration – whether collaboration is intended to gain support for public policymaking, induce voluntary actions by stakeholders, or something else – collaborative governance will have to mean very different things. Collaborative governance needs to be made fit for purpose. Drawing on the literature and a case study of road safety governance in Sweden, the article theorises on the interplay between strategic purpose and institutional design in collaborative governance and shows how two types of strategic purposes – policy support and voluntary actions – determine appropriate choices of institutional designs. This generates important insights that contribute to understanding institutional diversity and factors important to the success or failure of collaborative governance, and can aid practitioners who are designing collaborative forums within various policy fields.

Introduction

Collaborative governance has attracted widespread attention and support amongst scholars and policy practitioners, who see it as a vital element of governance efforts to deal with a wide range of complex policy problems as well as a fruitful way to restore trust in government, foster public engagement, and further the legitimacy and effectiveness of public actions (Gray 1989; Innes and Booher 1999; Fung 2006; Margerum 2011; Ansell 2012; Emerson et al. 2012). Collaborative governance is here defined 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” (Ansell and Gash 2008, 544).

The form and functioning of collaborative governance, as well as factors affecting its development and implementation, vary depending on the situational and contextual setting (Ostrom 1990; Koontz and Newig 2014; Prentice et al. 2019) as well as the specific task, function, or purpose they are set up to achieve (Agranoff 2006; Provan and Kenis 2007; Bryson et al. 2015). Collaborative governance is a “means to an end” (Scott and Thomas 2017, 193) that “public managers will (and should) structure /…/ differently depending on their goals” (Prentice et al. 2019, 802). The case made here is that collaborative governance is used purposively and strategically by public managers (Agranoff and McGuire 2003; Koontz et al. 2004; Koontz and Thomas 2006; Milward and Provan 2006; Scott and Thomas 2017; Imperial et al. 2018; Prentice et al. 2019) and that the purpose it is meant to achieve will significantly affect the appropriate choice of institutional designs in terms of the authority and resources invested in the arrangement, its membership and structure, and the roles and interactions of different actors (Agranoff and McGuire 1998; Huxham 2000; Koontz et al. 2004; Agranoff 2006; Milward and Provan 2006; Thomson and Perry 2006; Ansell and Gash 2012; Imperial et al. 2018). With more hopes being placed in collaborative arrangements, and more public resources being devoted to creating them, it is becoming increasingly important to develop a nuanced theoretical understanding of what types of collaborative governance are suitable for what purposes, that is, to make collaborative governance fit for purpose. This has been called a key research question that few studies so far have attempted to address (Gerlak et al. 2013; Bryson et al. 2015; Prentice et al. 2019), and that requires bridging the divide between institutional and managerial approaches to collaborative governance. This article aims to make an important contribution to this end.

Building on existing theoretical work on collaborative governance and an original case study, this article aims to theorise on collaborative governance as strategically used by public managers for different (public) purposes, and on how this interacts with the institutional design of collaborative arrangement. Two types of collaborative governance are elaborated on; policy support and voluntary action. Using a case study, the article empirically illustrates how the interplay between strategic purpose and institutional design is critical for the successes and failures of collaborative governance.

Following this introduction, research on the purposes and institutional design of collaborative governance is discussed. The research design is then presented, followed by the empirical analysis of two collaborative governance arrangements established to promote road safety in Sweden. Grounded in both theoretical argumentation and empirical analysis, the article concludes with important insights for collaborative governance research and recommendations for practice on how to make collaborative governance fit for purpose.

Collaborative governance: integrating strategic purpose and institutional design

Collaborative governance has developed into an influential research agenda, integrating research across such fields as collective action (Ostrom 1990), planning (Innes and Booher 1999), implementation (Hjern and Porter 1981), management (Kickert et al. 1997; O’Toole 1997), and public participation (Fung 2006). The concept encompasses a great variety of forums in which public policies and programmes are developed, implemented, or managed through a close and consensual collaboration between public agencies and non-state stakeholders (Ansell and Gash 2008). Distinctive for these forums is that stakeholders are not just consulted by government, but actively contribute to actions or decisionmaking (even if the final decision-making authority may be in the hands of public agencies), and thus are accorded policy influence and responsibility (Wood and Gray 1991; Margerum 2011).

Much theorising on collaborative governance adheres to institutional theoretical perspectives, focusing on structural conditions under which collaborative governance can be expected to work (Ostrom 1990; Margerum 2011). The second approach, emerging out of public management research, treats collaborative governance as an instrument used strategically and purposively by public managers to address complex policy problems in situations where working together is better than acting alone (Agranoff and McGuire 1998; Scott and Thomas 2017; Imperial et al. 2018; Prentice et al. 2019). Both approaches offer key insights into the dynamics of collaborative governance, and although there are tensions between them (O’Leary et al. 2009; O’Leary and Vij 2012) they need to be integrated to advance theory and further knowledge about how different strategic purposes interact with institutional design decisions.

Before discussing strategic purposes for collaborative governance, it is important to note that this article focuses on top-down collaborative governance initiated by the government, a process in which public managers can be expected to play a more direct role in structuring and managing collaboration in pursuit of a strategic purpose than in collaborative governance resulting from bottom-up processes within local communities (Koontz and Thomas 2006; Koontz and Newig 2014). While collaborative governance is generally heavily influenced by the government (Huxham 2000), these different processes have their own dynamics which are likely to be reflected in strategic purposes and institutional designs as well as the challenges and opportunities faced.

Strategic purpose of collaborative governance

A mutually agreed purpose is generally stressed as important for a collaboration, and together with mutual trust and interdependency is seen as an “embodiment of the principles of collaborative governance” (Huxham 2000, 351). The purpose serves to direct and structure collaborative arrangements and to influence actions and behaviours. However, different collaboration purposes can be entangled, changing, or hidden (even from the actors themselves). Diverse and conflicting interests and goals amongst individuals and organisations tend to make it difficult to define a joint purpose (Huxham 2003; Vangen and Huxham 2011). What collaborative governance is to achieve is thus most often treated as an open question; it is through the process of collaborating (and deliberating) that the purpose of the collaboration is manifested and made concrete. In other words, the purpose is seen as evolving over time as a result of actions, and while not ideal, the first thing that is important to do is to set up a collaborative forum and initiate action (Huxham 2000, 2003). From a management perspective, it is essential to envision a purpose of collaboration and to design a process by which this purpose can be carried out before initiating a collaboration (Gray 1989, 71). Collaborative governance means investing substantial public resources (time, energy, money), and, while collaboration is sometimes embraced as an end in itself, engaging in collaborative governance is not a decision to be taken lightly (Huxham 2000, 2003; Imperial 2005; Zachrisson et al. 2018). There is a tendency in the literature to be “celebratory and only rarely cautious” (O’Leary et al. 2009, 6), neglecting risks associated with collaboration such as agency capture, mission drift, loss of organisational autonomy, costs, difficulty evaluating results, and problems with upholding public accountability (Gazley and Brudney 2007; Kallis et al. 2009). Managers should have a clear purpose in mind and know what is to be achieved (in terms of solving public problems or creating public values) before engaging in collaboration; otherwise, it might not be worth the effort, and public resources should be directed elsewhere (Agranoff 2006; Huxham 2000, 2003; Imperial et al. 2018; Zachrisson et al. 2018). It is also important that appropriate institutional designs, including decisions on whom to invite and how to organise and assign roles, largely depend on the purpose of the collaboration. Collaborative governance needs to match the purpose it is intended to achieve; otherwise, the discrepancy between institutional design and purpose is likely to have negative effects on its effectiveness (Provan and Kenis 2007; Bryson et al. 2015). Consequently, public managers need to have a strategic purpose in mind when engaging in collaborative governance.

The strategic purpose of collaborative governance can be understood as situated in a tripartite structure of nested institutions (cf. Ostrom 2007) (Table 1). The initial conditions and general orientation of specific collaborative governance arrangements are generally set by the government through policy goals and frameworks (e.g., to alleviate water shortage using collaboration rather than regulation or market-mechanisms). The type of collaborative governance (in terms of what the collaboration is intended to achieve, i.e., its strategic purpose) is framed by these conditions, but is generally decided or substantially influenced by public managers, who also make strategic decisions about the institutional design. The institutional design shapes the collaborative process in which participants’ various interests and goals are shaped (through deliberation) into operational activities and joint actions (ideally decided through consensus) that, in turn, generate collaborative output. Public managers can influence collaborative governance by acting within the collaborative process (as well as advising government), but key managerial powers consist of deciding on strategic purposes and institutional design.

Table 1. Strategic purpose of (top-down) collaborative governance

Collaborative governance can have various strategic purposes. Agranoff and McGuire (1998) distinguish three types of networks according to their strategic purposes: to formulate and/or implement policy; to exchange resources; or to develop temporary projects. They later added “information seeking” and “to seek adjustment of government programs” (Agranoff and McGuire 2003, Chapter 4). Also, viewing networks as emerging or being created for specific purposes, Milward and Provan (2006) identify four types: service implementation, information diffusion, problem-solving, and community capacity-building. Both of these typologies theorise on a broader set of public management networks, and while information-seeking, capacity-building, and resource exchange may be the strategic purposes of these networks, in collaborative governance, they are better seen as key activities or functions that are integrated within most arrangements. Drawing from an institutional perspective, Margerum (2008, 2011) has developed a typology based on what collaborative groups deliberate and distinguishes between on-the-ground actions, organisational roles and programmes, and government legislation and policies. This typology is more closely aligned with how collaborative governance is defined here, but by building on institutional levels, it shifts the focus away from purpose, especially since collaborative governance tends to operate across these levels (cf. Emerson and Nabatchi 2015). Similar to the case made here, all three typologies emphasise the importance of matching form to function (cf. Gerlak et al. 2013).

Building on these typologies and integrating insights from the managerial and institutional approaches, we here elaborate on two types of collaborative governance distinguished by a strategic purpose: collaborative governance to generate policy support and collaborative governance to induce voluntary action. These types are best seen as existing at different ends of a spectrum ranging from government-oriented (collaboration to support government actions/decisions) to society-oriented (collaboration to support actions/decisions by non-state actors). The types also represent collaborative governance oriented towards different policy process stages – policy development and implementation, respectively. The typology is elaborated for analytical purposes, to help guide and organise empirical studies, to enable further specification in terms of appropriate institutional design, and to aid comparison (Bailey 1997; Emerson and Nabatchi 2015, 159). It is important to recognise that theoretical types rarely manifest in their “archetypal” forms in real life. To recognise this, we analytically distinguish between the main purpose – providing the defining characteristics of the type – and secondary effects. Also, collaborative governance is dynamic rather than static (Huxham 2003; O’Leary et al. 2009; Emerson and Nabatchi 2015, 168). This challenges researchers (and practitioners) to take a long-term view, not (only) focusing on formal aims or initial declarations of intent, but also engaging with questions of changes in strategic purpose, seen over time and through multiple perspectives (cf. Sabatier 2007).

The first strategic purpose of collaborative governance is to generate policy support. Here, the main purpose of collaboration is to gather input on public policy initiatives from non-state stakeholders, and thereby to improve policy, “ease” implementation, and/or gain legitimacy for government actions. Actors are given the opportunity to voice opinions, contribute expert knowledge, and, through deliberation and convincing argumentation, to influence policy formulation, implementation or management. Through learning, adaptation, and anticipation, this process can also influence participants’ perceptions and/or actions as a secondary effect, for example, if participants take voluntary action or pre-emptive action against public initiatives.

The second strategic purpose is to use collaborative governance to further policy implementation through voluntary actions (cf. Koontz and Thomas 2006). Here, the main purpose of the collaboration is to foster action amongst non-state stakeholders in support of public policies, goals, and programmes, and hence to increase effectiveness and efficiency. Public agencies facilitate interaction and provide expert support, convincing argumentation, and mediation to convince, persuade, or threaten non-state actors into taking voluntary action, for example, through voluntary agreements or joint action. The secondary effect of engaging with non-state stakeholders is that public agencies get input that can help them refine or change policies and programmes or implementation and management strategies.

Institutional design of collaborative governance

Research on collaborative governance has generated various theoretical frameworks that identify important factors for successful collaboration, showing the importance of both contextual conditions and the design of collaborative processes (Bryson et al. 2015). Political culture and traditions, constitutional and legal rules, and past collaborations are some initial conditions that shape and motivate collaborative governance (Gray 1989; Bryson et al. 2006, 2015; Ansell and Gash 2008; Sandström et al. 2014; Sørensen and Torfing 2019). In this section, we direct attention to what is “inside the box” (cf. Thomson and Perry 2006), and specifically to three important institutional factors amenable to design decisions by public managers: incentives, structures and procedures, and public management.

Incentives

There is a wide array of motives for non-state actors to voluntarily participate in collaborative governance (Gustafson and Hertting 2017). While the problem-solving potential and effects of deliberation may help engage participants and maintain their support, it is often critically important to design tangible incentives to get self-interested actors involved (Huxham 2000; Lubell et al. 2005; Ansell and Gash 2008; Segerson 2013). To incentivise collaborative governance is thus a pivotal task of public managers (Scott and Thomas 2017). Collaborative governance operates in the shadow of hierarchy (Héritier and Lehmkuhl 2008), and credible threats of government regulation or other interventions (e.g. stricter oversight) are one way to induce non-state stakeholders to collaborate in order to influence (or pre-empt) policy intervention (Koontz et al. 2004; Lubell et al. 2005). Another way is for public managers to stimulate key constituencies, shareholders, customers, or members to demand that businesses or other organisations act in a socially responsible manner and collaborate in the face of critical problems (Gunningham 2009a; Hall and Hysing 2019). Third, collaboration can be incentivised by providing “intermediate outcomes” (Ansell and Gash 2008, 561), for example, information and fact-sharing or access to state funding, technical assistance, and best practices (Koontz et al. 2004; Koontz and Thomas 2006; Segerson 2013). Finally, collaboration may offer opportunities to get public endorsement (i.e., reputational value) and to influence public policy and shape future government initiatives. Building and maintaining trustful relations and collaborations with the government may also be important beyond the specific issue at stake, by paving the way for future actions and generating social legitimacy (Sandström et al. 2014; Hall and Hysing 2019).

Structures and procedures

The organisational structures and procedural mechanisms employed in collaborative governance are critically important to building trust, developing a common recognition of interdependency, and establishing a shared sense of purpose (Gray 1989; Ostrom 1990; Innes and Booher 1999; Ostrom and Crawford 2005; Bryson et al. 2006; Ansell and Gash 2008; Emerson et al. 2012; Gash 2016). It is a characteristic of collaborative governance that decisionmaking is “consensus-oriented and deliberative” (Ansell and Gash 2008, 544). While consensus-building has beneficial effects extending beyond decisionmaking (Innes and Booher 1999), it is also prone to gravitate towards risk-averse agendas and lowest-common-denominator decisions, and to emphasise “reaching agreement” over “making decisions” that entail actual implementation. Public managers may thus be better served by concentrating on the substantial problem rather than a specific decision-making procedure (Agranoff 2006; Koontz and Thomas 2006; Margerum 2011).

Access is another fundamental institutional design feature (Ostrom and Crawford 2005; Ansell and Gash 2008). To improve legitimacy and enhance problem-solving capacity, collaborative governance should be inclusive enough to mirror the problem (involving those who have an effect, are affected, and have relevant perspectives and important knowledge), even when this means involving “troublesome” stakeholders (Ansell and Gash 2008, 556). However, large heterogeneous groups make it difficult to develop shared understandings and reach consensus. Being inclusive in terms of participants necessarily leads to the exclusion of issues and possible outcomes. It might be more effective for managers to strategically select stakeholders based on their capacity to act (Connelly and Richardson 2004; Ansell 2012; Karlsson 2012; Gash 2016). Collaborative governance tends to mirror rather than negate societal power relations. Even if they are granted access, participants can encounter powerful exclusionary mechanisms in the form of language (professional jargon), arbitrary legitimation of knowledge, cultural practices of interaction, and fixed worldviews (Gray 1989; Huxham 2000; Agranoff 2006; Brisbois and de Loë 2016a).

Finally, benefits and costs need to be assigned (Ostrom and Crawford 2005) so that it pays for participants to take part (see incentives) at the same time as collaborative governance proves its worth to government sponsors (Agranoff 2006; Gash 2016). Collaborative governance needs to be assessed in the light of alternative government strategies (Ansell 2012), but ultimately, it needs to produce substantial improvements within politically acceptable timeframes (Kallis et al. 2009). Securing the results of collaboration in the participants’ home organisations has proved a challenge (Milward and Provan 2006; Margerum 2008; Gash 2016). Consequently, in assessing collaborative governance, public managers need to consider the extent to which stakeholders actually align their policies, processes, and budget allocations in accordance with collaborative outputs (Newman et al. 2004; Margerum 2008; Brisbois and de Loë 2016b).

Public management

Collaboration has evolved into a key governance strategy in recognition of the government’s interdependency with non-state actors. While the role of public managers is disputed and likely varies across policy fields (Koontz et al. 2004; Margerum 2011; Agger and Sørensen 2018), they generally exercise considerable power and control over collaborative processes (Koontz et al. 2004, 21; Agranoff 2006; Ansell and Gash 2008; Gunningham 2009b; Brisbois and de Loë 2016b; Scott and Thomas 2017).

The question of why and how public managers engage in collaboration has engendered substantial research interest (Agranoff and McGuire 2003; Koontz and Thomas 2006; O’Leary and Bingham 2009; Scott and Thomas 2017). Collaborative governance is often a way to add capacity and resources to fulfil the missions of public agencies and/or a way to comply with government instructions, court orders, or legislation (Ansell 2012; Ansell and Gash 2008; Bryson et al. 2015; O’Leary et al. 2009). Alongside strictly managerial motives, there are also ideological motives related to strengthening participation and empowerment (Huxham 2000). When portraying collaborative governance as a strategic choice, researchers run the risk of treating managers (and others) as individual, purposeful and overly rational actors (Vangen and Huxham 2011). It is important to assume “bounded rationality” (Scott and Thomas 2017) and to view collaboration as pursued in accordance with individual and organisational preferences and interests that are framed and shaped by contextual and situational conditions (O’Leary et al., 2009; Imperial et al. 2018).

Public managers take on different roles in collaborative governance, acting as leaders, encouragers, followers, or even free-riders (Koontz et al. 2004; Scott and Thomas 2017). As a tool of government, collaborative governance is not based on formal authority and needs to be directed using hands-on management, facilitation, and leadership, as well as hands-off structuring, framing, and designing of collaborative arrangements (Sørensen 2006). (Facilitative) leadership is the key to successful collaborative governance. This role includes to help convene and sponsor collaboration and maintain its integrity; manage and arbitrate conflicts; and help identify and realise value-creating opportunities (Koontz et al. 2004, 24; Bryson et al. 2006; Ansell and Gash 2012). While both public and private actors can take on this role (Gray 1989, 71), leading collaborative governance is costly, time-consuming, and has uncertain outcomes, and thus, the task usually falls to the public agency (Gunningham 2009b; Klijn 2016; Scott and Thomas 2017; Zachrisson et al. 2018). Also, when it comes to actual implementation, this almost always becomes the responsibility of the public agency (Agranoff 2006).

Summary of theoretical arguments

This article makes three overarching theoretical arguments. First, the appropriate institutional design of a collaborative governance arrangement (incentives, structures and procedures, public management) depends on what it is intended to achieve. Drawing on the literature, this article contributes insights on the importance of integrating an institutional perspective that focuses on the structural design of collaborative arrangements with a management perspective that sees collaborative governance as resulting from purposive and strategic actions. This integrated theoretical approach to collaborative governance (which will be applied in the following empirical analysis) arguably makes us better equipped to understand institutional diversity as well as the success and failure of collaborative governance.

Second, the strategic purpose of collaborative governance is decided or at least fundamentally influenced by public agencies. Public managers, and what they hope to gain through collaboration, leave a deep “imprint on all facets of collaborative structure and action” (Koontz et al. 2004, 167). This is especially true for top-down collaborative governance. Thus, a key insight generated here is the need to explicitly direct theoretical and practical attention to what public managers want to get out of collaboration, that is, their purposes and rationale for engaging in collaboration (cf. Scott and Thomas 2017). The role of strategic purposes in collaborative governance is further developed by situating it in a tripartite structure and relating it to societal goals and the collaborative process (see Table 1).

Third, as more government resources (time, money, and energy) and hopes are devoted to collaboration (Koontz et al. 2004), it becomes all the more important to place the public purpose of collaborative governance at the forefront of theoretical and empirical work; otherwise, it is at risk of being lost in collaboration (cf. Gazley and Brudney 2007). From a practitioner’s point of view, it is critically important to understand the design of appropriate types of collaborative forums in relation to specific purposes in order to be able to act effectively (Milward and Provan 2006; Bryson et al. 2015). Here, we elaborate on two new types of collaborative governance based on strategic purpose – policy support and voluntary action. This research focus and the combination of theoretical arguments and empirical illustrations are well-suited to bridging the gap between theory and practice (O’Leary and Vij 2012) and thus can provide praxis-relevant insights to help managers to avoid a mismatch between the design of collaborative arrangements and the strategic purpose that they hope to achieve.

For the next step of theorising on the interplay between strategic purpose and institutional design, we turn to the case study of Swedish road safety governance. The case is used to illustrate the institutional design of two types of collaborative governance distinguished by strategic purpose – policy support and voluntary action – and how the interplay affected successes and failures in the cases. Elaborating on theory grounded in both literature and observed practices is a research strategy used (or argued for) in several studies on collaborative governance (Huxham 2003; Imperial 2005; Agranoff 2006; Prentice et al. 2019) and will be further discussed in the next section.

Research design

Why the case of Swedish road safety?

Road accidents are a critical public health issue, claiming more than 1.2 million lives annually, to a cost of 3% of the global GDP (WHO 2015). Around the world, governments struggle to find ways to make transport more sustainable by reducing road deaths and emissions without compromising mobility (Banister 2008). To find support to address these complex transport-generated problems, governments engage in collaboration with non-state actors (e.g., Baumann and White 2012; Hrelja et al. 2018; Paulsson et al. 2018; Tornberg and Odhage 2018).

Swedish road safety governance is guided by the Vision Zero (VZ) policy, which aims to eliminate road deaths and serious injuries by employing a systematic approach to road safety where road administrators, vehicle manufacturers, the police, transport providers, and other so-called system designers are assigned a shared responsibility for road safety (Government Bill 1996/97:137). VZ policies for road safety have been introduced in various countries, including Canada, Norway, and the UK, as well as in cities such as New York (Kim et al. 2017; Kristianssen et al. 2018). Collaboration is a key element of the VZ policy used by the Swedish Transport Administration (STA, until 2010 the National Road Administration, NRA) to support the work of external stakeholders and to further policy development and implementation (Belin and Tillgren 2012; Government Offices of Sweden 2016; STA 2018; Hysing 2019).

The empirical focus of this article is on two cases of collaborative governance: the Group for National Cooperation on Road Safety (GNS) and the Transport Quality (TQ) project. The cases display important similarities in terms of collaborative approach (top-down), scale (national level), societal goal (road safety), and institutional context (Swedish road safety governance). They differ, however, in terms of the strategic purpose of collaboration and in that they target different stages of the policy process (policy development and implementation). Thus, the case selection offers opportunities to study the interplay between purpose and institutional design in two collaborative governance cases within a similar setting (cf. O’Toole and Meier 2014).

Methods and materials

This article builds on qualitative data from an original case study conducted in 2018. The empirical material consists of documents (evaluations, annual reports, and presentation materials) and semi-structured interviews with 18 representatives of organisations, companies, and authorities engaged in road safety governance (Appendix I). Interviews were conducted in person or by telephone/Skype and lasted from 32 to 224 minutes (median 54 minutes). The interviews were based on, but not restricted to, an interview guide focusing on the attribution of responsibility and governance processes, i.e. how, why and by whom governance is conducted (Appendix II). Interviews were recorded and transcribed verbatim. All quotations from the interviews were professionally translated.

Respondents were strategically selected to represent a broad set of stakeholders identified through policy documents as well as snowball sampling (where respondents are identified as relevant during initial interviews). The sampling strategy was designed to ensure a variety of voices and perspectives regarding road safety governance in general and collaborative governance in particular, and included both insiders and outsiders of the two collaborative arrangements. The technique does not guarantee a full or representative sample, but it does help to identify actors whose insights, experiences and perceptions can help give a nuanced, multi-perspective picture of collaborative governance.

The material was analysed using a qualitative thematic approach (Bryman 2012, Chapter 24). Analytical dimensions were generated both from previous research and through lessons and insights generated from the empirical cases. In the end, four dimensions proved salient: purpose, public management, structures and procedures, and incentives. Documentation and interview transcripts were analysed together and coded according to these dimensions. Common categories and controversial issues were noted, as were the positions (insider–outsider) of the respondents. The case material for each collaborative arrangement was compiled and compared with regard to the four dimensions. By comparing cases, key similarities regarding collaborative governance characteristics and differences in purposes and institutional design features could be identified. Taken together, these insights were finally used to identify critical factors behind the successes and failures of the arrangements.

Strengths and limitations

Cases from various policy settings contribute valuable knowledge for theory development as well as improving practices of collaborative governance (Prentice et al. 2019). However, there are both strengths and limitations to this design. First, the empirical analysis provides important insights in its own right, but also supports the theoretical aim of the article by illustrating how the interplay between purpose and design may take place in practice, including how it affects the success of collaborative governance. This choice of cases based on differences in public purpose means that the two cases address different stages of the policy process. Further research on a wider and more diverse set of cases is needed, as are comparative studies of more similar collaborative governance to validate and elaborate on the findings. Second, this is a study of top-down collaborative governance within a single national and sectoral context. Sweden is characterised by a consensual political culture built on a neo-corporatist tradition of using collaboration to address policy problems (Hall and Hysing 2019). Also, the transport sector has long been characterised by collaboration. These contextual conditions likely influence the resulting collaborative governance (cf. O’Toole and Meier 2014), which cautions against generalising empirical findings across sectors and countries. Finally, the case-based approach contributes contextual and in-depth empirical knowledge based on multi-perspective accounts of the cases. This is necessary to understand the inner workings and dynamic character of collaborative governance (cf. Koontz and Newig 2014; Prentice et al. 2019) and strengthens internal validity, but also weakens external validity and replicability (Bryman 2012, Chapter 17).

Analysis

GNS

As an instance of collaborative governance, the GNS can be described as a long-standing, open-ended, formal collaborative forum where the public agency STA engages stakeholders in a consensual and deliberative way (cf. Ansell and Gash 2008). The forum was established in 1993 by the NRA as an arena for the exchange of knowledge and coordination between the agency and stakeholders considered to have a substantial impact on the road transport system. GNS is organised as the main body dealing with “strategic policy issues” (interviewed STA Director) involving 14 agencies, organisations and companies, and six expert groups that organise collaboration on specific issues (e.g., speed compliance) and involve a broader set of actors. Officials at the STA play key roles in organising, convening, and setting the agenda for the group, which has the formal “aim of improving their joint capability to actively contribute to Vision Zero” (STA 2018). The STA director heading the group described it as an important platform for anchoring public policy, sharing information, and deliberating on policies and measures. The main purpose of the group, as conceived by public managers, is to gain policy support from other public agencies and non-state stakeholders for public policy prioritisations and co-development of more operational strategies. The GNS has been producing an annual “Guidance for Operational Planning” document in which the group evaluates policy progress based on a set of indicators (see Thoresson 2018). It also makes recommendations about what interim targets that should be prioritised by the STA and other actors working with road safety. In 2018, the guidance document was replaced by a four-year action plan to guide the work on road safety in Sweden, developed by the STA in collaboration with GNS (STA 2018). Other outputs from the GNS are strategies related to specific issues, e.g. safe cycling, produced within the expert groups.

Most respondents, both inside and outside of the group, argued that the forum plays an important – and at times even decisive – role in road safety governance. As stated by representatives of an organisation that is not a member of the group:

Then there’s a group called GNS, it decides what interim targets we should have in Sweden and what we should measure /…/. And whatever isn’t an interim target is completely uninteresting.

As for its serving as a collaborative forum, this has reportedly changed over time. The introduction of the VZ policy in 1997 changed the direction of Swedish road safety governance and the designated roles and responsibilities of non-state stakeholders. This shift was also reflected in the GNS. A former Traffic Safety Director portrayed it as a cultural shift in how people interacted in the GNS, from a conflictual to a consensual form of collaboration. According to him, GNS was previously characterised by:

An old-fashioned kind of collaboration where /…/ everyone gathered around a table and told each other what they should do. /…/But then this journey began /…/ we think that the meetings should be about what you do yourself. And do you need anyone else’s help. And that started to be like a mantra, that you should talk about what YOU are going to do.

As for structures and procedures, GNS is a relatively closed forum with a limited membership invited by the STA. What actors are invited and what interests are represented were found to be critical issues for the legitimacy and functioning of the GNS. The official who headed the group described it as “society in miniature,” and when asked about the criteria for being invited, the director argued that invitations go to “major societal actors that have the wherewithal or for some other reason have the potential to be an important actor.” However, several members acknowledged that the basis for being invited “was a bit arbitrary,” and that to some extent tradition, personal relations, and a personal commitment to the VZ were valued above the importance of the organisation that participants represented (cf. Huxham 2000).

One particularly critical issue was the position of road user organisations. According to a leading STA official, “they are represented, just in a different way [invited to participate in the expert groups], but they aren’t all taking part in the GNS. And that’s largely for practical reasons and so that we can [work on] this policy- and systems level, and not single issues too much.”

Respondents representing one road user organisation, however, did not consider themselves part of GNS, after having explicitly been denied access to the main group (cf. Huxham 2003). They also did not see the point in participating in the expert groups, as they perceived that they were not listened to by the STA. The interview study clearly revealed a lack of trust between the STA (and some other GNS members) on the one side, and road users’ organisations on the other.

A frequently expressed rationale for not including road user organisations (and others) in the GNS is that the number of participants needs to be carefully weighed against efficiency: “It’s about having the critical mass necessary for the discussions to function without it swelling up and getting too big” (STA Director). Another reason given was that all road user organisations would then need to be invited and that some groups might not be sufficiently well organised to be able to participate (cf. Ansell and Gash 2008; Newig et al. 2018). Organisational capacity was also mentioned as a problem by interest organisation representatives participating in the GNS:

The STA is after all a very large organization [about 9000 employees] where a huge amount is going on /…/ it’s hard for us other actors who don’t have the same personnel resources to participate in everything.

Respondents expressed that the GNS does not represent the full range of perspectives on road safety in society, but only those of a tightly knit group of individuals knowledgeable about VZ and supportive of its position as the dominant policy discourse in Sweden (cf. Scott and Thomas 2017). Actors inside the GNS reflect (self-) critically about this:

Perhaps there’s a lack of critics, causing the slowdown we’re seeing in the statistics [that the reduction in fatalities had leveled off], that a successful collaboration needs new eyes and a new type of input in order to improve, and regarding this I think we’ve reached a certain level, but when it comes to this we need a new way of thinking.

As a collaborative governance project primarily striving to enhance legitimacy and get external input on government policies, it is highly damaging to be perceived as having unclear inclusion rules, holding an entrenched worldview, and engaging in the calculated exclusion of troublesome actors and contradictory knowledge claims. As expressed by respondents, the collaboration will be at risk of stagnating, with few new ideas and insights (cf. Newig et al. 2018), and it may also generate contestation at other stages and reduce the legitimacy-generating effect of collaboration. Excluding critical voices and only including long-time members supportive of previous policies may make the GNS consensual and well-functioning, but doing so puts its capacity to innovate and rethink past and present policy decisions at risk (cf. Agger and Sørensen 2018).

Another area of critical discussion relates to the output from the group. When describing the role of the GNS, the STA clarifies that neither GNS membership nor participation in the development of policies and strategies involves any commitment on the part of participants to take action (STA 2018; Thoresson 2018). The GNS accordingly contributes to decisionmaking and gives non-state stakeholders an opportunity to influence public policy, but there are no demands on members to take action or commit their own resources. This makes membership attractive and there is little need for public managers to provide additional incentives for participation. However, as part of implementing the 2016 Renewed Commitment to the VZ (Government Offices of Sweden 2016) – a government initiative to reinvigorate the work with road safety – the STA strived to get the participants to clarify their commitments, even if they did not necessarily make concrete promises, and to remake the GNS to be “a bit more action-oriented regarding the things we discuss, so that it’s not [just] nice words about things that need doing” (an interviewed public official). Here, the importance of clear mandates and broad support in the relevant organisations was highlighted: “It’s also important that you aren’t just sitting there on some sort of personal mandate made of air, but are able to represent your organization” (GNS participant) (cf. Huxham 2000; Thoresson 2018).

TQ

As an example of collaborative governance, TQ was a project initiated and organised by the NRA/STA from the late 1990s to 2012. It was designed as a long-term, open-ended collaborative arrangement with the overall goal of supporting public policy objectives concerning energy efficiency, environmental protection and road safety. The public agency aimed to engage public and private organisations and companies to voluntarily work systematically to ensure the quality of their transports, e.g. by making informed vehicle choices and promoting eco-driving, and to formalise these commitments in voluntary agreements or joint declarations of intent. While several respondents stated that mutual learning was an important outcome of the project, its main purpose was to advance voluntary action by transport providers (taxi, bus, haulage) and public and private organisations and companies that purchased transport services (NRA 2009; STA no date a).

NRA/STA officials were at the centre of the project and described their work as “advocacy” (Knutsson 2009), which included to inspire, support, cooperate, and exchange knowledge with transport providers and purchasers. One respondent described their role as to “sell,” “raise awareness” and “sow seeds” because, as the respondent put it, “there were very many organizations that had never thought to set requirements about anything else than price and reliability.” In addition to advocacy, the NRA provided informative and economic incentives to “sweeten the deal” and build capacity amongst the participants. For example, they could supply free test versions of new technical systems, provide expert support for the development of management systems or road safety policies, offer free speed monitoring of company vehicles, and fund educational and information activities. However, as argued by one respondent, this support offered to individual companies was increasingly questioned in the highly competitive transport market and was used more and more restrictively.

The structures and procedures of TQ were highly diverse and dynamic, ranging from one-to-one deliberation and collaboration with individual companies to broad “business networks”. These networks, set up by the agency as a way to make their work more effective, convened companies from similar segments of the industry, e.g. forestry, groceries, and fashion, but could also be industry-wide (Andersson and Vedung 2008). Access to TQ was open to a broad set of actors, but the NRA/STA successively narrowed the target group for collaboration. Companies were invited based on agency officials’ assessments of where the effects of their efforts could be greatest. This resulted in their prioritising companies that had strong brands, significant turnover, many employees, and substantial transport operations (either as transport purchasers or providers), and that operated across Sweden (Knutsson 2009). The networks were convened and organised by the NRA/STA who offered their knowledge and expertise and acted as “neutral” facilitators. However, the forums also provided an opportunity for participants to share experience and ideas, help develop common declarations of intent, and facilitate joint action, such as following up on demands set during procurement.

In 2010, the TQ actively involved 185 companies, most through the business networks (STA no date a). One example was the “Forum for Sustainable Transport” initiated in 2006 and involving five industrial companies, aiming to “demonstrate and strengthen the connection between transport and Corporate Social Responsibility (CSR) and to develop a tool to help companies procure more sustainably” (STA no date b). As explained by one respondent, voluntary commitments by companies in the form of CSR have increased, but only rarely include transport. During the period 2006–2010, the network developed a set of demands and a web-tool for procurement. It was disbanded in 2010/2011 as the network became part of the member-based organisation Q3 (www.q3.se) (STA no date b; Andersson and Vedung 2008).

The TQ project was terminated in 2012. There was some hope that the companies involved would continue with the activities to ensure the quality of transport. Yet, in the end, all the networks and collaborations were closed down when the STA cancelled its support to the project. One respondent reflected on this as follows:

It was a bit naive to believe that the private business sector would take this baton. I’d say it shows you don’t understand very much about how things work in reality /…/ Why would they? What’s in it for me?

When reflecting on the TQ project in retrospect, the interviewed agency staff regard it as a success, and several also highlighted the importance of this type of close interaction between the authorities and companies, not least in order to gain an understanding of each other’s working conditions. The most critical issue for successful collaboration was argued to be finding the right incentives for the participating companies and organisations to take voluntary action. When the project was initiated, information and expert support were important “intermediate outcomes” of collaboration (cf. Ansell and Gash 2008, 561), as much of this was new to the companies as well as to agency staff. Managers involved in the project thought that the winning argument was that there was money to be saved by the companies. Eco-driving was promoted as “best practice,” reducing fuel costs and emissions and increasing road safety. As the project proceeded and companies became more knowledgeable, creating market pressure became a key strategy, and the focus shifted towards branding. The NRA argued that “proactive environmental and safety procedures enhance your company’s image as trustworthy and reliable,” which in the long-run enhances profit (NRA 2009). For participating agency officials, it became clear over time that the best way to influence transport providers was through the companies purchasing their services. As the manager heading the project stated, as long as no one demanded and paid for safe and environmentally friendly transports, there was no way for the transport companies to change. As argued by a former NRA Director:

If you want to get something done, you have to go higher up in the hierarchy […] go up-stream; go to where the conditions are set, go to the leadership, go to the money flows.

Critical respondents argued, however, that while the TQ successfully involved the companies’ transport managers and procurement officials, the project failed to engage the companies’ management level; “Transport ends up pretty far down in an organization, it’s more a question of economy /…/ In most organizations you don’t take these questions to the board of directors” (former NRA director). Interviewed public managers and industry representatives argued that efforts to change corporate behaviour in the long term had largely failed. In reflecting on the TQ project, one public manager concluded that, in the end, “the Swedish business sector never did their part.”

The second critical issue for the success of the TQ project was the importance of showing measurable effects and contributions to the national transport policy objectives. Several involved STA officials described internal and external criticism that successively built up during the last years of the project, concerning the lack of follow-ups and inadequate monitoring of effects. While arguing that the TQ did show measurable positive effects, for example, that the average speed was lower amongst TQ participants than non-participants, and thus that the project contributed to both environmental and road safety objectives (STA no date b), several respondents acknowledged the difficulty of measuring and demonstrating the effects of this type of collaboration. Even though more structured follow-ups, documentation, and monitoring of the effects of measures were developed during the later phase of the project, the critique from key constituencies on the lack of measurable outcomes supporting the national policy objectives (cf. Bryson et al. 2006) was a key argument for terminating the project.

Comparative analysis

The GNS and the TQ were both designed as long-term and open-ended collaborative governance arrangements in which the public agency NRA/STA engaged non-state stakeholders to promote road safety. However, the strategic purposes behind the two arrangements, as defined by the NRA/STA, differed. The GNS can be described as a forum where the public agency could gain support for public policymaking, while the TQ can be described as oriented towards implementation, with the public agency wanting to induce voluntary actions by non-state stakeholders.

When it comes to institutional design, both can be characterised as voluntary, consensus-oriented, and deliberative. However, critical design features differed. While the GNS is a relatively small and closed forum for a homogeneous group of high-level policy actors sharing a commitment to basic norms related to road safety, the TQ project was institutionally more complex, being open to a large and more heterogeneous set of public and private organisations and companies. In both cases, the initiation of collaborative governance was top-down, and by all accounts, the NRA/STA played a leading role (sponsoring, convening, structuring, leading). In GNS, managers have a largely facilitative leadership role, while in the TQ their roles also entailed advocating and incentivising participation and action.

The interplay between strategic purpose and institutional design highlights critical factors for the successes and failures of collaborative governance in these cases. In the GNS, exclusion was a critical factor. For a forum intended to generate legitimacy and provide input for public policymaking, unclear and limiting inclusion rules are highly damaging. In the TQ project, the difficulties primarily centred around creating incentives that would make participation attractive for companies and, most importantly, would get self-interested actors operating in highly competitive markets to commit to voluntary action. In comparison, GNS membership does not demand concrete commitments or actions by invited stakeholders. Essentially GNS was offering an arena for getting policy influence for free, and thus creating incentives to get non-state stakeholders to participate was not an issue. Another critical factor for the TQ was to retain government confidence and support by demonstrating its contribution to national policy goals. Its failure to show measurable outputs proved detrimental to the arrangement. Table 2 provides a summary of the institutional designs and strategic purposes of the two cases and factors identified as critical for their successes and failures.

Table 2. Summary of empirical analysis: strategic purposes and institutional designs

Conclusions

Policy practice and research are increasingly placing hope in the power of collaboration. What public managers expect to get out of collaboration, which largely defines the strategic purpose of collaborative governance, has important implications for the design of collaborative forums. By integrating two separate strands of collaborative governance theorising – institutional and managerial approaches – the article elaborates on an integrative approach targeting the interplay between strategic purposes and institutional design. New theoretical insights on the role of strategic purpose in understanding collaborative governance are developed by situating it in relation to societal goals and collaborative processes, as well as elaborating a new typology consisting of two types of collaborative governance based on differing strategic purposes (policy support and voluntary action). The integrative approach is used in an empirical analysis of two cases of collaborative governance, bringing new empirical insights on an important policy area (road safety) and illustrating how the fit between strategic purpose and important institutional design elements (incentives, structures and procedures, and public management) is a critical (but not the only) factor for the successes and failures of collaborative governance.

Placing the strategic purpose of collaborative governance (as defined by public managers) at the centre of theoretical and practical attention arguably better equips research to provide useful theoretical tools that can aid practitioners charged with designing collaborative arrangements to fulfil specific public purposes and help them avoid a mismatch between design and purpose. In Table 3, important lessons for practice and research on how to design collaborative governance that is fit for purpose are summarised.

Table 3. Lessons for practice and research

Apart from providing new insights of practical importance, this article brings important insights for collaborative governance theorising and research. Collaborative governance theories need to (a) upgrade the importance of the strategic purpose and actions of public managers in explanations of institutional diversity and the outcomes of collaborative forums, (b) further differentiate between different types of collaborative governance when theorising on suitable institutional designs and their impacts, (c) engage more critically with collaborative governance, acknowledging that the positive normative underpinnings of collaborative governance often expressed in academia risk influencing practice to use this instrument even in situations where it is not suitable, or for purposes that are not in the public interest, and (d) give closer consideration to the methodological challenges involved in getting behind the facade of these forums, recognising, amongst other things, the importance of research on collaborative governance integrating multiple theoretical perspectives generated from across disciplines (public policy, public management, public administration). Hopefully, the theoretical arguments and empirical analysis presented here will inform continuous theoretical elaboration on collaborative governance along these lines with the aim of helping practitioners design collaborative governance that is fit for purpose, and thereby will lay the foundation for more effective and legitimate governance.