Meta-governance as partial organization

Abstract This paper develops a decision-based approach to meta-governance by drawing on insights from studies of organization outside of formal organizations. We argue that meta-governance can be analyzed as a form of partial organization of interactive arenas. Meta-governance has become an important concept in theorizing the role of government in contemporary governance, particularly in the context of shaping and providing guidance to different interactive arenas, such as networks and quasi-markets. Meta-governance is, however, also a broad and ambiguous concept, which limits its ability to guide empirical research on the actual practices of governing and the actors involved. Rather than presenting a new perspective on meta-governance, the framework of partial organization enables analysis of the processes behind the formation of meta-governance strategies. By focusing on decision-making, it offers a dynamic understanding of the stepwise development of meta-governance, reflecting an emerging, rather than predefined, rationality of governance arrangements.


Introduction
Governing has always been associated with creating control, coordination, and collective action. This task has, however, become increasingly challenging for government actors, who strive to retain control over a public sector characterized by dispersed power, "wicked" problems, uncertainty, and complex interdependencies (Ansell, Trondal, & Øgård, 2017;Klijn & Koppenjan, 2016;Newman & Head, 2017). To grasp this development and to theorize changes in how governing actors seek to steer society and the economy (Peters, 2010), scholars have developed the concept of meta-governance. Meta-governance is sometimes broadly described as the "governance of governance," and refers to the practices of (mainly) public authorities in coordinating different modes of governance (i.e., hierarchy, network, and market governance) and, more specifically, in providing guidance and some level of control over various "interactive" arenas, such as collaborative arrangements, partnerships, and quasi-markets (Gjaltema, Biesbroek, & Termeer, 2020;Torfing, Peters, Pierre, & Sørensen, 2012). The popularity of the concept is reflected in a growing number of publications and empirical applications across a wide range of policy areas. These areas include healthcare (e.g., Hammond et al., 2019), crisis management (e.g., Larsson, 2017;Stark, 2015), environmental policy (e.g., Fischer & Schl€ apfer, 2017;Hedensted Lund, Sehested, Hellesen, & Nellemann, 2012), natural resource management (e.g., Bell & Park, 2006;Kooiman & Jentoft, 2009), education policy (Hooge, Waslander, & Theisens, 2021), and employment and social policy (e.g., Damgaard & Torfing, 2010;Qvist, 2017). This paper develops a framework for analyzing the practices and processes behind the formation of meta-governance arrangements. While meta-governance captures an essential part of contemporary governance, it is also a broad and ambiguous concept. The most common conceptualization is to see meta-governance as a set of strategies that together cover a range of governing activities and interventions: from institutional design to active participation in interactive arenas (e.g., Hooge et al., 2021;Klijn & Koppenjan, 2016;Sørensen & Torfing, 2007. At the same time, the actual processes by which meta-governance strategies are created and change over time constitute a theoretically underdeveloped area of research. This is reflected in a recent review of the meta-governance literature, which found that both the "how" (i.e., the governing practices) and the "who" (i.e., the governing actors) are often not clearly defined in empirical research (Gjaltema et al., 2020).
This suggests, we argue, a need for analytical tools geared toward the decision-making behind meta-governance. In the wider literature on public administration, there is a growing interest in processual approaches (see, e.g., Mukhtar-Landgren & Paulsson, 2021) and in using frameworks from organizational studies to theoretically enrich administrative research (Egeberg & Trondal, 2018;Jacobsson et al., 2015;Whitford, Milward, Galaskiewicz, & Khademian, 2020). As argued by Whitford et al. (2020, p. 81), an organizational lens helps us understand "governance in action" by drawing attention to "the enactment of processes and practices that are evolving and contextually contingent and mutually constituent." Interactive arenas, such as governance networks, are organized entities (Kenis & Raab, 2020), so organizational theory provides useful tools for analyzing the governance of such arenas.
More specifically, we argue that recent work on organization outside formal organizations offers a particularly relevant framework for the study of "meta-governance in action." This theory of "partial organization" (Ahrne, Brunsson, & Seidl, 2016;Ahrne & Brunsson, 2011) is useful for understanding attempts to create structure and order in the spaces between different formal organizations, such as the formation of interactive arenas. By linking meta-governance to partial organization, this paper introduces a processual view that focuses on decisions regarding various elements of organization. It therefore serves as both a conceptual framework and a methodological guide for studying meta-governance as decision-making over time. This helps to highlight both the "how" and the "who" in evolving meta-governance processes.
After discussing previous conceptual work on meta-governance and presenting the organizational framework, we illustrate how it can be used to guide empirical research with the help of two empirical examples. The paper concludes with a discussion of the contribution of a decision-based approach and by pointing out some promising themes for future research.
The multiple meanings of meta-governance When introduced in the 1990s, researchers used the concept of meta-governance at a quite abstract level of analysis, focusing on the shaping of the foundational norms, principles, and ground rules of governance (Jessop, 1997;Kooiman, 1993). Conceptual work during the 2000s (e.g., Sørensen & Torfing, 2007) connected meta-governance to research into the management of networks (e.g., Kickert et al., 1997), introducing a more actor-centered perspective on strategies applied by politicians and public managers. In later research, the concept has also been situated more firmly in the context of public-sector reforms and is now widely used to theorize the attempts of governing actors to steer devolved governance processes (Peters, 2010). This task has become increasingly important with the proliferation of interactive arenas in the public sector such as quasi-markets, partnerships, and governance networks (Torfing et al., 2012), and the emergence of new policy areas where traditional structures and measures are no longer considered viable (Mukhtar-Landgren & Paulsson, 2021).
As argued by Gjaltema et al. (2020), the concept of meta-governance can be seen as a way of reconciling the old "from government to governance" debate, by acknowledging both the autonomy of devolved governance processes and the continued centrality of the government. At the same time, there continues to be a major difference between perspectives on meta-governance based on the associated views of the primacy of the state (Stark, 2015). Whereas some tend to emphasize the transformational capacity of interactive arenas (Torfing et al., 2012) and governance networks (Klijn & Koppenjan, 2016), others apply a more state-centric perspective in which autonomy and devolution are shaped under the authority of government actors (Bell & Hindmoor, 2009;Jacobsson et al., 2015). Furthermore, the concept is often combined with various theoretical approaches, such as rational actor perspectives, different strands of institutional theory (see Sørensen & Torfing, 2007), governmentality (e.g., Larsson, 2017), and, in some cases, organization theory (Egeberg & Trondal, 2018;Jacobsson et al., 2015).

Defining meta-governance
In developing theory on the decision-making processes behind meta-governance, we seek to build on existing definitions of the concept. The aim is not to develop an alternative theory, but to offer an approach for "bringing down" meta-governance research to the level of decision-making and practice. To be more precise as to the practices we have in mind, we start from a definition of meta-governance that is inclusive yet more specific than the notion of simply the "governance of governance." Based on their literature review, Gjaltema et al. (2020, p. 12) provided a definition that fits this ambition, defining meta-governance as "a practice by (mainly) public authorities that entails the coordination of one or more governance modes by using different instruments, methods, and strategies to overcome governance failures." As noted by Gjaltema et al. (2020), meta-governance often focuses on the governance of networks (or networks of networks). In practice, however, governance modes are often intertwined; for example, networks often include a considerable element of hierarchy, and contractual arrangements are often intermixed with different types of collaborations. To conceptualize the "target" of meta-governance, we use the notion of "interactive arenas," suggested by Torfing et al. (2012), which includes not only governance networks but also different types of partnerships and (quasi-)market arrangements. While broader than the notion of "the governance of governance networks," this understanding of meta-governance is still narrow in that it focuses on concrete governance arrangements and not on decision-making regarding broader and more abstract governing principles.
Based on this understanding, meta-governance can be understood as a process of balancing discretion with control (Peters, 2010;Voets, Verhoest, & Molenveld, 2015). An interactive arena implies the creation of a "governance space" where market or network actors are given room to maneuver within regulative, normative, and discursive frames that are supportive of problem-solving "from below" (Qvist, 2017;Skelcher, Mathur, & Smith, 2005;Torfing et al., 2012). This actor autonomy is sometimes discussed in the literature as a bottom-up process of "self-organization." As argued by Bell and Hindmoor (2009, p. 46), however, devolution and the relocation of authority are often driven by government actions. Hence, while interactive arenas may sometimes evolve spontaneously, the creation of a governance space often involves the externalization of processes and capacities from public-sector organizations. In some cases, this externalization may be limited to service provision only, for example, through contracting out. Often, however, especially in quasimarkets and networks, it includes the shifting of capacities for decision-making, planning, coordination, problem-solving, auditing, and sanctions to external actors and arenas. Such externalization implies that meta-governance requires attempts to steer and exercise control in spaces created outside of, and between, government organizations (see also Agranoff, 2014).
These attempts have been the main theme of the meta-governance literature and are usually conceptualized as meta-governance strategies. These strategies are developed to maintain some level of control over the processes within a governance space (Peters, 2010). In other words, for meta-governance to occur, there needs to be both a governance space that enables interaction and an active attempt (by a meta-governor) to steer, or create structure for, this governance space. The challenge, according to Torfing et al. (2012, p. 144), is to find ways to "facilitate, manage, and direct interactive governance arrangements without reverting to top-down command, and without undermining the self-regulatory capacity of networks, partnerships and quasi-markets." A useful and widely adopted categorization of meta-governance strategies is the distinction between "hands-off" and "hands-on" meta-governance: hands-off strategies are exercised at a distance and include framing and design; hands-on strategies entail the more active facilitation or participation in interactive arenas (Sørensen & Torfing, 2007;Sørensen & Torfing, 2017).
As actors, meta-governors are sometimes defined at an individual level (e.g., politicians and public managers), and sometimes at an organizational level (Gjaltema et al., 2020). Meta-governance strategies are often studied from a design perspective, which emphasizes the intentionality of meta-governors. For many research purposes this is, of course, a highly useful approach. However, if we want to delve deeper into how these strategies come about, and how they develop over time, we need theories that unpack the decision-making "hidden" behind the strategies of meta-governance.

An organizational perspective on meta-governance
Organizational theory is increasingly used by public administration scholars to develop a deeper and finer-grained analysis of phenomena such as implementation (Sandfort & Moulton, 2020), networks (Kenis & Raab, 2020), and public-sector reforms (Christensen & Laegreid, 2002;Jacobsson et al., 2015). Primarily, we want to introduce a process view of organization into meta-governance research, to refer to existing definitions and frameworks in this literature, which tend to apply a more static approach to actors and governance structures. The process view of organization is the subject of ongoing discussion in organizational theory; it offers various possible approaches, some more far reaching than others in their views of entities, such as actors, as constitutive of process (Hernes, 2014). Here, we suggest that a decision-based organizational perspective can be particularly relevant to conceptualizing the practices and dynamics of meta-governance. More specifically, we draw on the literature on organization outside organizations-or partial organization (Ahrne & Brunsson, 2011, 2019. This theoretical approach is processual but retains a focus on actors and positions that exist prior to process. It draws on the seminal work of March and Simon (1958) on decision-making as the key object of study in organizational research. At the core of the partial organization literature lies the ambition to theorize the attempts to create structure and order in the spaces between different formal organizations, and to do so by applying a distinct approach to organization that places decisions at the center of analysis. To pursue this approach, we must first clarify the concept of organization.

The concept of organization
The argument of Ahrne and Brunsson (2011) assumes that "organization" is defined differently from the traditional meaning of the term. The organization-as-actor notion, which is widely taken for granted in research, emphasizes (formal) organizations as collective actors. This claim is based on the assumption that organizations must be recognized or constructed as actors capable of interacting with other actors (e.g., Geser, 1992;King, Felin, & Whetten, 2010;Meyer & Jepperson, 2000). In contrast, Ahrne and Brunsson (2011) suggested a decision-based perspective in which organization is seen as a particular social order, "a decided order," that enables a separation between organization as an analytical concept and organizations as empirical phenomena. This not only attributes to organization a distinct meaning (Grothe-Hammer, 2019, p. 326) but also allows the acknowledgment of complex organizational phenomena without particular organizations as their obvious setting or, expressed differently, organization may be found inside, outside, or among formal organizations (Ahrne & Brunsson, 2011).
The perspective draws on the classics of organization studies and emphasizes decisions as the most fundamental aspect of organization (March & Simon, 1958), while challenging the distinction between organization and environment. Thus, the notion of organization as a decided order entails both a broadening and narrowing of meaning compared with the traditional organization-as-actor notion. It is broadening in that the concept is not restricted to formal organizations but encompasses decisions about organizational elements extending beyond these, permitting the notion of "partial organization" (Ahrne & Brunsson, 2011) or "degrees of organizationality" (Dobusch & Schoeneborn, 2015).
It is narrowing in that ways of ordering that are not decided on within or between organizations, but that can very well structure interaction, are excluded from the definition. Consequently, all forms of informal organization are left for other concepts to cover. This conceptualization of "organization" has some similarities to the concept of "organizing," as used by Weick (1979Weick ( , 1995 and Czarniawska (2004), but also important differences. There is a shared argument for not taking formal organizations as a point of departure, but instead studying the actions and interactions that (potentially) build them up. The difference, however, is that organizing may include any intentional actions, or the production of any social order, which makes it a much wider concepttoo wide according to Ahrne et al. (2016): We consider this suggestion problematic in that it undermines the potential for organization studies contributing a distinctive perspective on the social world-a perspective that would also be useful in other disciplines. … In contrast, the proposed concept of organization as "decided order" allows for the transfer of the term to other domains outside formal organization, while simultaneously preserving its distinctiveness. (p. 95) Accordingly, we find that a decision-based understanding of organization offers a fruitful and sufficiently narrow conceptual apparatus to enable the analysis of meta-governance.

Elements of organization
A decision-based understanding of organization concentrates on decisions regarding the required foundations for repeated interaction, which are typical of formal organizations: membership, rules, hierarchy, monitoring, and sanctions. Ahrne and Brunsson (2011) referred to these as "elements of organization," which therefore constitute the objects of decisions. Membership is central to formal organizations but can also exist outside formal organizations, for example, in the guild system, in which membership decisions are made in a market context. In the context of collaborations and networks, criteria for the inclusion and exclusion of partners or stakeholders are another example of membership criteria. Organization also entails hierarchy, which implies that some have the initiative and power, i.e., the authority, to make decisions that others need to comply with. Interactive arenas, such as governance networks, are sometimes governed by an administrative organization, sometimes created voluntarily by the members, and sometimes as a mandated feature in the process of network formation (Provan & Kenis, 2007). Another element of organization is rules, which are codified norms (decided upon and written) of how to act. Typical rules that exist beyond formal organizations are "standards," which are voluntary rules that one chooses to comply with (often produced by international standardization organizations with no formal authority). Furthermore, the ability to monitor compliance with rules and commands is an element of organization. Ratings and rankings are examples of monitoring that takes place outside the realm of formal organizations. Monitoring can be one of the tasks of a shared administrative organization in a network setting. Finally, organizers may decide upon sanctions, which may be both positive and negative. Examples of sanctions are grading systems, awards, and penalties. Not all these elements need to be present for a situation to be characterized as organization, which is what the distinction between complete and partial organization emphasizes. Hence, organization can refer equally to the spaces between formal organizations as to the orders within them. From this point of view, the degree and character of organization are empirical questions, because organization, in practice, occurs through decisions regarding one or several organizational elements that are connected to one another as well as recursively connected through time.
Building on Ahrne and Brunsson's (2011) idea of partial organization, several researchers have analyzed how organizational elements structure the fields of, for example, standards and standardization (Gustafsson & Tamm Hallstr€ om, 2018), social movements (Den Hond, De Bakker, & Smith, 2015), inter-organizational cooperation (Grothe-Hammer, 2019), crowdfunding (Nielsen, 2018), and markets (Brunsson & Jutterstr€ om, 2018). There are also studies of how configurations of organization change over time (Ahrne, Aspers, & Brunsson, 2015;Ek € Osterberg & Qvist, 2020). While focusing on different phenomena, these studies share the aim of exploring the organization and organizers of spaces between formal organizations, and the consequences this entails. These studies explain how, why, and by whom "fluid social collectives" (Dobusch & Schoeneborn, 2015) are created, upheld, and transformed through the successive introduction of various organizational elements, as well as the development of the decision-making capacity of these collectiveseven when they are not externally recognized as actors (Grothe-Hammer, 2019).
Furthermore, decisions can be seen as a particular form of communication (Luhmann, 1995) in that they convey information about specific content ("this has been decided") and about the fact that a selection has been made ("there were options"). Decisions make visible their own contingency while simultaneously "absorbing uncertainty" (March & Simon, 1958), in that they may be taken as given in ensuing communications and decision-making. This simultaneous uncertainty reduction/creation is essential to decision-making and thus essential to bringing nuance to strategies of meta-governance understood as decisions regarding organizational elements. Meta-governors must address both the uncertainty that results from making the existence of alternatives visible, and the fact that decisions can limit the alternatives available for further decisions.

A research approach
As a research approach to meta-governance, the decision-based framework of organization suggests that we should study the management of governance spaces the same way that we would study the management of formal organizations (cf. Ahrne & Brunsson, 2011). The elements of organization (i.e., membership, rules, hierarchy, monitoring, and sanctions) serve as a conceptual framework for determining the degree and character of organization in governance spaces. However, rather than simply identifying different elements of organization, the analysis should focus on how different configurations of elements emerge and how they develop over time. Moreover, it is important from a governance perspective to identify the key actors-the meta-governors-in the process, as well as the underlying rationale for the creation of meta-governance arrangements.
Studying meta-governance as partial organization has a very practical implication for research design. It suggests that governance should be studied as continuous decision-making and, consequently, that meta-governance can be understood in terms of decisions regarding organizational elements within a governance space. In contrast to a "toolkit" understanding of governance, the decision-based approach emphasizes a processual and dynamic understanding, reflecting an emerging, rather than predefined, rationality of governance arrangements. As decisions are interconnected, we need to study governance longitudinally, or-in other words-how decisions regarding organization elements configure and reconfigure governance spaces over time. Hence, the decision-based approach may serve equally as a methodological guide and as an analytical framework. Table 1 summarizes how meta-governance can be investigated and analyzed as a decided order. In addition to the elements of organization, the framework includes a set of analytical questions used to study the decision-making process. These questions connect to the key themes ("what," "why," "who," and "how") that, as suggested by Gjaltema et al. (2020), need to be covered empirically in any analysis of meta-governance.
Next, we draw on two empirical examples to briefly illustrate how the framework can be used to guide empirical research and, more specifically, how it can be used to capture the emergence of meta-governance arrangements. The examples are taken from studies of interactive governance in the areas of immigrant integration and transport planning in Sweden. Both cases are characterized by policy implementation taking place through the joint action of multiple actors, with the state applying a variety of metagoverning strategies that are well documented in research.

Studying strategies of meta-governance as decision-making
Undertaking a micro-level analysis of meta-governance requires that we break down broad governance strategies into a series of decisions regarding organization. Depending on the research question and the context of study, this research task can take many different forms and focus on various aspects of governance processes. Our purpose in this section is to illustrate the logic behind a decision-based approach, and not to conduct full case-study analyses. In the next section, we discuss some of the analytical gains from the presented organizational framework.
When presenting our two examples, "social procurement" in immigrant integration and "Strategic Choice of Measures" in transport planning, we outline the main characteristics in each case, beginning with the object of meta-governance (what), turning to the underlying logic or rationale behind the meta-governance strategy (why), and then identifying the main meta-governors (who). Last, we provide an overview of the main decisions regarding organization behind the means of meta-governance (how). We use the examples here merely for illustrative purposes. However, they stem from extensive field studies within two research projects whose empirical contributions have been reported elsewhere (Ek € Osterberg & Qvist 2020; Ek € Osterberg; Norb€ ack; Zapata, & Zapata Campos, 2021). Both projects were intended to create a nuanced understanding of governance practices, through qualitative studies using interviews and archival data as the primary sources of data. 1 In these studies, the decision-based framework gradually came to be the lens through which the empirical material was interpreted and served, we argue, as a useful tool for analyzing how strategies of meta-governance developed at the micro-level and the dynamics by which multiple actors became involved.

The example of social procurement in immigrant integration
In immigrant integration, a longstanding challenge has been to facilitate the involvement of the private sector and create incentives for employers to offer local labor market opportunities to newly arrived immigrants. In Sweden, refugees and their families constitute a large proportion of immigration. While collaboration in local partnerships is the main strategy in this endeavor, the introduction of social procurement creates a new arena where local authorities can further motivate, through contractual requirements, private firms to enter such partnerships. Here, the targets of governance (what) are networks of public and private actors, particularly in situations where private firms provide, through public contracts, goods and services to local authorities. Applying social procurement means that such contracts become conditional on labor market integration efforts, such as employment or internship for individuals facing difficulties in the labor market. To enable this, purchasing units and public procurers need to arrange to include employment criteria in their procurements, and providers need to support the initiative. Facilitating such socially conscious market exchanges, and thereby enabling procurement to be a lever of social reform, is the goal of the meta-governor(s) in this context. Hence, the rationale behind making use of social procurement (why) is to increase the governing capacity of the public actors and increase the pressure on local businesses to act socially responsibly by, in this case, offering employment opportunities to individuals outside the labor market.
In Sweden, many local authorities have decided that procurement should be used more frequently as a way of meeting local challenges in labor market integration, not least in the wake of the 2015 refugee crisis in Europe. These decisions are commonly manifested in overall political goals, targets for social requirements, and the development of social procurement strategies reflecting popular administrative policy ideas of using procurement to achieve societal outcomes. The main meta-governor (who) in this case is the local authority striving to increase the use of the opportunities available in procurement. However, seen in the light of local integration policy, it is a governance strategy that brings a series of new meta-governors to the field with the potential to reshape the governance spaces: the state (particularly the National Procurement Agency), through guidance on what could and should be done in relation to EU internal-market principles; the EU Commission, through soft law on how to gain support for and successfully implement social procurement practices; contractual rules, once requirements are set in specific procurements; and the European Social Fund, through financing and requirements for follow-up-to mention just a few.
Turning to the decisions regarding organization (how), the process began with local authorities deciding to use social procurement in immigrant integration. This introduced a number of new organizational elements into the local integration policy arena, which in turn created a need for additional decisions regarding how to manage and support the new procurement arena. In Gothenburg Municipality, widely seen as a "forerunner" in social procurements, decisions were made at an early stage to develop a process description for the various steps in the social procurement process. The municipal authorities also decided to create a support function to assist procuring units and private providers in working on preparation, contractual issues, matching activities, and follow-ups. In effect, the support function became an important hierarchical structure exerting major influence on what individuals were prioritized and what procurements were seen as suitable for social criteria. The support function was behind a number of successive decisions regarding rules for how to arrange interactions between purchasers and providers and between potential employers and employees. Over time, routines for monitoring and follow-ups were added, to credibly describe the scope and effects of the social procurement initiative. This was important, because it was not mandatory for private firms to apply the social criteria or to follow the process description. The support function became central to the spread and justification of the model, and intense activities took place to gain support for the model, with the intention of attracting as many employers as possible to embrace the idea. Arguments that the model was a win-win-win situation, a tool with enormous untapped potential, and a shortcut to securing recruitment needs for local businesses (especially those experiencing competence supply difficulties) spread and the number of employment opportunities steadily increased.
In sum, the use of social procurement in the meta-governance of immigrant integration in Gothenburg was based on a configuration of mainly soft rules, monitoring, and positive sanctioning, channeled through the support function, which has become a key organizer of social procurement. This hierarchical setup was never challenged, despite changing conditions for the work. Applying the decision-based perspective reveals how the space gradually developed into an increasingly formalized system, not least by linking the local interactive arena for procurement to new organizing elements at the national and international levels.

The example of Strategic Choice of Measures in transport planning
Our second example is taken from the area of transport infrastructure planning in Sweden. In recent decades the communicative model of planning (see, e.g., Innes & Booher, 2010), which emphasizes broad involvement of stakeholders, has become increasingly influential in the Swedish planning system. In 2013 the government decided on a new format for the early stages of planning, called Strategic Choice of Measures (SCM). This governance effort created a new interactive arena that targeted (what) both public authorities, such as regional and municipal authorities, and various stakeholders from the private sector and civil society, with the aim of promoting open and creative problem-solving in which the actors considered all possible solutions, including those that did not involve costly new construction.
The idea behind this reform (why) was to address transport problems from a multimodal perspective rather than focusing on one particular mode of transport (i.e., rail, road, air, or sea). The aim was also to avoid the tendency of path dependence in transport planning, in which key actors commit to expensive new construction at a very early stage, without considering other alternatives, such as innovative use of existing infrastructure or measures to change transport behavior.
The meta-governors in this case (who) are primarily the central government, responsible for the overall regulatory framework, and the Swedish Transport Administration (STA), which is the government agency responsible for managing the whole field of SCM processes around the country. The process of establishing this meta-governance arrangement (how) began with the government decision to create a rule that an SCM study was mandatory for all proposals for future infrastructure investments to be considered for government funding. In effect, the introduction of this organizational element created a whole new arena for planning. By linking the SCM to possible funding, the rule created a strong incentive for the formation of collaborative processes. After the reform was decided, managing the SCM field became a task for the STA. Over time, the STA has made a series of agency-level decisions to introduce additional elements of organization into the SCM field. These include a set of recommendations, or guidelines, for how the interactive processes should be organized. Of particular importance was the introduction of a distinction between "actors" and "stakeholders," with only the former (i.e., public agencies with responsibilities in transport planning) being granted decision capacity in SCM processes. In effect, this became a way to regulate membership in the SCM field. The STA also operates a project management system for all SCM processes, adding an element of hierarchy. In the role of project manager, the STA can decide on the duration, setup, and financial resources of each study.
To summarize, the meta-governance of SCM is based on a configuration of membership, hierarchy, and rules that has developed since 2013. The initial configuration, based on only a few rules, had a strong impact on the initiation of a large number of interactive planning processes. The STA, as the main meta-governor, has subsequently increased the degree of organization, giving itself more capacity to regulate the length, setup, and character of the SCM processes. This configuration has, however, also reduced the level of interactivity and the involvement of nonpublic actors.

Analytical gains from the organizational approach
The previous section described the research approach of studying meta-governance in terms of decisions regarding organization. It also introduced a processual understanding of interrelated decisions, illustrated in the two empirical examples. This section elaborates on the possible analytical gains from applying the organizational approach. We focus on two broad themes that we believe are of general relevance to research into the practices and processes behind meta-governance: (1) the configuration of the organizational elements constituting meta-governance strategies, and (2) identifying meta-governance actors. Again, the two cases of social procurement in integration policy and transport planning are used to illustrate the discussion.

Meta-governance strategies as configurations of organization elements
In both our examples, there is a clear desire to create, but also control, interactive arenas where various actors collaborate in the implementation of policy. This balancing act is a core feature of meta-governance strategies, but this is often an ongoing and evolving process. Initial decisions regarding creating a new governance space, or shaping the conditions of an existing one, are crucial and largely formative for subsequent events. In the example of transport planning, the collaborative processes of SCM created not only a new step in the Swedish planning system, but also an entirely new field of interaction between public and private actors as well as a new market for consultancy services. The latter includes services both in process management and in formulating the study reports on the proposed transport measures. In the case of social procurement, the situation was somewhat different. The decisions of local authorities to make procurement part of integration policy extended an existing arena for governing local integration to include the private sector in a more profound and systematic way.
What the cases have in common, however, is that the creation of a new governance space was only the first step in the development of what we, in retrospect, can identify as a meta-governance strategy. What we see at the micro-level, however, are decisions regarding elements of organization, such as rules in the case of SCM. Over time, elements of organizations are often added, combined, and sometimes removed. Meta-governance strategies can thus be understood as particular configurations of organizational elements.
The arena created to enable increased use of social procurement entailed several organizational elements: the procurement regulation and international soft law, as well as monitoring and sanctioning through local-level hierarchical solutions created to support implementation. Here, meta-governance was based on a configuration of mainly soft rules, monitoring, and positive sanctioning channeled through the support function, whose role as the key actor also formed part of the initial decisions regarding how to arrange the arena.
In the case of transport planning, the reform creating SCM was followed by a surge of SCM processes around the country (about 200 SCM processes are ongoing each year). For the main meta-governor, the STA, this created problems in terms of control: processes sometimes became lengthy, lacked a multimodal perspective, and, in many cases, there were difficulties deciding on appropriate transport measures (Tornberg & Odhage, 2018). The STA responded by issuing guidelines (soft rules) regulating mandates and access to the planning process (membership) and subjecting the SCM processes to a project management system (hierarchy). The initial configuration, based on only a few rules, had a strong impact on the pattern of network formation, and later configurations substantially increased the degree of organization, giving the STA more capacity to regulate the length, setup, and character of the SCM processes.
A similarity between our examples is that meta-governance is an emerging decided order, in which decisions both lock in and enable future courses of action. An initial creation of a governance space was followed by decisions restricting the level and scope of interactions.
In the case of SCM, meta-governance was clearly a way to retain some sort of control over planning processes that risked becoming all too complex and creating excessive demand for costly transport solutions. In the case of social procurement, meta-governance was a way to promote and motivate a particular course of action, i.e., to utilize the enhanced governing capacity that procurement offers. Through step-by-step decisions about organizational elements, both examples illustrate the continuous transformation of the characteristics of meta-governance, which can be explained by the interconnectedness of decisions over time and the local dynamics of which they are part.
The examples are used to emphasize development over time, but configurations of organizational elements, i.e., rules, hierarchy, membership, monitoring, and sanctions, can also be analyzed using a "snapshot" approach. By focusing on a formative or important moment, the framework of organizational elements can be used to unpack a particular configuration and provide the basis for an in-depth analysis of the consequences for the functioning of the interactive arenas to which the organizing attempts are directed. While comparative analysis admittedly falls outside the scope of this paper, the empirical illustrations indicate that such configurations could be fruitfully used to analyze similarities and differences through cross-case comparisons.

Explicating the actors of meta-governance
It is already well known that meta-governance often involves multiple actors, rather than a single, unitary meta-governor (see Gjaltema et al., 2020), so identifying the metagovernor in a particular context can be more complicated than it may initially seem. Our two cases differ in this respect. Both involve several meta-governors, but the example of transport planning is more straightforward because it reflects the traditional structure of a top-down implementation process. The new SCM step in the planning system was initiated by the government, passing a law in parliament, and the task of managing and providing overall coordination of interactive processes was then handed over to the STA.
The case of social procurement is much more complex. It involves several public organizations and levels of government, but it also has an important bottom-up dimension. In Sweden, the decision to use social procurement in immigrant integration policy resides with the semi-autonomous municipalities. It was these local authorities that introduced new elements of organization, connecting public procurements to the existing work of finding employment for newly arrived immigrants. Furthermore, these local organizational decisions created new linkages to other meta-governors and other levels of government. In other words, the local adoption of social procurement created-from the bottom up-an arena for the multi-level meta-governance of integration measures. First, it activated a role for the National Procurement Agency, which is a key meta-governor of social procurement in Sweden. The Agency applies several organizational elements to provide guidance to local arenas, mainly in the form of soft rules and positive sanctioning through widespread narratives of good examples. Moreover, the various EU institutions regulating the field of procurement became important arbiters of what should and could be done, and of the limits in relation to EU internal-market principles. Local authorities needed to handle the activities within their interactive arenas but also take account of the decisions of these distant actors.
This brief overview of the meta-governors in social procurement suggests that it can sometimes be difficult to start a research process by defining who the key meta-governors are. In the literature, the meta-governance actors are defined in different ways. Some studies focus on individuals, but it is also common to apply the notion of organization-as-actor. The organizational approach, however, suggests a different point of departure. Drawing on the classics of organization theory, this approach places decisions at the center of attention (March & Simon, 1958). Starting with a decision, such as the transport planning reform of 2013 or the adoption of social procurement in Gothenburg, the researcher can then trace actors from decisions. This will sometimes be fairly simple, but when we consider that decisions are often interrelated, and sometimes occur at different levels, identifying the actors of meta-governance can be a research task in itself.
Studying meta-governors from the perspective of decision-making also draws attention to the relationships between actors. This may include relationships between different meta-governors, but also relationships and interactions with the objects of meta-governance strategies, i.e., the envisioned participants in interactive arenas. Inevitably, the study of such relations leads to questions about accountability and the democratic legitimacy of meta-governance. As pointed out by organizational theorists, decisions make visible their own contingency: choosing one course of action means that another course of action could have been chosen, but was not. For example, creating a governance space for interactive processes can sometimes be a controversial choice over a more traditional, expert-driven system. In addition to the "who" question in studies of meta-governance, the focus on decisions sheds light on such contingencies and the accounts of the responsible meta-governors.

Conclusion
While meta-governance captures an essential aspect of contemporary governance, knowledge is scarce regarding the micro-level dynamics behind meta-governance strategies, i.e., how these strategies emerge and change over time. With inspiration from research using organizational theory in public administration (e.g., Egeberg & Trondal, 2018;Jacobsson et al., 2015;Kenis & Raab, 2020;Whitford et al., 2020), we suggest that the literature on partial organization (Ahrne et al., 2016;Ahrne & Brunsson, 2011) offers an analytical lens for the analysis of meta-governance practices and processes. Governance and the more specific notion of meta-governance have become key concepts in administrative theory and research, although they remain rather broad and imprecise concepts for guiding empirical research. The partial organization approach enables us to come closer to the actual practices and actors involved in metagovernance, which has been pointed out as central to future research (Gjaltema et al., 2020).
What the literature on partial organization offers, more specifically, is a conceptual and analytical framework that puts decisions regarding organizational elements at the center of analysis. We argue that meta-governance can be understood as a configuration of organizational elements, reflecting an emerging decided order in governance spaces characterized by interactive processes and arenas. The two cases of social procurement and transport planning in Sweden are examples in which stepwise decisions regarding organization created increasingly formalized systems of governance. They also illustrate how the decision-based approach can be used to identify the key meta-governors. In this respect, the transport planning case reflects a more traditional top-down process, whereas the local decisions regarding social procurement created an area for multi-level meta-governance.
A conclusion of the paper is that administrative research into meta-governance may benefit from the decision-based organizational framework in three main ways. First, the framework offers a research approach to break down broad governance strategies into a series of decisions regarding organization, with a conceptual apparatus sufficiently narrow to enable micro-level analysis. While case studies may be criticized for being fragmented and all too context specific, the framework could serve as a basis for comparing settings and thereby specifying under what conditions and how certain forms of meta-governance develop. Configurations of organizational elements, and the development of these configurations over time, could be a promising analytical focus for such comparative studies. Second, it enables researchers to account for the dynamics and contextual features of governing processes, which are sometimes overlooked in studies of meta-governance. This could contribute to a deeper understanding of developments over time and the sequential character of decisions related to meta-governance, and of the links created between various meta-governors. Third, the framework visualizes who the meta-governors are by following decisions regarding organizational elements and tracing actors from decisions.
The processes behind meta-governance can provide important insights into the ongoing reforms of public administration, especially the proliferation of interactive arenas such as governance networks and market arrangements. A better understanding of these processes can also suggest new themes and questions and create links to other areas of administrative research. One example is accountability, which is already an important topic in meta-governance research (see, e.g., Hammond et al., 2019;Skelcher, Klijn, K€ ubler, Sørensen, & Sullivan, 2011), but is emphasized by a focus on decisionmaking rather than governance strategies. Decisions are communicative acts that reveal their own contingency, and how meta-governors handle (or try to reduce) the uncertainty that results from making the existence of alternatives visible is of major importance from an accountability perspective.
Connected to accountability is the wider issue of the democratic legitimacy of metagovernance. Governance reforms are often evaluated in terms of effectiveness and output legitimacy. Research has also focused on the legitimacy and "democratic anchorage" of interactive arenas such as governance networks, with meta-governance being seen as a means to enhance the democratic legitimacy of these arenas (Klijn & Skelcher, 2007;Skelcher et al., 2011). The democratic legitimacy of meta-governance itself has, however, received less attention. In this context, a focus on decisions could supplement existing frameworks by drawing attention to the processual norms of democratic decisionmaking, such as transparency, legality, and accountability.
The paper has argued for why the partial organization framework is useful in the analysis of meta-governance. There are, however, limitations to this framework. Of particular importance is that its definition of organization focuses on decisions regarding organizational elements, and how these decisions are justified, and not on informal or discursive organizational practices. The analytical gain is a more distinct conceptual framework, but this does not mean that discourses are unimportant for the analysis of meta-governance. Instead, if we follow the argument of Ahrne and Brunsson (2011), these organizational aspects of meta-governance are left for other theoretical concepts to cover. While we believe that an organizational approach is appropriate for developing a processual understanding of meta-governance, it should be emphasized that other processual theories from organization studies may offer additional insights. The ambition here is to spur continued discussion of the application of organizational theory in governance research. Note 1. The transport planning study was conducted in 2013-2017, while the immigration integration study was initiated in 2018 and is ongoing.

Notes on contributors
Emma Ek € Osterberg is associate professor at the School of Public Administration at the University of Gothenburg. Her research focuses on public sector governance, in particular in market-based arrangements.