From government to governance…to meta-governance: a systematic literature review

ABSTRACT

This study aims to disentangle the concept of meta-governance by systematically reviewing 79 articles. By investigating the who, what, why, and how of meta-governance, this article paints a detailed picture of its empirical manifestation. We conclude that the literature could benefit from better conceptualization and operationalization. From the results, we define 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.’ Meta-governance therefore goes beyond the unproductive dichotomy in the ‘from government to governance’ debate.

KEYWORDS:

• Meta-governance
• systematic literature review
• government
• shadow of hierarchy
• governance

1. Introduction

The concept of meta-governance has attracted quite some attention in the governance literature in recent years. In general, meta-governance is seen as a solution to governance failures (Jessop 2011, 113–116; Lund 2009, 245–246; Torfing and Triantafillou 2011, 2). It is, for example, regarded as an important activity for enhancing accountability and transparency in governance networks, limiting the fragmentation of global sustainability governance, or successful governance of natural resource regions, amongst other things (Glasbergen 2011; Morrison 2016; Sørensen and Torfing 2009).

At the same time, scholars also acknowledge that we know little about what meta-governance entails and how it actually works in practice (Bell and Hindmoor 2009, 69; Torfing and Triantafillou 2011, 2). Meta-governance is often referred to in very general terms such as the governance of governance or the organisation of self-organisation (Jessop 1998; Kooiman and Jentoft 2009; Sørensen and Torfing 2009). Because of this conceptual ambiguity, meta-governance risks becoming a catch-all phrase and a solution for every complex problem. It is therefore necessary to make the highly abstract concept of meta-governance more concrete. We do this by conducting a systematic literature review of the meta-governance literature. We analyse the main characteristics of the empirical cases portrayed in the literature as meta-governance. The findings of such an endeavour can strengthen the conceptualization of meta-governance, which in turn can increase the explanatory value of the concept.

Conceptually, meta-governance appeared for the first time in the public administration and political science literatures in the mid-1990s. Jessop (1997) and Kooiman (1993) were among the first to use the concept. However, their accounts of what constitutes meta-governance differ substantially. Jessop, a prominent scholar in critical state theory, considers meta-governance as the involvement of the state in strategically organizing the context and ground rules for governance. Kooiman, whose work has been influential in the governance literature, describes meta-governance as the process in which the discussion, formulation and application of values, norms and principles for governance takes place (Jessop 1997; Kooiman and Jentoft 2009). Both accounts of meta-governance are, however, very abstract and theoretical. Since these early works, many conceptual and empirical studies have been conducted by a wide variety of scholars from a range of disciplinary backgrounds, focussing on different aspects of meta-governance. However, there is currently no systematic overview of how meta-governance is operationalized and studied empirically. The central question of this research therefore is: how does the empirical literature portray the concept of meta-governance?

Four questions structure our systematic review which allows us to paint a detailed picture of how the current literature portrays meta-governance without imposing our own theory and conceptualizations on the data (McNaughton 2000, 407). The four questions revolve around the who, what, why, and how of meta-governance. The first question, ‘who is portrayed as the meta-governor in the literature?’, provides insights into the characteristics of meta-governors. The second question, ‘what is portrayed as the object of meta-governance?’, aims to show what is being governed by the meta-governor. The ‘why’ of meta-governance is analysed by asking ‘what rationales for meta-governance are found in the literature?’ Finally, the fourth question, ‘through what means is meta-governance enacted?’, provides insight into the ‘how’ of meta-governance. Investigating these questions allows us to synthesise the literature, critically reflect on the contribution of the concept, and provide scope for further research.

The remainder of this article is structured as follows: Section 2 discusses the systematic literature review methodology used and explains how we selected and analysed the literature. Section 3 presents and discusses the results of the review. In Section 4 we relate our results to broader theoretical discussions and formulate ideas for further research. This article ends with main conclusions in Section 5.

2. Methodology

This study applies a systematic literature review method (Gough, Thomas, and Oliver 2012). Systematic reviews differ from more traditional reviews because they use ‘systematic and explicit methods to identify, select, and critically appraise relevant research, and to collect and analyse data from the studies that are included in the review’ (Moher et al. 2009).

The systematic literature review method is best known in the health sciences, where systematic literature reviews usually contain statistical meta-analyses that aggregate available evidence from randomized controlled trials (Delaney et al. 2016, 4; Gough, Thomas, and Oliver 2012, 2). These aggregative studies ‘“add up” findings from multiple, similar studies’ and are used for theory testing (Gough, Oliver, and Thomas 2017, 7). However, next to these aggregative systematic reviews, there are also configurative forms of systematic reviews (Gough 2015). Our study falls into this latter category.

Configurative forms of systematic reviews are adapted to suit more qualitative and heterogeneous research objectives (for a brief overview, see Barnett-Page and Thomas 2009). Configurative systematic reviews ‘aim to configure or “organise” findings of studies’ in order to develop concepts and theories (Gough, Oliver, and Thomas 2017, 3–7).

Compared with an aggregative analysis, a configurative data analysis has a lower level of replicability because of the iterative manner in which the often fuzzy and heterogeneous data from qualitative literature are organized and integrated, and the important role of the researcher therein. This means that there is always a level of interpretation by the researcher. However, the collection and analysis are still systematic because they are executed in a transparent manner, and decisions in the process are specified and reported.

In order to ensure a high standard of reporting quality, our study links to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement, which includes a checklist and a flow diagram (Moher et al. 2009). The PRISMA checklist is included in the Supplementary Material; the flow diagram is depicted in Figure 1, which provides a schematic overview of the review steps. These steps are briefly discussed in the following paragraphs. Additional information can be found in the Protocol (see Supplementary Material).

Figure 1. The systematic review process, based on the PRISMA flow diagram.

2.1. Identification of the literature

To identify literature on meta-governance, we consulted three digital databases: Scopus, Web of Science, and SocINDEX. The first two were selected because of their broad coverage of academic journals; the latter was added because of its specific focus on social science research. We did not limit our search to specific journals or subsets of the literature. We searched for (the different spellings of) meta-governance in the title, abstract, and/or keywords of all articles in the three databases. The search was performed in April 2017. This resulted in the collection of 325 records. All records were transported to the reference management software package EndNote X7. Subsequently, duplicates (n = 119) were removed from the dataset, resulting in a primary database of 206 articles.

2.2. Screening and eligibility of the literature

Inclusion criteria were developed for the eligibility assessment of the documents retrieved. The articles were screened for source (article, peer-reviewed) and language (English). Articles not meeting these criteria were excluded (n = 31). Because we are interested in the empirical characteristics of meta-governance in the literature, we screened the remaining 175 articles to ensure that our study included only empirical cases of meta-governance. To meet this criterion, articles needed to contain an explicit reference to methodology for primary data collection and analysis, contain or refer to a definition of meta-governance, and describe or analyse at least one instance of meta-governance. 79 articles met all inclusion criteria (a list of the included articles can be found in the Supplementary Material).

2.3. Analysis of the articles

The articles were transferred to the computer programme Atlas.ti for configurative analysis. We read all articles completely and coded all text that referred directly to the four guiding questions. This was done because relevant information was often found in sections other than the results section.The method used to analyse the literature is inspired by McNaughton’s (2000) application of Miles and Huberman’s approach to qualitative data analysis (See Dixon-Woods et al. 2005). This method includes a deductive within-case analysis and subsequently a more inductive cross-case comparison. In the first phase, the researcher starts with a list of codes for broad categories. In our case, we used our who-what-why-how-framework. For ‘who’ and ‘what’ we looked at ‘level’ (local to international) and ‘domain’ (public, market, civil society, mixed) because they are general and informative descriptive characteristics. For the ‘why’ of meta-governance, we looked at ‘rationale’ and used open coding to capture this. Similarly, for the ‘how’ we used open coding to capture all relevant activities reported in each article. A substantive and detailed picture was thus developed of all cases of meta-governance in our sample.

The second step was a cross-case analysis to identify similarities and differences across the cases (Dixon-Woods et al. 2005, 49). For this analysis, all data on the coding categories were exported to a data extraction table in Microsoft Office Excel. In this phase, more specific sub-categories were inductively developed. All phrases attached to a respective code were copied to a new Excel sheet and compared and clustered. From this process, new sub-categories emerged. For example, for the ‘why’ of meta-governance, we developed seven sub-categories (e.g. ‘coordination’ and ‘participation/democracy’). The new sub-categories were added to the extraction table. Thereafter, the data on the categories and sub-categories in the extraction table were counted, and we looked for co-occurrence between the different components of the framework. This led to the identification of the main characteristics of meta-governance and the development of four dominant types of meta-governance, which are introduced in the discussion.

3. Results

This section contains our main findings. We start with a short overview of some general characteristics of the meta-governance articles included, such as year of publication and countries covered.

Although Jessop and Kooiman coined the term meta-governance in the mid-1990s, the first empirical study on meta-governance identified in our review was only published in 2001. In this study, Rutherford (2001) analyses the development of local labor market boards in Ontario, Canada as a form of state-led meta-governance. Between 2001 and 2013, the number of publications with empirical studies of meta-governance ranged between zero and seven per year. From 2014 onwards, ten or more publications were identified for each year, with a peak of 14 articles for 2016.¹

Our results show that the debate on meta-governance is scattered over the academic landscape. Our 79 articles were published in 63 peer-reviewed journals with substantially different outlooks and sectoral foci (all the journals are listed in the Supplementary Material). The maximum number of articles published in the same journal is three. This demonstrates that meta-governance as a concept attracts a wide variety of scholars and is picked up in many different academic discussions.

The sectoral focus of the meta-governance articles is also highly fragmented, ranging from studies on meta-governance in the crisis management sector (Stark 2015) to studies on meta-governance in the sports sector (Geeraert 2014). We identified 15 sectoral categories in which meta-governance cases were studied (the sectors are listed in the Supplementary Material). Meta-governance cases in planning and urban development (n = 18) and in the social domain (n = 12), e.g. social housing or healthcare, were most frequently identified.

Contrary to the dispersed sectoral focus, the geographical focus of meta-governance studies is very much concentrated around a small group of countries (see Supplementary Material). As much of the meta-governance literature builds upon earlier academic debates on, amongst other things, network governance and New Public Management, the United Kingdom and the Nordic countries are most frequently discussed in the literature: of the 79 articles reviewed, 24 discuss at least one case of meta-governance in the UK. Our current understanding of meta-governance is thus based almost exclusively on meta-governance in (a specific subset of) Western countries.

In the remainder of this section, the key findings on the who, what, why and how of meta-governance are discussed. Table 1 provides an overview of the descriptive information of our data for these categories.

Table 1. Data on the who, what, why and how of meta-governance.

Level	Number of articles	Type	Number of articles
Who
Local	9	Public	59
Regional	9	Market	1
National	21	Civil Society	1
International	7	Mixed	9
Multilevel	25	Undefined	12
Undefined	12
What
Local	20	Public	7
Regional	15	Market	2
National	8	Civil Society	2
International	5	Mixed	30
Multilevel	5	Undefined	42
Undefined	31
Why		Number of Articles
Coordination		13
Meta-Governors’ aims		11
Participation/Democracy		10
Political administrative changes		6
(Geo)Political conflicts		3
Compliance		6
Not Specified		38
How		Number of Articles
Instrument – Authority		41
Instrument – Informational		35
Instrument – Economic		33
Process design		41
Process management		19
Collibration^a		5

ᵃCollibration is defined as the altering of ‘the weight of individual modes of governance so that the overall ensemble of governance arrangements at a higher or more comprehensive level of social organization is better adapted to coordinate complex social relations’ (Jessop 2016, 80).

3.1. The who of meta-governance

Our analysis of the who, i.e. the meta-governors, leads to four interesting insights. Our first finding is that the who question is not applicable to a small number of articles under review (n = 4/79).² These four articles do not take an agency approach towards meta-governance, but instead conceptualize meta-governance as a structural force that shapes actors’ behaviour. Brooks and Fairfull (2017), for example, analyse how organizational culture influences the form and effectiveness of governance of coastal zone management. These authors define organizational culture as a form of meta-governance. The vast majority of articles (75/79), however, consider meta-governance as an act of agency in which specific actors (i.e. meta-governors) interact to influence the activities of other actors. The remainder of this section focuses on the findings from these articles.

A second key insight is that the literature most often analyses cases in which meta-governance is performed by public actors. Sørensen (2006, 102–103) argues that ‘metagovernance can potentially be exercised by any resourceful actor – public or private. All it takes is resources and a desire to influence activities performed by self-governing actors’. Although this could potentially be the case, our findings show otherwise (see Table 1). The vast majority (n = 59/79) of articles contain public meta-governors. Most of these articles describe public meta-governors that operate at the national level (n = 21). Studies with multiple public meta-governors at different levels were also frequently found (n = 21), while studies of a public meta-governor at the international (n = 3), regional (n = 8) and local (n = 8) level were relatively little found. Besides different levels, the unit of analysis, a ‘public meta-governor’, varies substantially in the literature. Some authors define public actors at an institutional or organizational level, such as the state or a ministry, whereas others define public actors at the individual level. In the latter type of study, the level of analysis is geared towards specific public actors such as civil servants or politicians. Fotel and Hanssen (2009), for example, compare the different meta-governance roles of politicians and administrative representatives and their effect on the democratic legitimacy of regional governance networks in four Nordic countries.

A third key insight is that, in a substantial number of articles (n = 25/79), meta-governance is performed by multiple meta-governors at different levels targeting one network (see Table 1). This multilevel meta-governance exists in different constellations. Rutherford (2001), for example discusses how both the national government and provinces in Canada aim to meta-govern local boards for training and adjustment in Ontario, whereas Whitehead (2007) discusses how local and national actors act as meta-governors of local urban partnerships in the UK. Although there are quite a few cases of multilevel meta-governance in the literature, not much attention is paid to this multilevel character in the conceptualization and discussions of meta-governance. Thuesen (2013) is one of the exceptions; she investigates how local action groups in Denmark perceived the meta-governance activities conducted by the EU, the Ministry of Food, Agriculture, and Fisheries, and several regional and municipal actors. Her analysis investigates the roles (central vs. peripheral meta-governors) and respective activities (hands-off vs. hands-on tools) that differed per meta-governor.

In multilevel meta-governance cases, the roles of the multiple meta-governors can easily become blurred and it is very possible for them not to act in concert or work towards the same goals. They do not automatically ‘ … work in a unified, unidirectional, coherent, or functional way in support of capital or a specific state project. There is an important element of struggle, negotiation, and contingency about how such arrangements operate which must be explored historically-empirically’ (Allmendinger and Haughton 2009, 621; see alsoTemmerman, De Rynck, and Voets 2015; Voets, Verhoest, and Molenveld 2015). This element of interplay between multiple meta-governors is not limited to public meta-governors, as highlighted in Fransen’s (2015) article. He investigates the role of private meta-governors that aim to address coordination problems between multiple voluntary sustainability standards. However, the emergence of multiple and overlapping private meta-governors can be problematic, because they have divergent interests that result in a reproduction of the coordination problems that they intended to solve (Fransen 2015).

Our final finding is that the meta-governor or meta-governors are often not clearly defined in the cases. This is signified by the high numbers (n = 12) of undefined meta-governors in Table 1. Furthermore, direct references to a specific actor as the meta-governor were sparse. We could regularly detect information about the meta-governor(s) only by close reading and interpreting the case descriptions and results sections. This indicates that the concept of meta-governance is often used loosely in the literature and thereby leaves much to the reader’s interpretation.

3.2. The what of meta-governance

But what do these meta-governors actually govern? Our analysis of the what of meta-governance results in five key findings. First of all, in the reviewed literature, meta-governance is often (n = 30/79) directed towards networks consisting of actors from multiple domains, i.e. public, market, and/or civil society (see Table 1). The specific mix of actors varies across the literature, including public–market networks, public–civil society networks, and public–market–civil society networks. Bristow et al., for instance, investigate a variant of the latter in their study on EU Structural Funds Partnerships created by the Welsh Assembly ‘on the basis of strict equality of representation across the public, private and voluntary sectors’ in order to ensure inclusivity of different interests (Bristow et al. 2008, 903).

Second, we observed that the meta-governed networks often operate at a local (n = 20) or regional (n = 15) level (see Table 1). For example, Vabo and Røiseland (2012) investigate the meta-governance of nine Norwegian local urban networks, and Doberstein (2016) analyses how a regional collaborative network on homelessness in the Vancouver region in Canada is managed by bureaucrats. The meta-governance of international or national networks is less prominent in the literature.

The third finding is that the literature differs greatly about what constitutes a network. In some cases, networks are abstract groups of actors without clear linkages, whereas in other cases networks are very well-defined groups of interdependent actors working towards a shared goal, such as public–private partnerships. This difference becomes visible if some examples are compared. Baker and Stoker’s (2015) study, for instance, analyses very loose networks, i.e. the national nuclear power governance networks in the UK, US, France, and Finland. The relationships between the actors in the network and the form of interaction are not clear. In contrast, Bristow et al.’s (2008) and Vabo and Røiseland’s (2012) previously mentioned meta-governance examples investigate clearly defined networks, i.e. EU Structural Fund Partnerships and local urban networks in which the network actors cooperate and develop policies together.

A fourth key finding is that meta-governance is used to describe not only an activity directed at networks, but also an activity of balancing different modes of governance. This refers to the coordination of market, hierarchical, and network governance, rather than the coordination of only networks as discussed above. Few articles have this distinctive understanding of meta-governance in which a meta-governor deliberately switches between governance through networks, through markets, and through hierarchical steering (n = 4/79).

Lastly, echoing our findings regarding the who of meta-governance, it was often hard to identify the what of meta-governance in the literature. This is signified by the high number of articles that fall into the ‘undefined’ category in Table 1; this means that the object of meta-governance could not be clearly established. For the majority of articles (n = 42/79), it was not possible to identify the type of actors in the network being meta-governed, and in 31 articles the level at which the networks operate was not clear, mostly because of a lack of information in the case descriptions.

3.3. The why of meta-governance

As stated in the introduction, meta-governance is often perceived as a solution to governance failures. Failures, however, can mean different things, and hence it is necessary to delve more deeply into the reasons why meta-governance is introduced. In Mayntz’s (1993) words: ‘in the realm of governing one cannot sensibly speak of possible therapies if there is no diagnosis’. So, what was the diagnosis, or rationale, for meta-governance in the cases in the literature studied? We scrutinized the literature for rationales linked to the actual cases of meta-governance under study. As our aim is to further our understanding of meta-governance based on the cases in the literature, we excluded purely theoretical or conceptual references to the rationale for meta-governance.

Our first finding is that the analyses of the empirical meta-governance cases in the literature often do not include information about the rationale for meta-governance. The rationale for meta-governance is neither explicitly mentioned, nor could it be easily inferred from the text, in almost half of the articles (n = 38/79, see Table 1). For the remaining 41 articles in which one or more reasons for meta-governance were identified, these rationales were not the main focus of study, except for Lund (2009). Clearly, little attention is paid in case studies to the why of meta-governance.

The second finding is that, for the articles that do discuss reasons for meta-governance, most (n = 13/79) referred to the need for coordination as the reason, for example to decrease fragmentation, create coherency, aim for procedural harmonization, or reduce overlap. Howard and Bakvis (2016), for example, identified the proliferation of new statistical agencies as the major driving force behind the meta-governance interventions by the Australian Bureau of Statistics, and Whitehead (2003) discussed the meta-governance activity by the UK Government Office as a way to improve coordination and integration of urban policy.

Third, from the cases studied, it becomes clear that the meta-governor is often not a neutral broker in the governing process, but rather an individual or organization that tries to further its own policy goals (n = 11/79). Examples include politicians and administrators that meta-govern a policy network to garner support for the construction of a bridge and a national government using a big-cities policy to achieve its own goal of economically competitive cities (Kokx and van Kempen 2010; Torfing, Sørensen, and Fotel 2009). It is not clear from the literature whether the meta-governor’s aims can be equated with advancing public goals or whether they are more in line with ideas of state capture.

Fourth, meta-governance is used to increase the democratic legitimacy of governance networks (n = 10/79). Holmen (2011), for instance, describes local governance networks in Norway in which elected representatives and administrators played an important role and aimed for ‘a relatively broad participation in order to anchor the strategic plan regionally’. Nevertheless, the objective and the effect of meta-governance do not have to align. Or, in line with Mayntz’s (1993) metaphor, the diagnosis may be right, but the therapy might not cure the disease. This is quite evident when we look at the objective of democratic legitimacy. Screening the outcomes of several meta-governance studies, we found that meta-governance by no means automatically leads to more democratic governance (Bristow et al. 2008; Evans 2007; Qvist 2017; Sørensen 2006). Sørensen (2006), for example, concludes that the way meta-governance is performed by local political leaders in Denmark weakens democracy at the local level, and Qvist (2017) concludes that increasing democratic legitimacy through stakeholder incorporation was not successful because in the end not many different stakeholders were included. Lastly, Bristow et al. (2008, 918) found that deliberate inclusion of specific actors in the partnerships under investigation did not lead to ‘meaningful equality of influence’ but instead led to negotiations between elite groups.

Our final finding is that, besides intentional reasons such as the goal of more coordination or better democratic practices, some studies identified contextual reasons for meta-governance. Examples include changes in political-administrative systems, (geo)political conflicts, and external requirements by superior institutions or the law (see Table 1). Walls et al. (2005), for example, discuss how highly contentious debates about the regulation of genetically modified crops and mobile telecommunications technology impelled the UK government to meta-govern, and Thuesen (2013) describes how a ministry as a superior institution required a municipality to fulfil a meta-governing role. Such an obligation to comply can also have the form of a legal obligation (Fransen 2015; Haughton, Counsell, and Vigar 2008).

3.4. The how of meta-governance

Our review so far has shown who are portrayed as meta-governors, what they are targeting, and the reasons why meta-governance is performed in the cases in the literature. The fourth, and final, question looks at how meta-governance is performed, in other words we want to unravel the means that meta-governors use. We coded all the means used by the meta-governors (the who) in relation to the meta-governed (the what) in an open fashion. This resulted in a long and diverse list that can be classified in several ways. The categorization in Table 1 builds upon different classifications – i.e. Hoods and Margetts’s (2007) NATO model, the network management methods of Klijn and Edelenbos (2007) and the hands-off and hands-on tools of Eva Sørensen and Jacob Torfing (2009) – to demonstrate the variety of means used. Definitions of the categories can be found in the Supplementary Material. Below, we discuss four key insights that we derived from our findings.

The first insight is that many of the means of meta-governance do not differ much from the means of traditional governmental interventions. A vast majority of the meta-governance articles (n = 57/79) include meta-governance activities that can be characterized as authority instruments, economic instruments, and/or informational instruments. These policy instrument categories are well-known in research on the means of governmental steering (Hood and Margetts 2007). At first sight, it might be surprising that the same instruments are also used in meta-governance, because meta-governance is often characterized as remarkably different from hierarchical steering (Sorensen and Torfing 2009). What is remarkably different from hierarchical steering, however, is that the instruments are often applied in an indirect manner: meta-governors use policy instruments to govern networks, which in turn govern society.

Table 1 shows the frequencies of the three policy instruments. Authority instruments are used most frequently (n = 41/79), followed by informational instruments (n = 35/79) and economic instruments (n = 33/79). Authority instruments covered in the literature range from strict mandates and task descriptions for governance networks (See for example Damgaard and Torfing 2010; Haveri et al. 2009; Lund 2009), to broader overall visions including political priorities and guiding core principles (e.g. Hansson 2013; Sehested 2009; Voets, Verhoest, and Molenveld 2015). Informational instruments covered in the literature range from the dissemination of knowledge, either in a one-way fashion (for example by issuing publications) or as a two-way street (in the form of interactive information exchange focused on learning and dialogue), to developing narratives and frames. Examples of economic instruments found range from positive financial stimuli such as the provisioning of subsidies and co-financing projects (Frantzeskaki, Wittmayer, and Loorbach 2014; Harada and Jorgensen 2016), to negative financial stimuli, such as financial clawback in the case of non-compliance (Whitehead 2003). Other studies did not look so much at the actual financial contributions of the meta-governor to the meta-governed but mentioned the development of financial frameworks as a steering instrument (e.g. Haveri et al. 2009).

A second key finding regarding the how of meta-governance is that meta-governance often comprises a form of process design in which networks are institutionalized. Hence, meta-governance is not only a response to the networked reality, but also instigates this reality. Process design refers here to activities that bring together actors and institutionalize cooperation between them. This institutionalization can range from providing a channel for interaction, to the development of a fully-fledged organization. More than 50 per cent of the articles discuss this form of meta-governance (n = 41/79). Chesire, Everingham, and Lawrence (2014), for instance, describe the creation of coordinating groups as a meta-governance strategy of local authorities governing the mining sector in Australia. A second form of process design centres around the selection of members that can participate in the governance network. Meta-governors can choose different participation strategies when developing networks (Palm and Thoresson 2014) and therefore include or exclude certain stakeholders.

Our third finding is that meta-governance is often discussed in the literature as a form of direct governance, or hands-on, meta-governance. Sørenson and Torfing (2009, 47) were the first to discuss this form of meta-governance in which the meta-governors closely interact with the governance network at hand. They distinguish between hands-on network participation and hands-on network management. Their account of these two forms of meta-governance is clear about the goals but not about the means of hands-on meta-governance. It is, for example, unclear what participation itself entails, more than the simple presence of the meta-governor in the network. Therefore, we did not include hands-on participation in Table 1 as a specific means of meta-governance.

Many hands-on activities identified in the case descriptions can be characterized as process management. This includes activities of designing, measuring, monitoring, and improving the network’s processes, through activities such as time management, trust building, and organizing a process of continuous rethinking and learning: 19 of the 79 articles discuss activities that fall into this category of meta-governance. Doberstein’s (2016) research on homelessness policy in Vancouver, Canada, for example shows how the meta-governor steered the network’s deliberations towards less contentious issues first, in order to build trust and a sphere of cooperation.

Our fourth finding is that, when meta-governance refers to the coordination of network, market, and hierarchical governance, this implies specific means of meta-governance. Jessop (2016, 80) refers to this as collibration, which is the altering of ‘… the weight of individual modes of governance so that the overall ensemble of governance arrangements at a higher or more comprehensive level of social organization is better adapted to coordinate complex social relations’. It is a reflexive activity of calibrating optimal mixes of these modes of governance. A few studies (n = 5/79) included collibration strategies. Meuleman (2010), for example, discusses how the meta-governor chooses specific governance modes and accompanying instruments and switches and adapts the mixture of these modes in response to a changing context.

4. Discussion

This research aimed to uncover how meta-governance is manifested in the academic literature. To this end, we have analysed who meta-governs what, in what manner this is done, and for what reasons. These questions are indispensable for a thorough understanding of what meta-governance means both conceptually and empirically.

As discussed in the results section, the majority of cases described in the literature analyse meta-governance activities performed by a public actor, such as a ministry or local government. Furthermore, our review revealed that a wide variety of means can be found in the meta-governance literature. We encountered meta-governance through the use of instruments (hierarchical, economic, informational), through network management methods (process management or process design), and through collibration strategies (switching modes). In most cases, these means were used to influence network governance. Additionally, a few cases described how meta-governance was directed not only at network governance, but also at market and hierarchical governance. Finally, we found that meta-governors aim to overcome a wide variety of governance failures, such as coordination problems, lack of democratic legitimacy, and failure to reach certain substantive outcomes.

From these findings, we define 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’. Although more specific than ‘the governance of governance’, our definition of meta-governance remains necessarily broad to capture the richness of the literature. However, within these boundaries, we can make a distinction between different types of meta-governance. To identify these ideal types from the literature, we looked at co-occurrence between the who, what, why, and how of meta-governance across the articles. It should be noted that it was only possible to establish co-occurrence between the who and the what, but not in relation to why and how meta-governance was performed. The latter was impossible because many articles include multiple cases and often it was not possible to relate a specific means of meta-governance (how) or a reason (why) to a specific case in the article. Additionally, the sheer number of different possible combinations and variations in the variables made it impossible to establish meaningful inferences about the relationships (see also Faling et al. 2019). Nevertheless, by combining the data on the who and what of meta-governance, it is possible to identify four ideal types of meta-governance, namely: network meta-governance; multilevel meta-governance; meta-governance of multiplicity; and meta-governance of modes (see Figure 2)

Figure 2. Four ideal types of meta-governance.

The most frequently identified ideal type is network meta-governance. In this variant, a public meta-governor steers a governance network that comprises actors from multiple domains. Nyseth (2008) provides a quintessential example of this form of meta-governance. The study investigates how the local authority in Tromsø, Norway, steered a governance network of private and civil society actors in the realm of urban planning.

A second meta-governance ideal type that we found in the literature is meta-governance in which multiple meta-governors from different levels are active, i.e. multilevel meta-governance. This encompasses the activities of multiple public³ meta-governors towards the same public–private⁴ network. These meta-governors are located at multiple levels, ranging from the international to the local level. We understand multilevel meta-governance as an arrangement in which multiple meta-governors at different levels steer a specific network (see Figure 2). Fotel and Hanssen’s (2009) article is a nice illustration of this type of meta-governance, as they analyse how public national, regional, and local authorities all meta-govern a regional network.

A third type of meta-governance is the meta-governance of multiplicity. In this variant meta-governance is directed towards a network of networks. The public meta-governor aims to steer multiple governance networks in concert (see Figure 2). la Cour and Andersen (2016) provide a typical case of this type in their study of the governance of the Danish housing sector. The national government acted as meta-governor through a steering reform that encouraged cooperation between municipalities and local housing associations. In doing so, the Danish state stimulated collaborative governance at the local level in a multitude of Danish municipalities.

The first three ideal types of meta-governance we identified from the reviewed literature are all examples of what Jessop (2011, 115) calls ‘first-order’ meta-governance: the governance of one specific governance mode, namely, network governance. Our fourth and final ideal type of meta governance emerging from the literature is where the public meta-governor seeks to balance hierarchical, market, and network governance (see Figure 2). Jessop refers to these as is second-order meta-governance with multiple modes and interactions (Jessop 2011; Meuleman 2019). We found a few articles that fall within this category. Meuleman’s study is a nice example where he shows how public managers in the UK, Germany, the Netherlands, and the European Commission try to create ‘situationally optimal mixtures’ of hierarchical, network, and market governance in environmental policy (2010, 67). Meuleman concludes that governance approaches are context specific and therefore there is not one best practice, but different situationally best approaches in governance modes and mixes.

In the remainder of this section, we reflect on our findings in relation to the broader governance literature and develop three arguments. As a first argument, we contend that meta-governance can be seen as the synthesis of the ‘from government to governance’ debate. In this debate which has been ongoing for decades, it is argued that there is a shift from government to governance, where hierarchical steering by government is increasingly challenged by governance through self-organizing networks of actors from civil society and the market. In this debate about ‘governing without government’, some argue that the state is ‘hollowed out’ (Rhodes 1996; Milward and Provan 2000), meaningthat the steering capacity of the state has been reduced and that government is seen as ‘weakened and as incapable of “steering” as it had in the past’ (Peters and Pierre 1998, 224; Rhodes 2007).

The concept of meta-governance moves beyond the ‘from government to governance’ thesis and brings government back in at a more central stage, thereby questioning the nature of governance as pure self-organizing networks in which governments fulfill a limited steering role. Instead, meta governance literature argues that many such network and collaborative governance arrangements operate rather ‘in the shadow of hierarchy’ (Bell and Park 2006; Fotel and Hanssen 2009; Haveri et al. 2009; Rhodes 1996; Scharpf 1994; Whitehead 2003). Although Scharpf’s work on the shadow of hierarchy predates discussions about meta-governance, his idea of the shadow of hierarchy resonates clearly in the meta-governance literature; Scharpf concludes that networks are embedded in a hierarchical structure and government has the power to approve and disapprove negotiated outcomes. It is often the government that created the networks or the procedures on which these networks operate (Scharpf 1994, 41).

Thus, the meta-governance literature demonstrates a dialectical development in the debates on the shift from government to governance. The thesis of hierarchical steering by the sovereign state was followed by the anti-thesis of the hollow state in the new networked reality. Meta-governance can be perceived as its synthesis: it often implicitly tries to reconcile the government versus governance debate by looking into the new role of the state in the networked society. Meta-governance is ‘government plus governance’ and can be constraining as well as enabling, as it ‘combines control and facilitation’ (Damgaard and Torfing 2010, 260).

The pitfall in the current literature on meta-governance is the constant danger of falling back in the dichotomy between hierarchy and self-organization and drawing the final conclusion about the power of the state vis- à -vis self-organizing networks. Instead, it is more worthwhile to acknowledge the context specificity of cases and investigate under what conditions and in which ways public authorities have the power to meta-govern. This requires more insights into the causal mechanisms behind meta-governance and into the necessary and sufficient conditions under which meta-governance works. Our review has given some initial hints about circumstances that can lead to meta-governance (e.g. political-administrative changes or the need for compliance with rules) and important rationales for meta-governors (e.g. the need for coordination or for democratic legitimacy), but further research on the cause and effect chains in meta-governance is necessary.

Our second argument is that meta-governance should be perceived as an umbrella concept in the governance literature that ties together different branches of literature. Our review shows that it covers parts of the literatures on network management, orchestration, and collibration. Articles on meta-governance are often not explicit about how they relate to these strands, especially to network management and orchestration, although they have much in common. Therefore, we argue that the meta-governance literature could benefit from connecting more closely to literature on network management and orchestration. For example, the network management literature pays considerable attention to process management as well as process design – the two methods that we also found repeatedly in the literature reviewed. Future meta-governance research could, for example, learn from network management studies about different management strategies and the effects of these different strategies on the outcomes of network governance (Meier and O’Toole 2001; Klijn, Steijn, and Edelenbos 2010).

A similar case can be made for connecting to literature on orchestration. In contrast to hands-on network management, orchestration has a more hands-off character and international focus. Orchestration is defined as a specific governance mode in which an orchestrator ‘enlists and supports intermediary actors to address target actors’ in pursuit of the orchestrator’s goals (Abbott et al. 2015, 4). Orchestrators provide an intermediary organization, e.g. a governance network, with ideational and material support. It is a soft form of steering because it is not based on hard control through legal means. Future research on meta-governance could, for example, learn from studies on orchestration how indirect governance works at the international level and the instruments that are available at that level.

Our third and last argument is that, in order to increase the explanatory relevance of meta-governance, future research should pay more attention to the relationship between meta-governance and democratic legitimacy and governance failure. Our findings indicate that democratic legitimacy is an important rationale for meta-governance but that meta-governance does not automatically lead to a higher level of democratic legitimacy of governance. It is therefore worthwhile to further investigate the relationship between meta-governance and the democratic legitimacy of governance networks. A complicating factor is that different democratic theory traditions are used in the analysis of meta-governance. From a liberal democratic perspective, it is argued that the oversight of governance networks by elected bodies or politicians (i.e. meta-governors) will increase the democratic legitimacy of the respective network because networks often selectively favour the inclusion of private, self-interested stakeholders, and these networks lack electoral accountability (Bristow et al. 2008; Fotel and Hanssen 2009; Plüss 2014). By linking representative democracy to these networks, meta-governance can protect the public interest and safeguard accountable, transparent, and representative governance. On the other hand, some authors argue that governance networks are mechanisms based on ‘trust, cooperation, and informal interaction’ that stimulate deliberative processes and better stakeholder representation (Plüss 2014, 492; Sørensen and Torfing 2009). From this perspective, oversight and interference by the state or other public actors can endanger this deliberative process and diverse representation. Sørensen and Torfing (2005, 2009) concept of ‘democratic anchorage’ could be used to examine these different understandings of the relationship between meta-governance and democratic legitimacy.

5. Conclusion

The central question guiding this systematic literature is: how does the empirical literature portray the concept of meta-governance? For this, we looked at four dimensions of meta-governance: the who, the what, the why and the how. Investigating these dimensions allowed us to synthesise the literature, reflect on the contribution of the concept, and provide points for future research. From our findings, we conclude that meta-governance is described in the literature 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.

In the discussion section, we reflected on how meta-governance relates to the wider governance literature. We contend that meta-governance should be perceived as an umbrella concept in the governance literature. It ties together different branches of literature, such as network governance and orchestration literature. We also reflected on meta-governance as the synthesis of the classical ‘from government to governance’ debate. It does not focus only on government steering society, neither does it envision society only as a sphere of self-governing networks. Instead, meta-governance literature acknowledges the importance of both government and governance. It is a useful concept because it enables us to analyse the relationship between public authorities and governance networks. That said, this systematic literature review also shows that the meta-governance literature is still in its infancy. The concept of meta-governance is used loosely and without much operationalization in case studies in the literature. This is problematic, as it does not allow us to reflect critically on what meta-governance entails in practice, and it prevents us from advancing it theoretically. For example, in order to be able to say something about the effects of meta-governance, a thorough conceptualization and operationalization of meta-governance is indispensable, something which is currently lacking in the meta-governance literature. We hope that our study makes a modest contribution to strengthening the conceptual and empirical research on meta-governance.

Acknowledgments

A previous version of this paper was presented at the Annual Work Conference 2017 of the Netherlands Institute of Government where we received useful comments. The authors also wish to thank Peter Tamás for his methodological advice as well as the three anonymous reviewers for their valuable comments.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here

Additional information

Funding

Contributions by JG and KT were funded by the Netherlands Organisation for Scientific Research [NWO grant no. 409-14-017] Contributions by RB were funded by the Netherlands Organisation for Scientific Research [NWO grant no. 451-17-0064140].

Notes on contributors

Jonna Gjaltema

Jonna Gjaltema, MSc. is a PhD candidate at the Public Administration and Policy group at Wageningen University & Research, the Netherlands. Her research addresses the relationship between public and private governance of sustainable global value chains. After finishing her Master’s in International Relations and Diplomacy, she worked as a policy advisor to a Dutch Member of the European Parliament and as a policy advisor at the Dutch Ministry of Foreign Affairs.

Robbert Biesbroek

Robbert Biesbroek is Associate Professor at the Public Administration and Policy group of Wageningen University & Research, the Netherlands. His research interests include comparative public policy, mechanisms of complex decision making, dynamics of policy (dis)integration, and the political and bureaucratic responses to climate change. He is recipient of a Dutch NWO-VENI research grant (2018–2021) and (co)authored over 50 scientific articles in the fields of public policy and climate change.

Katrien Termeer

Katrien Termeer is Professor and Chair of the Public Administration and Policy group at Wageningen University & Research, the Netherlands. Her research addresses the governance of wicked problems in the interrelated fields of food, agriculture, climate and energy. Before, she worked at the Erasmus University of Rotterdam; Technical University of Delft; the Ministry of Agriculture, Nature, and Food; and Sioo, a Centre for Organizational Change and Learning. She is also a Crown-appointed member of the Social and Economic Council of the Netherlands.

Notes

1. NB. Only the first four months of 2017 are included.

2. Articles with a structural focus are included in the category ‘undefined’ in Table 1.

3. in our data, 86 per cent of multilevel meta-governance relates to public meta-governors.

4. Of the multiple public meta-governors cases, 71 per cent correspond with public-private networks, undefined cases excluded.