2.1 Introduction

A key starting point for polycentric governance scholars is the idea of heterogeneity in governance, which the Oxford English Dictionary loosely defines as “composed of diverse elements or constituents.” Accordingly, “the Ostroms pointed toward heterogeneity, diversity, context, and situational logic as critical elements in the analysis of institutions, governance, and collective action” (Reference AligicaAligica, 2014, p. 5). While heterogeneity can take many forms, referring to diversity in capabilities, preferences, beliefs, information, but also social, cultural, or linguistic aspects (Reference AligicaAligica, 2014, pp. 4–5), this chapter focuses on the conceptual consequences of such heterogeneity for theorizing the role of policy evaluation in the shift from polycentricity to polycentrism (see Chapter 1).

To do so, this chapter provides an overview of polycentric governance theory in terms of positivism, normative elements, and key variables. It then disentangles three foundational ideas or assumptions on which polycentrism builds, namely that actors can and do self-organize in order to address pressing governance challenges, that context matters in governance, and that governance centers, while independent, interact to fully generate the hypothesized benefits of polycentric governance. The chapter explains the origins of these ideas, specifies their theoretical implications for polycentric governance, and draws together key existing empirical research. This is done in order to assess how the idea of monitoring currently features in work on polycentrism, and how related key insights may be developed in order to analyze what role evaluation may play to contribute to the shift from polycentricity to polycentrism with a view to the foundational ideas. The chapter ultimately combines polycentric governance and policy evaluation literatures in novel ways to advance polycentric governance and policy evaluation theory. It concludes by further elaborating and specifying the key empirical research gaps, which Chapter 1 already flagged, and which will be addressed later.

2.2 Positivism, Key Variables, and Normative Theory

From the outset, and as Chapter 1 recognized, it is critical to appreciate that theory on polycentrism contains a subtle blend of both normative and positive elements (Reference AligicaAligica, 2014; Reference McGinnisMcGinnis, 2016). In his early and later republished work, Vincent Ostrom explains that polycentrism does not simply provide an explanation of the status quo, but is rather a theory which is capable of making normative prescriptions, such as identifying necessary conditions for polycentrism to work (Reference Ostrom and McGinnisOstrom, 1999a). In more recent contributions, the normative element has become even more pronounced, as for example Michael D. Reference McGinnisMcGinnis (2016) specify what “polycentric governance requires” (p. 15; emphasis added). In sum, the polycentric approach must be understood as both a positive and a normative project.

Such normative considerations should, however, not obscure the extraordinary amount of empirical work that scholars in and around the Ostrom Workshop in Political Theory and Policy AnalysisFootnote ¹ have been undertaking for at least five decades in order to gauge the (normative) polycentric approach against empirical realities. Elinor Ostrom’s book on Governing the Commons (Reference Ostrom1990) draws on vast empirical evidence to validate and further develop key polycentric ideas. Her “design principles”Footnote ² for common pool resource governance systems have since found further empirical support around the world (Reference OstromOstrom, 2005). Given decades of empirical work and particularly Elinor Ostrom’s affinity for “grounded research” and interdisciplinarity (Reference OstromOstrom, 2005), it would be inappropriate to relegate polycentrism to the realm of purely normative governance theories. By the same token, polycentrism clearly contains normative elements, which will likely become stronger as the approach gains traction and application on numerous issues, including climate governance. In fact, Reference AligicaAligica (2014) argues that Elinor and Vincent Ostrom have moved from empirical explorations towards more normative elements over time. Therefore, “Certain normative assumptions and preferences are undoubtedly and inescapably embedded at a very basic and intuitive level in the perspectives advanced by scholars, like the Ostroms, who explore collective action and institutional arrangements” (Reference AligicaAligica, 2014, p. 17). This book thus endeavors to make these normative elements explicit and engage with them in the context of studying policy evaluation.

The presence of normative aspects in polycentrism derives from Elinor and Vincent Ostrom’s general scholarly approach, which seeks to elevate theory over methods (Reference Aligica, Sabetti, Sabetti and AligicaAligica & Sabetti, 2014b, p. 2). This approach reacts to the positivist doctrine starting in the 1960s, where scholars endeavored to build theory starting from empirical insights (for a treatment of this intellectual history, see Reference Ostrom, Sabetti and Dragos AligiaOstrom, 2014a). Therefore, Elinor Reference Ostrom, Sabetti and Dragos AligiaOstrom (2014a) advocates that “the development of theory precedes the choice of appropriate methods to test a theory” (p. 218). She furthermore elaborates that “theory has also come to mean for many political scientists a set of logically connected statements without the requirement that assumptions used in a theory have themselves already been established as empirical laws” (p. 218; emphasis in original). However, reverting back to an earlier point, Elinor Ostrom also had a strong affinity for empirical work. She argues that while “theory precedes empirical work …, empirical studies help to refine our theoretical understanding of the world” (Reference Ostrom, Sabetti and Dragos AligiaOstrom, 2014a, p. 222). In sum, Elinor Ostrom thus advocates a dialectic relationship between theory and data, but an approach that allows normative elements because theory comes before empirics. This general stance may in part explain the presence of normative elements in scholarship on polycentrism.

There has, of course, been a strong movement in political science and related fields to develop context-independent and generalizable theory. As Reference Benjamin and OstromBenjamin (1982, p. 69) argues, “During periods of relative social-economic and political stability, social scientists are lured into a false sense of security regarding the ahistorical validity of empirical generalizations.” Thus, if social conditions are ever changing and unstable, Reference Benjamin and OstromBenjamin (1982, p. 93) holds that

According to Reference Austen-Smith and BanksAusten-Smith and Banks (1998, p. 259), a “[p]ositive political theory is concerned with understanding political phenomena through the use of analytical models which, it is hoped, lend insight into why outcomes look the way they do and not some other way.” These models typically include assumptions such as rational individuals or the way individuals interact in game-theoretic situations (Reference Austen-Smith and BanksAusten-Smith & Banks, 1998). While polycentrism provides a normative panoramic vision of the governance landscape, many of the inner workings – both in normative and empirical terms – have yet to be fully explored.

2.3 Polycentrism: Three “Foundational” Ideas

Building on the underlying ideas of heterogeneity in governance (see above Section 2.1), the polycentric governance approach flows from and finds support in three foundational ideas. The first foundational idea of polycentrism is that polycentric governance emerges precisely because actors at various levels have the capacity and, given adequate circumstances, the willingness to self-organize. In earlier writings, Vincent Ostrom has pointed to the “self-organizing tendencies” of such actors in polycentric systems (Reference Ostrom and McGinnisOstrom, 1999a, p. 59). In order to self-organize, (new) actors need governance systems that are sufficiently open and flexible, a sense that they have some capacity to affect and change the rules to which they are subjected, and a feeling of motivation to actively participate in enforcement (Reference Ostrom and McGinnisOstrom, 1999a). In this process, self-organizing actors may thus benefit from sufficient, place-sensitive information on previous climate policies that is readily available and accessible. If these conditions are met and actors self-organize, outcomes may be normatively “better” than top-down solutions.

Polycentric governance theory holds that this is because those who have knowledge of the particular “local” governance context tend to be better placed and willing to make rules and regulate their own behavior – the key idea of subsidiarity. In recent decades, empirical evidence from common pool natural resource management literatures has built up to emphasize this point. Crucially, the assumption that actors will always deplete common pool resources in the absence of coercion from a higher authority or property rights (Reference HardinHardin, 1968) does not withstand empirical scrutiny across all cases (Reference OstromOstrom, 1990), although Elinor Ostrom recognizes potential drawbacks of polycentric governance arrangements, such as the possibility of free riding and potential under-provision of public goods (Reference OstromOstrom, 2010c). In fact, across multiple natural resource types including fisheries, and water or timber production, local actors managed to build enduring institutional systems to self-govern their local resource use (Reference OstromOstrom, 2010c). Thus, in some cases, actors appear to exhibit the capacity to self-organize and outperform top-down solutions. This book assesses to what extent this proposition materializes in the case of climate policy evaluation.

The second foundational idea is that context matters and that no rule or policy will produce effects irrespective of their wider context (Reference AligicaAligica, 2014). Elinor Ostrom and others conceptualize the influence of “context” through the Institutional Analysis and Development (IAD) Framework. According to Michael D. Reference McGinnisMcGinnis (2011, p. 51), “The IAD framework contextualizes situations of strategic interaction by locating games within social, physical, and institutional constraints and by recognizing that boundedly rational individuals may also be influenced by normative considerations.” This line of reasoning underpins one of Elinor Ostrom’s key messages, namely that there are no policy “panaceas” that will hold in all situations irrespective of the context (Reference Ostrom, Janssen and AnderiesOstrom et al., 2007). Different contexts require different approaches as there is no one-size-fits-all approach.

This insight has long been acknowledged in international climate governance. In 1992, the UNFCCC stated that in order to address climate change, “policies and measures should take into account different socio-economic contexts” (Article 3[3]). In consequence, it is only by paying close attention to the context that analysts can understand how actors and rules generate particular effects (Reference AligicaAligica, 2014) – and by extension policy evaluation should, therefore, also be place and time specific. Furthermore, because context and “local” conditions matter, multiple solutions at various governance scales including many actors may thus generate “better” outcomes than a single, hierarchical approach. This is one of the most central ideas of polycentrism. However, “[n]o a priori judgment can be made about the adequacy of a polycentric system of government as against the single jurisdiction” (Reference Ostrom, Tiebout and WarrenOstrom et al., 1961, p. 838). The effectiveness of polycentric governance depends, at least in part, on its fit with the wider context into which it is placed. Building an understanding of the successes and failures of polycentric governance systems thus requires close attention to context – including in (public) policy evaluation. Therefore, learning across contexts requires intimate knowledge of contextual variables – including historical, geographical, cultural, or ideational aspects to name but a few (Reference Aligica and SabettiAligica & Sabetti, 2014a).

The third foundational idea holds that if polycentrism is to emerge, governance centers need to interact, but without generating strong interdependencies. But what is a “governance center”? Scholars in the polycentric tradition differ in their understanding. For example, Elinor Reference OstromOstrom (2012, p. 355) writes that “[a] polycentric system exists when multiple public and private organizations at multiple scales jointly affect collective benefits and costs,” thus taking an “organization” as the core unit of analysis. In a slightly different way, Vincent Ostrom et al. (1961, p. 831) write about “centers of decision-making” as the core unit, with less emphasis on “organizations.” In a different vein, Elinor Reference OstromOstrom (2005, p. 257) stresses that “complex, polycentric systems of governance that are created by individuals,” thus focusing on people. In other places in the same book (Reference OstromOstrom, 2005, p. 269), Elinor Ostrom writes about “the presence of governance activities organized in multiple layers of nested enterprises” (emphasis added). These differing definitions show that what constitutes a “governance center” remains unclear, as it may range from individuals to all types of organizations or enterprises all the way to more fuzzily described “centers of decision-making.” To complicate things more, a recurring theme in Elinor Ostrom’s scholarship is that governance centers are “nested” (see quote above), which creates the challenge to not only tell governance centers apart in a horizontal, but also in a vertical, way and to understand their potential linkages. Looking across the relevant literatures, the ideas of “decision-making” and “independence” run deeply and are probably theoretically more relevant than the exact nature of the organization (or the number of people involved) that make up a governance center. This book defines governance center in a broad sense, that is, as any organization or organizational unit that has authority to make some decisions and is reasonably independent in doing so (see Reference Ostrom, Tiebout and WarrenOstrom et al., 1961). This definition therefore encompasses the level of the nation state and supranational organizations like the EU.

Linked to the idea of “nesting,” what drives interactions between centers of governance in polycentric systems? There are numerous potential mechanisms. Vincent Ostrom believed that governance centers will interact more or less automatically if they have sufficient incentives to do so (Reference Ostrom, Tiebout and WarrenOstrom et al., 1961). Overlapping jurisdictions may be one reason why centers interact. For example, writing on the IAD, Michael D. Reference McGinnisMcGinnis (2011, p. 52) proposes that interaction may take place through a “network of adjacent action situations” (NAAS) where individuals or organizations simultaneously participate in multiple rulemaking venues in a polycentric system. These individuals or organizations become bridges between different governance centers to foster interaction. In other cases, interaction may emerge because of market-like competition – for example, when different governance centers offer the same service. If two municipal governance entities provide the same service, people are likely to choose the one that they see as most favorable, depending on the dimension that matters most to them (e.g., cost; quality of the service). However, scholars from other fields have proposed a range of additional mechanisms. For example, policy diffusion and transfer scholars distinguish between learning, competition, coercion, and mimicry as forms of interaction (Reference Marsh and SharmanMarsh & Sharman, 2009). While multiple disciplines have identified these kinds of mechanisms, scholars differ significantly on which mechanisms matter more and, importantly, how much external stimulus may be required to kindle interaction. By definition, the polycentric approach excludes ideas around top-down coercion, as governance centers are per definition thought to be independent.

In climate change governance, the threat of “carbon leakage” provides one potential (external and market-driven) incentive for governance centers to experiment with reducing their carbon dioxide emissions efficiently and potentially cooperatively. Carbon leakage generally refers to the idea that actors may shift activities that cause carbon pollution from jurisdictions with more regulation to those with less rules in a classic “race to the bottom” (Reference OstromOstrom, 2014b). Thus, if public policymakers perceive carbon leakage as a threat – such as heavy industry moving to other countries, with corresponding job losses – they may have significant incentives to identify the least intrusive ways to reduce carbon emissions and ensure that other governance centers take equivalent action.

An additional reason to look beyond one’s own governance center is to learn from the successes and failures of others, especially because policymakers tend to be risk averse (Reference HowlettHowlett, 2014). While the concept of policy learning has been subject to much scholarly debate, multiple authors point to learning as some change in behavior or beliefs, following the impact of experience, new information, or changing circumstances (Reference Bennett and HowlettBennett & Howlett, 1992). Of particular interest to this book is “lesson-drawing,” which is one form of learning that focuses on using the “lessons” or experiences from one governance context in another (Reference RoseRose, 1991, Reference Rose1993). Thus, Reference RoseRose (1991) explains that “a lesson is here defined as an action-oriented conclusion about a programme or programmes in operation elsewhere; the setting can be another city, another state, another nation or an organization’s own past” (p. 7). Crucially, rather than being compelled by some top-down authority, “lesson-drawing tends to be voluntaristic” (Reference RoseRose, 1991, p. 9) and thus fits well with ideas on polycentric governance. Climate policies may, for example, generate socially desirable side effects, such as improvements in human health or reducing congestion (Reference Thompson, Rausch, Saari and SelinThompson et al., 2014). Learning about experiences with such (beneficial) side effects and their political consequences may thus be another incentive to seek information about experiences in other governance centers. Lesson-drawing is not the “normal” state of affairs, but rather emerges from an underlying level of “dissatisfaction” with the status quo that prompts a search for lessons from elsewhere (Reference RoseRose, 1991). The aforementioned risk aversion among policymakers may be one such source of “dissatisfaction.” In the area of climate change, where there are currently no examples of far-reaching policy success in addressing this global issue, governance centers may be especially interested in the experiments of others as a key source of lessons (Reference AligicaAligica, 2014, p. 66; Reference GoodinGoodin, 1996; Reference Hildén, Jordan and HuitemaHildén et al., 2017).

An issue, of course, emerges with regard to the previous points about context. If context matters in policymaking, how can one learn from others? Following Reference McConnellMcConnell (2010), there are those who argue that policy is so contextual that nothing can be learnt across governance centers. By contrast, others contend that policies work irrespective of the context through set mechanisms (e.g., the power of the market to efficiently allocate resources). Between these arguably extreme positions is what Reference McConnellMcConnell (2010, p. 200) terms the “familial way” of contexts. While contexts may differ on a range of conditions, some settings are more similar than others. For example, if a country has a democratic parliamentary political system, all else being equal, a successful policy tends to be more likely to succeed in another country with a similar political system rather than a very different one (e.g., an authoritarian state). Thus, it may be possible to determine to what degree contexts are reasonably similar. This is, of course, no guarantee of success (Reference McConnellMcConnell, 2010). However, if a governance center wishes to learn from the experiences of another, it may be helpful to decipher which contextual conditions were critical for the success of a particular intervention, and if those conditions are present elsewhere (see Reference Benson and JordanBenson & Jordan, 2011).

This view of automatic interactions driven by a range of incentives contrasts with insights from other governance literatures that point to the need to stimulate interaction in some circumstances (e.g., Reference Jordan and SchoutJordan & Schout, 2006). There are reasons to believe that self-organization and consequently “taking into account” may not be automatic, something which has stimulated voluminous debates on “meta-governance.” In the absence of strong market signals or other powerful incentives – which is often the case in the public sector where duplication of services may be seen as a waste of resources – other mechanisms may be necessary in order to generate enough pressure to compel governance centers to pay attention to one another. In other words, it may be necessary to externally induce some of the dissatisfaction that Reference RoseRose (1991) considers essential for lesson-drawing to happen. This is also because while it may be perfectly rational from a collective standpoint to learn from others and continually improve governance practices, numerous factors such as vested interests, path-dependent behavior, preexisting institutions, and general political inertia bolstered by overburdened policymakers, may prevent such learning in practice, thus necessitating other forms of coordination. Hierarchies are one way to achieve this (see Reference PetersPeters, 1998), but hierarchy does not sit well with the Ostroms’ ontology of self-organization and may in some cases not even be possible (notably in the international climate regime at the time of writing).

In increasingly networked arrangements, where neither markets nor hierarchies force coordination, mutual taking-into-account, or what others have termed “policy coordination,” may thus be subject to substantial collective action dilemmas (Reference Jordan and SchoutJordan & Schout, 2006), the key issue that polycentric governance seeks to address (Reference OstromOstrom, 1990). Even though the system as a whole could benefit from learning, individual governance centers may not be able or willing to draw lessons from others or to provide their own lessons. For these reasons, some higher-level incentives, if only through coordination, may be necessary to drive interaction in polycentric systems (Reference Hale and RogerHale & Roger, 2013; Reference Jordan and SchoutJordan & Schout, 2006). To make this happen, “political pressure” or some resource provision from “on high” may be needed (Reference Jordan and SchoutJordan & Schout, 2006, p. 271).

Polycentric governance scholars have over time acknowledged the need for “higher-level institutions” to some extent. In her work on polycentrism, Elinor Ostrom advocates a subtle blend of self-organization by local actors and “some larger-scale jurisdiction” (Reference OstromOstrom, 2005, p. 282). Ostrom is less clear, however, on the origin and precise nature of this “larger-scale jurisdiction.” On the one hand, she argues that sometimes preexisting higher governance levels (e.g., state structures) are ineffective and it may therefore be advantageous to grow higher-level structures from lower levels: “Success in starting small-scale initial institutions enables a group of individuals to build on the social capital thus created to solve larger problems with larger and more complex institutional arrangements” (Reference OstromOstrom, 1990, p. 190). On the other hand, she argues on the same page that in a key case study on Californian water governance, recourse to preexisting institutions such as the (public) court system proved vital in fostering self-organization among local actors. In a similar vein, Reference AligicaAligica (2014) stresses “an over-arching system of rules” (p. 57) as one of the “three basic features” of polycentrism (p. 58) – which may be an “institutional and cultural framework” (p. 58) that determines who participates in a polycentric governance system (p. 59).

Based on the latter reasoning, Reference MansbridgeMansbridge (2014) argues that Elinor Ostrom in fact frequently alluded to higher-level governance functions that are often – but not always – conducted by states. Based on her reading of Ostrom, Mansbridge emphasizes that: “Ostrom’s polycentric model assumes some levels higher than the local, which can threaten to impose other solutions, provide neutral information, provide venues and support for the local negotiation, and, crucially, sanction non-compliance” (p. 9). Mansbridge goes on to argue that more traditional public actors including states may deliver some or all of these four functions. Notably, although Mansbridge does not specifically define what she means by “the state,” her discussion of fairly wide-ranging functions included in the above quote appears to allude to a broad definition of what precise institutions are thought to form part of the state. This is in line with a relatively broad description in the Concise Oxford Dictionary of Politics, which defines “the state” as “[a] distinct set of political institutions whose specific concern is with the organization of domination, in the name of the common interest, within a delimited territory” (Reference Burnham, McLean and McMillanBurnham, 2009). Taken together, scholars working in the polycentric tradition would conceive of the state fairly broadly, including institutions forming the legislative, executive, and judicative branches. In sum, coordination or “taking each other into account” may in some cases happen automatically, but in others requires conscious effort and coordination. These questions have a direct bearing on the central questions of this book, namely where these “lessons” are going to emerge from (i.e., who generates the lessons) and whether the lessons are provided in a way that can at least in principle enable lesson-drawing across governance centers (and thus the shift from polycentricity to polycentrism that Chapter 1 explains).

Crucially for this book, a focus on information provision and enforcement via monitoring is a central and explicit component in polycentric governance theory. As Elinor Reference OstromOstrom (1999) explains, “If all self-organized resource governance systems are totally independent and there is no communication among them, then each has to learn through its own trial-and-error process” (p. 525). Some scholars highlight “that a polycentric arrangement has a built-in mechanism of self-correction” (Reference AligicaAligica, 2014, p. 48) and advance the (big) claim that “reflexivity is a systemic feature” (Reference AligicaAligica, 2014, p. 66). As Elinor Reference OstromOstrom (1999) writes, “Thus, a self-organized resource governance system with a higher level of in-migration or greater communication with other localities is more likely to adapt and change rules over time than is a system where new ideas concerning how to use rules as tools are rarely brought in” (p. 525). But because reflexivity requires knowledge and critique of ongoing approaches, it depends on mechanisms to provide that knowledge. Otherwise, polycentrism, or “a system of reciprocal monitoring and assessment in dynamic interdependence” (Reference AligicaAligica, 2014, p. 66) may not materialize. But who will provide this information, will it appear with or without central stimulation, and will what emerges be of sufficient quality to be useful? Reference AligicaAligica (2014) was rather optimistic, assuming that “A system of ‘reciprocal monitoring and assessment for the range of institutions available in society’ is thus put spontaneously in place, but in addition a system of broad checks and balances emerges” (Reference AligicaAligica, 2014, p. 66). Others, such as Reference MansbridgeMansbridge (2014), envision a much stronger role for traditional public actors such as states, which could “help monitor compliance and sanction defection in the implementation phase” (p. 8, emphasis added). However, alternatively, states or other governance actors may shy away from the costs of collecting information about the experience in other governance centers or from making relevant changes once they know that another approach may generate better results. Thus, taken together, the question that runs through the literatures on common pool resources, polycentrism and policy coordination centers on who provides “collective” or “public” goods, which may include the extent to which governance centers monitor their own practices and in turn pay attention to one another in order to learn and, perhaps, coordinate their activities.

2.4 Monitoring: From Common-Pool Resources to Climate Policy

Common pool resource scholars in the polycentric governance tradition highlight that monitoring is an absolutely essential part of successful CPR governance. As Elinor Reference OstromOstrom (1990, p. 45) emphasizes, “[w]ithout monitoring, there can be no credible commitment; without credible commitment, there is no reason to propose new rules.” A fairly general definition holds that monitoring may be defined as “[a] continuing function that uses systematic collection of data on specified indicators to provide … indications of the extent of progress and achievement of objectives and progress in the use of allocated funds” (OECD-DAC, 2002, pp. 27–28). In other words, monitoring refers to “recipe[s] for the selection, organization and retention of large amounts of information” (Reference Dahler-Larsen, Rist and StameDahler-Larsen, 2011, p. 65). Elinor Ostrom strongly links monitoring with the idea of preventing rule defections (i.e., policing).

But what makes monitoring particularly successful? Evidence from resource management literatures suggests that there is no general recipe for organizing monitoring activities. For example, Reference Ostrom and NagendraOstrom and Nagendra (2007) use multiple methods to show that the success of forest management depends critically on the fit of monitoring institutions with wider ecological, social, and political environments (or context, see above Section 2.3). Furthermore, the success of a monitoring regime often hinges on whether it is perceived as legitimate, which tends to be the case when people who are affected by the regime are involved in its creation and maintenance (Reference Ostrom and NagendraOstrom & Nagendra, 2007). Participants may then even be willing to bear some of the cost of monitoring themselves (Reference Ostrom and NagendraOstrom & Nagendra, 2007). The key lesson to take from these smaller-scale studies is that in some cases, decentralized monitoring appears to work “better” than centralized activities for the reasons outlined above. But, again, the success of a particular monitoring regime depends critically on its fit with the particular context, including existing institutions, cultures and the nature of the resource. When monitoring is successful, it can not only prevent rule defections, but also provide knowledge that may be of use to other governance centers – driven by self-organizing actors.

When moving to larger common-pool resources (such as the atmosphere and a stable climate), Elinor Ostrom argues that the more successful governance systems tend to organize “appropriation, provision, monitoring, enforcement, conflict resolution, and governance activities … in multiple layers of nested enterprise” (Reference OstromOstrom, 1990, 101, emphasis added). This is because “[e]stablishing rules at one level, without rules at the other levels, will produce an incomplete system that may not endure over the long run” (Reference OstromOstrom, 1990, p. 102). Thus, monitoring by a single actor at a single level is unlikely to work in these instances.

At any level, monitoring is neither an easy nor a “cheap” activity (Reference LeeuwLeeuw, 2010; Reference OstromOstrom, 1990; Reference Schoenefeld, Schulze, Hildén and JordanSchoenefeld et al., 2019, Reference Schoenefeld2021; Reference Schoenefeld and JordanSchoenefeld & Jordan, 2020). Reference Kusek, Rist and MathisonKusek and Rist (2005, p. 301) have noted that “[t]he reality is that putting in place even a rudimentary system of monitoring, evaluating, and reporting on government performance is not easy in the best of circumstances.” Whether monitoring natural resource use or public policy, doing so requires significant and sustained effort, time, resources, and buy-in by multiple parties to set up and operate monitoring activities (Reference OstromOstrom, 1990, p. 202). But not all monitoring activities are created equal. Importantly, Elinor Reference OstromOstrom (1990) argues that for natural resources, “Monitoring costs are affected by the physical attributes of the resource itself, the technology available for exclusion and appropriation, marketing arrangements, the proposed rules, and the legitimacy bestowed by external authorities on the results of institutional choices …” (p. 203). Furthermore, “[f]actors that enhance the capacity of users to see or hear one another as they are engaged in appropriation activities tend to lower monitoring and enforcement costs” (Reference OstromOstrom, 1990, p. 204). Additionally, “[t]he availability of low-cost facilities for recording and disseminating information about regulated activities will also decrease monitoring costs” (Reference OstromOstrom, 1990, p. 204). In other words, the more detectable an activity – and potential rule-breaking – is, the easier it is to monitor.

The physical size of a resource also has a strong bearing on monitoring. Generally, “[t]he larger the resource, the greater the costs of ‘fencing’ and/or patrolling the boundaries to ensure that no outsider appropriates” (Reference OstromOstrom, 1990, p. 203). And if frequent monitoring is required, costs tend to increase (Reference OstromOstrom, 1990, p. 204). Finally, it is important to recognize that the nature of the rules to be monitored also affects the ease of monitoring: “Rules that unambiguously state that some action – no matter who undertakes it – is proscribed are less costly to monitor than are rules that require more information about who is pursuing a particular behavior and why” (Reference OstromOstrom, 1990, p. 204). Furthermore, “[r]ules that place a limit on the quantity of resource units that can be produced during an entire season or year are costlier to enforce” (Reference OstromOstrom, 1990, p. 205). The smaller and the more visible a resource and its use are, and the clearer the rules that govern it, the easier it is to monitor.

In cases where more technical scientific knowledge may be required to monitor a resource (such as overall fish stocks to determine fishing quotas), Elinor Ostrom points to the self-organizing capacities of local actors through community organizations. She argues that

Information generated in this way may be more sensitive to the interests and needs of the local actors who fund them – and help systemic learning. Crucially, “Associations of local resource governance units can be encouraged to speed up the exchange of information about relevant local conditions and about policy experiments that have proved particularly successful” (Reference OstromOstrom, 2005, p. 283). Self-organization may in turn support interactions between governance centers.

There is thus a strong argument to consider the “institutional fit” between what is being monitored and the institutions to do so. As Reference Keohane and OstromKeohane and Ostrom (1994) explain: “Another implication of research on local CPRs and public goods and on international regimes for international environmental institutions is the importance of achieving a match between the characteristics of a successful monitoring and sanctioning scheme and the characteristics of specific situations” (p. 22; emphasis added). Table 2.1 summarizes the key insights from monitoring common pool resources with a view to applying them to monitoring climate policy in the next section in light of the three foundational ideas of polycentric governance theory identified above. For example, the nature of the resource relates to context, whereas information exchange through associations relates to interacting governance centers.

What can we glean from these insights on monitoring natural resources for monitoring climate change policy? A first thing to note is that humans cannot readily detect carbon dioxide and other greenhouse gases without significant technical equipment, making monitoring technically much more challenging than, say, monitoring the number of fish that have been taken out of a fishery. Monitoring greenhouse gases requires significant expertise and equipment, and has been subject to contestation, especially when there are direct policy consequences of monitoring decisions. For example, Canada and the EU have quarreled intensely about the greenhouse gas content of tar sand oil (e.g., Reference NeslenNeslen, 2011). It can also be extremely costly to accurately measure or estimate carbon emissions from certain sources – and may thus not viable in some cases (Öko-Institut et al., 2012). Second, the expanse of the atmosphere is vast and it is thus exceedingly difficult to establish boundaries for monitoring and “appropriation.” Following Elinor Ostrom’s rationale, the physical nature of greenhouse gases makes monitoring their emissions rather challenging. It is hard to imagine how individuals may conduct such highly complex policy evaluations as they have been shown to do when monitoring individual resource governance.

But monitoring greenhouse gas fluxes is only one way of looking at policy outcomes, as other factors, such as impacts on congestion, public health, or employment are often equally at the center of policy discussions – and the “goals” of a policy may indeed be subject to significant contestation. Similarly, supply-oriented climate policy aims to leave significant amounts of hydrocarbons in the ground. Monitoring complex outcomes such as “public health” typically requires the use of indicators, which “summarize or otherwise simplify relevant information, make … visible or perceptible phenomena of interest, and quantify, measure, and communicate relevant information” (Reference GallopínGallopín, 1996, p. 108). Using indicators to monitor policies is by no means a politically “neutral” or “innocent” activity, because these indicators embody underlying value orientations regarding what matters and what does not (Reference GudmundssonGudmundsson, 2003; Reference Lehtonen, Jordan and TurnpennyLehtonen, 2015) and are frequently constructed from information that is either readily available or can be generated (Reference GallopínGallopín, 1996). Even choosing indicators such as greenhouse gas emission reductions to compare climate policies embodies a deeply normative choice (Reference Schoenefeld, Hildén and JordanSchoenefeld et al., 2018). The key difference is thus that CPR monitoring can often rely on direct measurement and observation of appropriation, whereas monitoring climate policies requires other tools to do so; and the goals of policy may be multifarious and sometimes fuzzy.

Furthermore, monitoring the effects of climate policies differs from that of common pool resources because policing and detecting rule defection is only one and possibly not the major objective of monitoring, which may also aim at learning, an aspect that features only partially in Elinor Ostrom’s discussions of monitoring. Related to the idea of indicators, climate policies may also generate a range of intended and unintended effects and potentially interact with other policies – as discussed above, it is thus often necessary to use (multiple) indicators rather than direct observation; and it involves many more actors and jurisdictions. Last, because much may be at stake, climate policy monitoring tends to be so politically sensitive (Reference Schoenefeld, Hildén and JordanSchoenefeld et al., 2018; Reference Schoenefeld, Schulze, Hildén and JordanSchoenefeld et al., 2019) that top-down monitoring has proven difficult if not impossible to install at the international level (see also Reference Schoenefeld and JordanSchoenefeld & Jordan, 2017).

In order to apply insights from CPR monitoring to climate policy monitoring, it is first necessary to somewhat relax the definition of “local.” Clearly, the idea of local monitoring where one fisher(wo)man may observe the behavior of her or his colleagues only has limited value when considering the monitoring of national, regional, or even local climate policy. But if one allows the idea of more localized monitoring to apply to the nation state, it quickly becomes clear that some states and/or regions (and actors therein) do have the capacity to monitor their own climate policies. The same is also true for some subnational actors, as monitoring at the city or the company level has shown. Thus, actors at “more local” levels (here understood as national versus international) may be better placed – and viewed as more legitimate – to regulate their own actions. This view is certainly also in the spirit of the Paris Agreement, which relies on nation states putting forward their own contributions and assessing their progress over time (Reference Mor and GhimireMor & Ghimire, 2022; Reference Schoenefeld, Hildén and JordanSchoenefeld et al., 2018). Similarly, when “self-organization” is understood as an activity done at the nation state level (or certain actors within the nation state, which nevertheless do not necessarily have to be individuals), it becomes more feasible to apply these concepts.

Table 2.2 summarizes the key conclusions from the discussion in this section. Similar to what was done above, it organizes the points by the three foundational ideas, namely self-organization, context, and interacting governance centers.

Drawing on Table 2.2, scholars working in the polycentric governance tradition would thus likely ask with respect to climate policy monitoring: how do actors monitor climate policy, what do they include (or ignore), who conducts the monitoring, and how do those engaged in monitoring interact with one another and their context? These three core ideas relate closely to the foundational ideas of polycentrism, and thus become the basis for discussing the role of policy evaluation in the following section.

2.5 Evaluation in Polycentric Governance Systems

Some form of knowledge generation on the effectiveness of policy approaches in different governance centers is part and parcel of polycentric governance. Empirical research on monitoring in common-pool resource systems (see above Section 2.4) contains necessary, but not yet sufficient insights to interrogate what role – if any – policy evaluation could and potentially already does play in polycentric governance systems. Compared with the definition of monitoring at the beginning of this section, ex post policy evaluation is a related, and yet substantially different activity. Recall that this book follows Reference VedungVedung (1997, p. 3), who defines policy evaluation as the “careful retrospective assessment of the merit, worth, and value of administration, output and outcome of government interventions, which is intended to play a role in future practical action situations” (see Chapter 1). Monitoring data may be an ingredient of evaluation, but evaluation goes a key step further than monitoring in making a value-based assessment, and evaluation can take a much broader view and consider factors and data that limited monitoring may struggle to detect.

Policy evaluation is thus a broader activity than monitoring, and therefore, its role in polycentric governance system must also be considered in broader terms (see Reference Schoenefeld and JordanSchoenefeld & Jordan, 2019). There are two headline reasons why policy evaluation may, in principle, play a role in polycentric settings – and which are also frequently cited as reasons for evaluating to begin with (see Reference Borrás and HøjlundBorrás & Højlund, 2015; Reference SandersonSanderson, 2002): first, related to Elinor Ostrom’s ideas about detecting rule defection via monitoring, policy evaluation may play a role in enabling accountability relationships in polycentric systems (Reference Versluis, van Keulen and StephensonVersluis et al., 2011, p. 206). Reference BovensBovens (2007) defines accountability as “a relationship between an actor and a forum, in which the actor has an obligation to explain and to justify his or her conduct, the forum can pose questions and pass judgment, and the actor may face consequences” (p. 107). A key issue in “new” governance contexts – including potentially polycentric governance – is that traditional forms of accountability, which are enacted through often long principal-agent chains, are becoming increasingly problematic (Reference StameStame, 2006). Whereas democratic states usually boast civil servants who answer to elected leaders who in turn answer to parliament, which itself answers to its voters, such a conceptualization of accountability struggles in polycentric settings, where it may be much less clear who answers to whom. This state of affairs has given rise to “new forms of accountability,” such as diagonal or horizontal ones where policymakers may be accountable to civil society or to ombudspersons (Reference BovensBovens, 2007). From this perspective, policy evaluation may enable accountability in polycentric settings (Reference BovensBovens, 2007). Reference Alkin, Christie and AlkinAlkin and Christie (2004, p. 12) have also highlighted that “[t]he need and desire for accountability presents a need for evaluation.” Relatedly, Reference HanbergerHanberger (2012) focuses on the role of evaluation to support political accountability in different governance systems, including state systems, regional-local systems, and network governance. Policy evaluation may thus make a significant contribution to enabling accountability (Reference FischerFischer, 2006). But more than providers of “objective” policy information, evaluators may also be seen as mediators between societal discourses and discussion about the merit of particular policies to achieve a number of different – and layered – goals (Reference FischerFischer, 2006). In this model, evaluators are not aloof from society, but inextricably bound up and working within and through a system of values and facts that are at stake when a policy is evaluated.

The second, and certainly no less widely discussed reason why policy evaluation may play a role in polycentric governance systems is as an enabler of learning (see Section 2.2 above for a definition and discussion of the concept). Scholars have already highlighted potential links between ex post evaluation and learning. For example, Reference HaugHaug (2015, p. 5) stresses that “[e]x-post evaluation of programmes or policies … is a widely applied group of approaches aimed at stimulating learning in environmental governance.” There is still an ongoing and largely unresolved debate on what exactly is learned, which depending on one’s philosophical position may range from “facts” to learning about value-based discourses (Reference Borrás and HøjlundBorrás & Højlund, 2015; Reference HaugHaug, 2015; Reference SandersonSanderson, 2002). This book focuses on the learning-related factors that feature most strongly in debates on polycentric governance, namely the importance of context, as well as learning as one vehicle of interactions between governance centers (see above Section 2.3).

While accountability and learning may be two theoretically relevant concepts for understanding a potential role of policy evaluation in polycentric governance systems, it is important to recognize that policy evaluation happens in a political environment and may therefore also be done for political – that is strategic and “irrational” – reasons that have little to do with either accountability or learning. “Political” in this context is understood as both processes and struggles that happen inside familiar governmental arenas, but also as a more pervasive process that happens when power operates, and regarding what is discussed and addressed in public and what is not (Reference HayHay, 2002; Reference LukesLukes, 2005; Reference Mansbridge and MacedoMansbridge, 1999, p. 214). Numerous scholars have already highlighted the political characteristics of policy evaluation (Reference Bovens, Hart, Kuipers, Moran, Rein and GoodinBovens et al., 2006; Reference GreeneGreene, 1997; Reference House and HoweHouse & Howe, 1999; Reference Lascoumes and Le GalesLascoumes & Le Gales, 2007; Reference Nilsson, Jordan, Turnpenny, Hertin, Nykvist and RusselNilsson et al., 2008; Reference Owens, Rayner and BinaOwens et al., 2004; Reference VedungVedung, 1997). First, Reference WeissWeiss (1993) argued that because government programs emerge through political processes, political pressures are unlikely to disappear at the evaluation stage (though they could arguably change over time). Second, “as social scientists increasingly recognize, no study collects neutral ‘facts’: all research entails value decisions and to some degree reflects the researcher’s selections, assumptions, and interpretations” (Reference WeissWeiss, 1993, p. 102). Third, policy evaluation may also be political because it has the potential to affect the range of decisions political actors can take and thus act as a “destabilizing” force (which links with the points on “dissatisfaction” with regard to lesson-drawing above in Section 2.3). For example, a sympathetic evaluation may provide vital support to continue or extend a climate policy, whereas a negative evaluation may deprive decision-makers of the possibility to do so and can potentially lead to policy dismantling (Reference Gravey and JordanGravey & Jordan, 2016). Fourth, because policy evaluation has the potential to affect resource distribution across society (Reference Bovens, Hart, Kuipers, Moran, Rein and GoodinBovens et al., 2006), it may be used in a strategic fashion such as to delay a political process or to move a decision to another forum. Thus, policy evaluation is political because it operates in a political context, can destabilize resource distribution, and can be used in a strategic way (see Schoenefeld & Jordan, 2019).

These political, and often strategic and normative elements of evaluation generate crucial but difficult questions for the role of policy evaluation in polycentric governance settings. If evaluation is done for more strategic and political reasons (see also Reference PollittPollitt, 1998), then its outcome may be less than optimal from a polycentric governance perspective, and thus expectations towards evaluation may have to be tempered. By the same token, in situations of considerable political contestation, it is also possible that evaluative knowledge may emerge through self-organizing capacities by individual governance actors (see below Section 2.5.1). For example, society-driven actors may conduct or commission their “own” evaluations in order to contest points made by state-driven evaluations. A whole range of evaluations may therefore generate a more “complete” body of evaluative knowledge that does not rely on a single perspective. Thus, a polycentric governance perspective on evaluation would highlight the need for a broad range of evaluation perspectives and actors so as to generate diverse knowledge of policy effects.

Against this background, the remainder of this section reviews debates in existing literature on policy evaluation insofar as they relate to the three foundational ideas of polycentrism, namely self-organization, context, and interaction between governance centers. Where pertinent, the review connects with ideas on accountability and learning, while keeping in mind the political nature of policy evaluation.

2.6 Conclusion

The previous sections endeavored to make a theoretical case for examining the importance and actual roles of policy evaluation in facilitating climate governance by contributing to the shift from polycentricity to polycentrism. They show that literatures on polycentric governance and policy evaluation have already engaged with concepts that are highly relevant to this question, yet often ill developed and with little connection to the body of literature on the other side. The respective debates have by and large taken place in relatively self-contained, and often self-referential, scholarly communities with their own set of dedicated journals, conferences, and networks. For example, evaluation literatures have already debated the role of context in evaluation, as well as the role of multiple actors and – to a much lesser extent – the notion of interacting governance centers. But to date there is a severe paucity of studies that consider these factors together. Insights from this kind of integrative research across different factors could help shed light on the potential and actual roles of (climate) policy evaluation in polycentric governance systems. The above review shows how information provision via policy evaluation is in many ways implicit in Elinor and Vincent Ostrom’s polycentric governance theory, but its precise role and to what extent this happens in practice have yet to be explored.

This chapter set out to identify the basic theoretical building blocks of polycentrism, which as a theory contains both normative and positive elements. The foundational insights are: (1) context matters in governance; (2) actors can and sometimes do self-organize to muster governance solutions; and (3) interaction between otherwise independent governance centers appears indispensable in order to move from polycentricity to polycentrism (see Chapter 1). Bearing in mind arguments about scale in governance, the chapter shows that we can draw key theoretical insights from monitoring studies in common pool resource governance systems in order to conceptualize the role of policy evaluation in polycentric governance systems. Crucially, policy evaluation can potentially make significant contributions to the emergence of polycentrism, but in order to do so, it must exhibit certain features outlined in the sections above. Moving forward, this newly developed theoretical approach thus provides some yardsticks against which we can evaluate the practice of climate policy evaluation in the next chapters.