The Utility of Combining the IAD and SES Frameworks

1. Introduction

Elinor Ostrom played a pivotal role in developing two analytical frameworks used by many policy analysts and scholars: the IAD (Institutional Analysis and Development) and the SES (Social-Ecological Systems) frameworks. Both frameworks have facilitated collaborative research and communication among scholars and practitioners from a wide range of disciplinary and professional backgrounds. However, certain aspects of each can result in confusion among readers unfamiliar with the Bloomington School of Political Economy (Aligica and Boettke 2009), or result in applications that only partially realize the full potential of either framework. In this paper we aim to clear up these confusions and encourage users to pursue more sophisticated analyses based on an integrated framework, namely, the Combined IAD-SES Framework, or CIS for short.

We begin with brief summaries of the foundational components of the IAD and SES frameworks, and use elements of each framework to serve as a counterweight to problematic tendencies commonly found in applications of the other. At their core, the IAD and SES are very closely related frameworks, and our discussion highlights their fundamental similarities while attempting to overcome their separate limitations. But our aspirations are limited: we do not pretend to offer a fully developed causal theory or formal model.¹ We are nonetheless confident that the particular combination we propose here can serve as a solid foundation for advances in the study of how institutions and policy outcomes evolve in tandem.

In an effort to forestall misunderstandings, we apply this combined framework to a case familiar to many scholars interested in the origins and operation of self-organized systems of resource governance, namely, lobster fisheries off the coast of Maine. We conclude with a brief discussion of related lines of research by scholars drawing upon other frameworks, in order to locate this CIS framework within its proper context.

2. The IAD framework

The IAD framework that Elinor Ostrom developed alongside colleagues from the Workshop in Political Theory and Policy Analysis has been widely adopted by social scientists to evaluate the effects of alternative institutional arrangements, and to investigate the processes by which institutions emerge and change over time (see, e.g. Tang 1992; Rudd 2004; Andersson 2006; Ostrom 2007a; Ostrom and Cox 2010). It is widely recognized as one of the leading frameworks in the policy sciences (Weible and Sabatier 2018) and provided the analytical foundation upon which Elinor Ostrom built a collaborative research program on community-based management of natural resources, for which she was named a co-recipient of the 2009 Nobel Memorial Prize in Economic Sciences. Ostrom (1986, 1989, 2005, 2007a, 2010, 2011; see also Kiser and Ostrom 1982) envisioned this framework as a means whereby scholars from multiple disciplines could more effectively communicate with one another, regardless of their diverse perspectives, to better understand complex policy settings.

As shown in Figure 1, the IAD framework represents institutional processes by a series of boxes within which different variables or processes are located. At the heart of the IAD framework is an action situation, an abstraction of decision environments in which individuals and corporate actors interact with one another, making choices that jointly determine outcomes of some aspects of a policy question. Individual choices and collective outcomes are influenced by the beliefs and incentives of the relevant individuals, as shaped by the responsibilities and social expectations attached to official positions they may hold, and by the information available to them.

The specific nature of the decision problem faced by actors within a particular action situation is shaped by preexisting contextual conditions,² grouped for analytical purposes into three categories: (1) the “nature of the good” under consideration, including all relevant biophysical conditions; (2) “attributes of the community,” social ties and the cultural context within which individuals interact and (3) “rules-in-use,” the existing configuration of laws, regulations, rules, norms, and shared understandings held by the participants to be relevant to their deliberations. Outcomes from an action situation are evaluated by the relevant actors (some of whom may not have been involved in the original decision). Feedback from these outcomes and evaluations can reinforce or induce changes in these contextual conditions, which in turn sets the stage for future interactions.

The IAD framework differentiates among three different types (or levels) of action situations: (1) operational-choice settings in which actors’ choices directly impact tangible outcomes, (2) policymaking or collective-choice settings in which actors shape the rules that constrain actors in operational-choice arenas, and (3) settings for constitutional-choice in which decisions are made concerning which actors have a stand in different choice situations as well as which kinds of alternative institutional mechanisms are available to them as they make their collective deliberations and operational-level choices (Ostrom 2005, 58–62). In all types of action situations, biophysical, legal-institutional, and socio-cultural factors interact in complex ways to shape patterns of interactions and outcomes.

Each action situation describes a nexus in which a group of decision makers jointly confront important decisions related to some policy concerns. Potential outcomes are differentially valued by actors with partial control over the final determination of results. Ostrom (1986, 2005) explicitly frames an action situation as a generalization of standard game models. To define a game, modelers must specify the actors involved, the information available to them, their options and the outcomes their choices can jointly generate. Similarly, an action situation is configured by interlocking “working components,” related in the following manner:

Participants, who can either be individuals or any of a wide diversity of organized entities, are assigned to positions. In these positions, they choose among actions in light of their information, the control they have over action-outcome linkages, and the benefits and costs assigned to actions and outcomes. (Ostrom 2005, 188; italics added)

The specific nature of each of these italicized components will have been determined by processes occurring previously or simultaneously in other strategic interactions, that is, in other action situations occurring at any of the three levels of choice.

The strength of the IAD framework stems from “its systematic theoretical focus on the impact of rules and norms on individual incentives in complex systems, its emphasis on empirical analysis of outcomes and consideration of multiple dynamic systemic processes that jointly affect those outcomes” (Rudd 2004). In short, the IAD framework recognizes the complexity of the world, but provides researchers with a tool to cope with this complexity and integrate knowledge across multiple studies and methods. The framework is sufficiently general to explore collective action problems of all kinds, but its greatest influence has been in the area of environmental policy where it was used to uncover the institutional foundations of sustainable community-based natural resource management (Ostrom 1990; Gibson et al. 2000; Cox et al. 2010). Nonetheless the framework has also been used, albeit to a lesser extent, to explore the processes by which institutions for common-pool resource management emerge (Ostrom 1990, Chapter 4; Lubell et al. 2002; Grossman 2019), and understand the factors influencing the delivery of public services in metropolitan areas (Bushouse 2011; Oakerson and Parks 2011).

Although most published applications of the IAD framework include a version of Figure 1, that figure inappropriately conveys the image of an atomistic action situation, which is not how Ostrom intended this framework to be understood. Her writings clearly stress that no action situation exists in total isolation from others (see Kiser and Ostrom 1982; Ostrom 1986, 1989, 2005). Figure 2 shows how a single action situation can be connected to a series of “adjacent” action situations, each of which has the effect of setting the values for one or more of the working components that define that focal action situation. In general, no single action situation can be fully understood without careful consideration of the web within which it resides.

A close examination of the “canonical” IAD representation of Figure 1 reveals an implicit clue that points towards the necessary involvement of at least one more action situation beyond the one being represented. Evaluative criteria do not come out of nowhere, but must have been determined by actors establishing the relevance of alternative evaluative criteria to particular situations. Presumably, individual actors in the focal action situation will engage in various forms of evaluation as they consider their next choices. In addition, for any single (type of) action situation, there must exist at least one corresponding collective choice arena in which actors have, either individually or as a group, selected a list of criteria they deem appropriate for evaluation of its outcomes.

Figure 2 does not indicate that each action situation is associated with its own unique categories of contextual conditions. Instead, it is more useful to presume that many of the same biophysical conditions, community attributes, and rules-in-use may apply to multiple nodes in any network of adjacent action situations. Implicitly, then, all factors within a given category that are deemed to be relevant to at least one action situation in that network could, for analytical purposes, be collected together into a comprehensive list of relevant variables within that category. Then, the specific contextual conditions applying to any particular action situation would be a subset of factors drawn from these comprehensive lists. This same idea was incorporated into the SES framework.

3. The SES framework

The IAD framework was promoted as an interdisciplinary research tool with which social and natural scientists could build knowledge concerning the sustainability of social-ecological systems. However, after many years of criticism and complaints that the IAD framework paid insufficient attention to the diversity and complexity of natural systems and processes (Berkes and Folke 1998; Young 2002; Agrawal 2003; Ostrom and Cox 2010, 6) and to alternative social science theories (such as Young 2002), Ostrom and her colleagues began developing a framework specifically designed for the analysis of closely-coupled social-ecological frameworks (Anderies et al. 2004; Ostrom 2007b, 2009; Ostrom et al. 2007; Ostrom and Cox 2010; McGinnis and Ostrom 2014).

Figure 3 shows the basic components of the SES framework, as modified by Epstein et al. (2013). It includes seven categories of attributes of any social-ecological system: Resource Systems (RS), Resource Units (RU), Governance Systems (GS), Actors (A), Related Social, Economic and Political Systems (S), Ecological Rules (ER), and Related Ecosystems (ECO). Those first-tier components are further decomposable into second-, third-, and in some cases even fourth-tier variables (as in Table 1). This nested structure supports finely-grained analyses of social and ecological conditions, which is the chief strength of the SES framework.

The SES framework has not been applied to as wide a range of policy settings as IAD (Schlager and Cox 2018), but nonetheless has been used to investigate the governance of forests, fisheries, irrigation and rangelands (i.e. Ostrom and Cox 2010; Fleischman et al. 2010; Gutierrez et al. 2011; Addison and Greiner 2016). Although it was intended to facilitate consistent data coding for purposes of assessments across numbers of cases (see, e.g. Ostrom and Cox 2010, 9; Fleischman et al. 2014; Ban et al. 2017), considerable difficulties remain to coordinate the collection and coding of data by different research teams (Thiel et al. 2015; Frey 2017; Gari et al. 2018; Partelow 2018).

The core element of the IAD framework–action situations–lies at the very heart of later versions of the SES figures to which Ostrom contributed directly (Ostrom and Cox 2010; McGinnis and Ostrom 2014). This potentially gives the SES a deeply dynamic orientation. Unfortunately, most applications of the SES framework generate long lists of factors that may have contributed to one or more outcomes (Basurto and Ostrom 2009; Fleischman 2010; Blanco 2011; Gutierrez et al. 2011; Ban et al. 2017). Although this approach could be used to identify combinations of SES attributes associated with the full range of relevant outcomes, these lists of variables can provide no basis for understanding the process by which these social, institutional and ecological factors combined to shape collective-choice and operational processes to generate those outcomes, let alone predicting or prescribing changes to social-ecological conditions over time. As a result, scholars are left to choose between a finely-detailed but ultimately static analysis with the SES framework or a more dynamic but underspecified analysis with the IAD framework.

SES improves upon IAD by expanding the number of categories of relevant and decomposable, contextual variables. A close comparison of Ostrom’s two frameworks reveals that many, and perhaps all, of the SES variables are contained, if only implicitly, within the IAD framework. Two IAD categories are directly analogous to first-tier SES categories: “attributes of the community” include the attributes of actors (and their connections) and the term “rules-in-use” has often been used to encompass all aspects of a governance system. Furthermore, the IAD category of biophysical conditions (a generalization of the initial “nature of the good” label) could be divided into two parts, one referring to specific resource units and the other to the resource system as a whole. In this way, each of the first-tier categories of the SES framework closely match the “Biophysical Conditions,” “Community Attributes,” and “Rules-in-Use” boxes of the IAD framework. Finally, the remaining categories for other social or ecological processes in the SES framework can be seen as holders for the potential effects of action situations remote from focal action situations that lie at the heart of any empirical analysis.

Yet, with a few exceptions (Cox 2014b; Vogt et al. 2015; McCord et al. 2017) most applications of the SES framework fail to attend to the critical importance of identifying and analyzing sets of focal action situations. As shown in Table 1, the variables listed under the Action Situation column on the right are not comparable to those listed in the columns corresponding to high-level categories of contextual or explanatory variables. Instead, a range of functionally different types of action situations are listed, without any attention to how these action situations are likely to differ from one another, or to the density of connections amongst them. The core attributes of an action situation are the working components, as shown in Figure 2, but that aspect of the IAD toolbox is generally missing from SES-based applications.

The literature on institutional analysis of social-ecological systems is currently dominated by two strands of research: (1) variable-oriented research which identifies configurations of social-ecological attributes associated with outcomes; and (2) process-oriented descriptive case studies that describe how institutions emerge and change, influencing behaviour and outcomes with very little interplay between the two approaches. This paper can, and in at least one case already has (McCord et al. 2017), help to address this gap by encouraging variable-oriented researchers to be more attentive to processes, and vice versa. But our help can only take us so far, because the purpose of a framework is not to explain outcomes, but to provide a tool for identifying, categorizing and organizing variables and processes for analysis. The CIS framework can help scholars to organize and integrate insights from variable-oriented research (i.e. most applications of the SES framework) and process-oriented research involving networks of action situations, but it does not in and of itself provide tools for inference.

4. An overview of the combined IAD-SES framework

We argue that anyone using the SES framework should carefully examine all action situations directly germane to the operation of that system and the values of the working components (see the interior part of Figure 2) that comprise it. They should also dig more deeply into the system to identify how these situations are interlinked, with particular attention to locating adjacent action situations that are responsible for the functional components within a focal action situation. The combined framework we outline here is primarily intended to help SES researchers make more effective use of the analytical strengths of the IAD framework by shifting the focus towards the analysis of action situations rather than variables.

Although Ostrom did not use the term “action situation” in her initial versions of SES (Ostrom 2007, 2009), she responded positively to suggestions by colleagues in the informal “SES Club” to explicitly add that term (see McGinnis and Ostrom 2014). Her original figures had placed interactions and outcomes at the center of the SES framework, but when she added the action situation term, she collapsed action situations, patterns of interactions, and outcomes (each of which had its own separate box in the IAD) into a single entity. By applying the same aggregation to the IAD framework of Figure 1, we end up with a much simpler version of the IAD which helped inspire the combined framework we offer here.

The canonical form of the IAD framework has multiple paths through which feedbacks can occur, with many of those feedback paths connecting outcomes and interactions to action situations and outcomes to potential changes in the pre-existing contextual conditions. All but the last of these feedback paths would lie within a single box, if we were to follow the SES expedient of aggregating action, interactions, and outcomes together in a single amalgam. This one link between outcomes and contextual conditions is worth careful consideration because this is the path through which the contextual conditions are endogenously determined within the broader network of adjacent action situations.

The simplicity of this one tie-back to contextual conditions enables us to rearrange the overall figure in a way that more directly represents the natural temporal order of before, during, and after the operation of action situations within the SES-style amalgam. That is, we can locate a network of action situations between an array of boxes on the left, representing the context from which those action situations emerged, and a similar list of the same boxes on the right, to denote how those variables had been changed by feedback from the actions, interactions, and outcomes occurring in that network. Figure 4 shows this simple arrangement of a set of contextual factors setting the stage for a network of interconnected strategic action, interactions, and outcomes, with the outcomes of those interactions feeding forward to reset conditions for the next iteration of the relevant action situations. This seems a much more natural representation of the temporal sequence of before, during, and after.

As we argued above, the evaluative action situations in which the evaluative criteria were determined should also be included, where possible, within a network of focal action situations. There are two potential pathways through which this feedback or, more appropriately, “feed forward” into the next iteration, can occur. Feedback that has been shaped by the application of existing evaluative criteria can be interpreted as following a more indirect path than would feedback that directly changes the values of contextual factors. Such direct effects need not involve conscious intervention on the part of the humans involved in this system and could instead operate independently of human agency. We argue that in many cases it is useful to explicitly separate direct (unmediated) and indirect (through some evaluative process) pathways of change in conditions shaping future interactions.

As the SES framework already expresses feedback paths from an integrated action-interaction-outcome amalgam to each of the four categories of potentially relevant contextual variables, it too could be easily rearranged to fit the obvious temporal sequence of before, during, and after the operation of the dynamic processes occurring within that amalgam. The distinction between direct and indirect (human-mediated) paths of feedback are especially relevant for the types of policy settings to which the SES framework was designed to be most applicable. As emphasized by Epstein et al. (2013), the SES framework as originally conceptualized did not allow for the operation of ecological processes that remain untouched by human agency, even though that possibility can hardly be denied when considering micro-level biological, physical or chemical processes that occur over time spans too short for human intervention, as well as macro-level processes of global change that may be extraordinarily difficult for any analyst to fully observe with a single human lifespan.

Figure 4 may be a deceptively simple representation of our proposed CIS framework, and it’s worth reviewing how we constructed it by rearranging component parts each of these frameworks into a coherent whole that highlights their complementary strengths.

From the IAD we take the realization that contextual factors set the stage for any nexus of strategic interaction among actors seeking to effect or forestall changes in some policy setting, and that the patterns of interactions and outcomes that emerge from those action situations can in turn re-shape the values of these same contextual variables, and thereby open up new opportunities, or impose additional restrictions, on subsequent choices and events.

We depart from the traditional understanding of the IAD framework in three important ways. First, we insist that no action situation stands in complete isolation, and that our scope of analysis must encompass a broader network of adjacent action situations whose outcomes mutually shape the contextual conditions under which other nodes in this network operate. Second, we insist that evaluation can be seen as a distinct mode of collective action, worthy of being treated as an action situation in its own right. Third, by incorporating sites of evaluative interactions into this network representation, we distinguish between direct effects and indirect pathways mediated by individual or collective evaluations and reactions, both of which contribute to the ways in which the outcomes of action situations shape their own future operation as well as the concurrent (and subsequent) operations of other action situations to which it is connected.

From the SES framework we adopt its assertion that processes operating at the level of resource units and resource systems are equally important, increasing the number of categories of relevant contextual conditions to four (from the three in IAD). However, in doing so we redirect our analytic attention away from these variables themselves to focus instead on the dynamic processes through which those conditions are determined, and through which they might be changed by conscious design. Although the multiplicity of relevant action situations is alluded to in later versions of the SES, we prefer an explicitly networked interpretation.

We also simplify some aspects of the SES framework which have not proven to be critical in many applications. For example, Figure 3 indicates that the four categories of contextual factors should also be expected to occur in multiple variants, to depict the simultaneous influences of distinct resource systems or governance systems, or diverse resource units and actors within those systems (McGinnis and Ostrom 2014). We prefer to represent this multiplicity instead in the networks of action situations located at the center of Figure 4. We realize that multiple resources and resource systems are present in all policy settings, but they are typically interconnected in some systematic manner, and those interconnections can be more effectively studied by focusing on the network of interconnected action situations through which those conditions are shaped. We suggest that analysts collect all the relevant contextual factors from different manifestations of systems, units, or actors into comprehensive lists, from which analysts focusing on any action situation can draw out a subset of those factors most directly relevant to its operation, instead of requiring that each action situation has its own associated list.

In addition, Figure 4 does not include any reference to the “related” ecosystems or social-economic-political systems that bracket the core SES box in Figure 3. Nor do we include a separate category for “ecological rules,” as suggested by Epstein et al. (2013). We do, however, acknowledge that there remain some ecological or physical process that lie outside the scope of direct human intervention, either because of the current limits on our scientific knowledge or a lack of instruments through which we might effect change directly into those processes. But these might be better understood as dynamic processes that have not yet been incorporated within a network of action situations in which human agency can be realized.

Thus, further scientific research might move some of these hidden processes within the realm of human intervention. Demographers, economists, and other social scientists have identified the operation of dynamic processes that operate on human society in ways that are not directly observable by individuals, as well as deep and hidden structures of power that shape human society in ways that are not easily discerned by the humans operating within those structures (Lukes 2005). When such hidden processes or structures become more fully understood, then they may prove amenable to change via collective action. Or, in other cases, particularly powerful actors may have the capacity to prevent significant change. Our combined framework should provide a way of dealing with that contingency as well.

Clement (2010), for instance, expands the IAD framework in a natural manner by adding two additional categories of contextual variables to highlight the potential consequences of social-economic foundations for political power and the ways in which public discourse can be shaped to favor the interests of influential actors. Kashwan (2016), meanwhile, illustrates how power asymmetries and differential access to institutional arenas in a network of adjacent action situations can fundamentally shape policy outcomes in the particular context of forestry governance in India. Both of these analyses highlight an important point made most clearly by Knight (1992), that when actors work together to construct any new institutional arrangement, or to revise an existing one, the result is fundamentally shaped by any asymmetries in power among those actors. He argues that all institutions have distributional consequences, and that actors’ expectations of these consequences will shape the details of that institutional arrangement. Knight’s particular model specifies that these details will work in the favor of those actors who have the stronger “outside option,” that is, those who are less directly dependent on the continuation of this relationship. We are proposing a mode of analysis that encourages analysts to identify the action situations in which those outside options are themselves determined. Any outside actors seeking to nudge the direct participants into establishing a more equitable relationship may find it more effective to intervene in those adjacent action situations, rather than insert themselves directly into the negotiation process.

Analysts working within the Bloomington School tradition tend to focus on the problem-solving aspect of policy problems and downplay the extent to which powerful actors can prevent outcomes that would significantly reduce their own power. In our proposed framework, power asymmetries can limit the ability of potential reformers to effectively participate in certain institutional arenas.³ After all, actors who control access to sites of constitutional deliberation need not be willing to listen to new voices, nor to adopt new modes of decision making.

In summary, this CIS framework borrows aspects of both the IAD and SES frameworks, while changing each in significant ways. In the next section we illustrate how this CIS framework can be applied to a specific empirical setting, as an example for researchers interested in adopting a similar mode of analysis in their own work. As we emphasize above, the purpose of a framework is not to explain outcomes, but to provide a tool for identifying, categorizing, and organizing variables and processes for analysis. The CIS framework can help scholars to organize and integrate insights from variable-oriented research (i.e. most applications of the SES framework) and process-oriented research involving networks of action situations, but it does not in and of itself provide tools for inference. In the next section we demonstrate how the CIS framework can be used to organize a body of analysis, and in that illustration we rely upon inferences drawn in well-respected peer-reviewed studies of Maine’s lobster fishery.

5. An illustration of the CIS framework in action

Maine’s lobster fishery has been richly described and analyzed in a series of highly regarded publications, including Acheson (1988, 1997, 2003, 2013), Acheson and Wilson (1996), and Wilson et al. (2013). To help focus the example, we draw particularly upon a specific article in which Acheson (1997) introduces a distinction between different phases of the evolution of the governance of Maine lobster fisheries from around 1940 to the time of that publication.

We recommend that scholars using the CIS framework begin with an inventory of the relevant contextual and explanatory variables emphasized in the relevant research literatures. Many of these variables should be located within one of the four primary categories of the SES (and CIS) framework, but researchers should feel free to draw explicitly upon additional conceptual categories as needed. We emphasize that any inventory of relevant variables is only the first stage of analysis. They must then look for significant changes in the values of key outcomes or contextual variables and posit a set of potential breakpoints that can be used to divide the time span under investigation into shorter periods during which a connected set of processes can be identified for further investigation. These variables should be understood as pointers to specific action situations within the relevant network of focal or adjacent action situations, that is, to those sites of interaction where those changes are most likely to have occurred (and thus where future policy interventions might prove especially efficacious).

Then comes the critical question of deciding which of these adjacent action situations are the most promising for subsequent analysis. These decisions should be guided by lessons gleaned from the relevant research literatures, as well as the analysts’ own previous experience. All of these should be concentrated on the primary goal, namely, to identify the forces driving changes in the core network of inter-connected action situations during each period.

By comparing the values of key variables at different points in time, we can bracket times of significant changes in those and other variables throughout that time period. The values of contextual variables in effect at the beginning of each period denote a partial equilibrium that must somehow have been disrupted in a way that generated changes at the end of that period.⁴ Times when new laws are passed or new organizations established should prove especially useful candidates for potential breakpoints, but exogenous shocks, such as the introduction of a new technology or climate event, may also be particularly relevant in some circumstances. There can be no hard and fast rules about the selection of breakpoints and focal action situations, and any research should fully expect to experience multiple iterations of this search process.

For this exercise Acheson has already done most of the work for us, in his extensive evaluations of the major changes that occurred during this entire period (see especially Acheson 2003 and Acheson and Steneck 1997). We use this procedure to demonstrate how the critical locus of action changes from period to period, after new action situations have been established or existing ones start to move in new directions.

Although Acheson considers several factors in marking breakpoints between different periods, we find it useful for our purpose of demonstrating the virtues of the CIS framework to focus our attention on specific years when the nature of the relevant governance processes changed in a significant manner. Our first break point is 1872, when the state of Maine enacted its first laws limiting the kinds of lobsters that could legally be harvested and sold. The next major breakpoint comes in 1933, when the first double gauge law set both the minimum and maximum sizes of lobsters. In 1977 the national government became directly involved in regulating Maine lobster fisheries, with Congressional passage of the FCMA (Fisheries Conservation and Management Act). Our final break point comes in 1995 with the formal establishment of a zone-based co-management regime involving local fishers as well as national, state, and local officials. Although the conditions of the Maine lobster industry continued to change after this time, the following overview of the periods between these breakpoints should be sufficient to illustrate how the CIS framework can be used to structure an analysis concerning dynamic institutional processes.

By selecting our breakpoints on the basis of significant changes in governance, we are positing that these were indeed consequential events that substantially changed the later evolution of that fishery and the governance system associated with it. If we were to attempt a more complete investigation of this empirical case, we would need to also consider other potential types of breakpoints, and look into the processes through which those changes were produced as well.

In Table 2 we offer an overview of our understanding of the analysis in Acheson (1997), based on our choice of specific breakpoints. The top row lists basic conditions of lobsters as a resource unit and the resource system in which they live, as well as the nature of the primary actors and the rules that govern this resource. Every second row revises these conditions or introduces new ones in order to reflect the changed conditions observed for subsequent breakpoints. Rows placed between these breakpoint rows summarize the most important patterns of interactions and outcomes observed in the focal action situations of particular importance for that period of time.

We use bold font to denote significant changes in the values of key contextual variables in effect at each of the breakpoints. Analysts should pay particular attention to the changing cast of focal action situations at different times. It is here that the most detailed analysis should be focused. We designate (in bold) those action situations which were newly established by the outcome of the breakpoint opening that period, or previously important action situations that now operate in a way much different than in the preceding period. After setting the stage by reviewing the pre-1872 condition of Maine lobster fisheries, we will highlight significant changes in each period in the remainder of this section (and in Figures 5 through 7).

6. Discussion and further extensions

The CIS framework effectively leverages the dynamism of the IAD framework, while incorporating the details of the SES framework to analyze important questions about institutional development and change. The detailed listing of key characteristics of this resource system in different time periods highlights those factors most important in marking the transition from one phase to the next. To implement this framework, institutional analysts should look for significant institutional or ecological changes and identify candidate action situations to which these changes might reasonably be attributed.

Other scholars are exploring modes of research that seem very much in the same spirit of the CIS framework. To begin with works clearly located within the Ostrom tradition, McCord et al. (2017) draw explicitly upon an earlier version of the ideas we present here to organize their investigation of Kenyan water governance. Mincey et al. (2013) use a network of adjacent action situations to examine conditions under which forests can contribute to sustainable urban ecosystems. Lubell et al. (2014) examine the complexity of water management through the application of the “ecology of games” approach, which is closely related in spirit to the idea of a network of action situations, but with a more concentrated focus on the strategic logics of collective choice in inter-related games (Lubell 2013).

Some research teams have devised more explicitly dynamic frameworks for the study of social-ecological systems. Villamayor-Tomas et al. (2015) represent food and energy production processes as “value chains” that effectively consist of sequences of action situations of resource extraction, production, distribution and consumption. In an analysis of Maine lobster fisheries, Webster (2015) introduces an “action cycle” in which a problem generates signals which may or may not trigger a response on the part of the policy makers. We find that representation to be overly simplistic in requiring all policy responses to have either positive or negative effects on the sustainability of that resource, but the explicit sequence of challenge and response is worthy of further elucidation.

Other scholars frame their research methods as offering a means of diagnosing problems in social-ecological systems. In a recent work, Heikkila and Andersson (2018) emphasize the importance and value of clearly specifying action situations of interest and taking a diagnostic approach for contextually-specific institutional design. They make the important point that even though the possibility of sustainable self-governance is more viable than generally realized, it is not always the best solution.

Moving outside the Ostrom tradition, de Loë and Patterson (2018, 567) study institutions of water governance using a diagnostic approach, by which they mean “a structured process of context-specific inquiry into both the biophysical and human aspects of a problem situation. Diagnostic approaches should provide systemic but strategic ways of identifying and evaluating external factors in particular situations.” They advocate the explicit adoption of a “user-oriented perspective (i.e. thinking [as] an analyst, who could be a research, policymaker, or practitioner), and give specific regard to the challenge of suitably capturing relevant external factors.” (568) Despite being outside the Ostrom tradition, this approach to institutional analysis is pertinent to Bloomington School concerns, especially the explicit focus on the critical problem of determining the appropriate scale at which a given policy setting should be understood, and more specifically on the problem of where to draw the boundary between core elements of a social-ecological system and exogenous effects from outside that system. As de Loë and Patterson acknowledge, there are no clear and fast rules for making boundary choices. They recommend that analysts initially should “frame” the problem as “tightly” as possible, and then “critically reflect” on the boundaries they are provisionally considering. Thus begins a process of sequential and repeated consideration of potential extensions by first “looking inward” to more fully understand the internal structure of that action situation and “spiraling outwards” to consider social and environmental factors that are closely tied to those core components. “This spiraling approach is important,” it is asserted, “because it allows for progressively expanding the scope of analysis while continuing to re-visit each key SES variable category in light of previous reasoning.” (571)

These suggestions are remarkably similar to the mode of institutional diagnosis offered by the CIS framework. As Ostrom (2005) argues, the application of the IAD framework requires analysts to “zoom in and out of inter-connected action situations.” Our suggested method of applying the CIS framework asks institutional researchers to identify significant changes in key contextual variables, and to use those observed changes as a symptom pointing towards the underlying processes, that is, to the appropriate action situation responsible for that change. Then they must diagnose the operation of that action situation by tracing its structure back to the sources that determined the operating characteristics (Ostrom’s working components) of that action situation. In short, we ask analysts to trace the relevant network of adjacent action situations in hopes of identifying critical points at which policy intervention might prove most efficacious, based on empirically-informed theory.

We admit this is an eclectic selection of research projects. Some research teams draw on specific aspects of the IAD or SES frameworks, others integrate IAD or SES components with perspectives from quite distinct research traditions, and some have no explicit connection whatsoever to any aspect of the Bloomington School. Overall, however, this variety suggests that we are in good company, since others are pursuing many of these same ideas. We are confident that other researchers will continue to draw upon the rich legacy of the IAD and SES analytical frameworks, whether or not they use the particular combination we offer here.

Although this paper highlights the potential benefits from proper utilization of a particular framework for analysis, we do not mean to downplay the critical importance of other modes of research designed to understand the outcomes of particular variables. One especially important (and frequently used) form of social scientific analysis relies on comparative case studies selected so as to allow researchers to develop and test hypotheses concerning which explanatory (or independent) variables have the strongest influence on determining the value of outcomes (or dependent variables) in those cases. The success or failure of efforts towards collective action is a dependent variable that is particularly important in the types of policy settings for which the IAD and SES frameworks are best suited. Thus, any application of our combined framework is likely to require researchers to draw upon the current state of knowledge generated by the cross-disciplinary study of the factors that facilitate successful efforts at collective action.

In Governing the Commons, the primary outcome variable of interest is the long-term sustainability of institutional arrangements for managing those resources. Before Ostrom (1990) could identify her famous “design principles,” she had to shift her attention away from the value of specific variables to focus instead on the overall configuration of each governance system as a whole.

Other researchers in related projects focus on identifying a reasonably small number of variables that are important determinants of the dependent variable in that analysis. For example, Agrawal (2001) lists some 30 variables as “key enabling conditions” for the successful operation of common property institutions. In the Social-Ecological Systems Meta-Analysis Database (SESMAD) project, resource management and international relations researchers worked together to organize over 100 variables into attributes of the environment, actors, and governance systems (Cox 2014a). Specialized studies then identified the factors most conducive to successful collective action among actors engaged in different policy settings (e.g. Fleischman et al. 2014).

We acknowledge that this mode of focused analysis on particular processes and outcomes remains a critically important tool for social scientists and policy analysts. After all, the CIS framework, as a framework, can only take us so far. It can help researchers to consider a broad range of potentially relevant factors relating to their unique or iterative focal action situations. Researchers then need to draw upon theories and models of particular processes to move beyond the framework stage.

Collective action is of critical importance in all policy settings, but it is rarely (if ever) the only game in play. To fully understand any empirical policy setting, institutional analysts will need to become familiar with the forces driving particular processes of ecological change, resource appropriation, production, transportation and communication, market exchange, organizational behavior, legislative and judicial deliberation, electoral dynamics, public opinion, or the diffusion of knowledge or cultural traditions and other forms of social change. The primary contribution of this CIS framework is the focus it provides on ensuring that institutional analysts begin by identifying the full range of dynamic processes they will need to examine before their task is done.

In sum, the IAD framework is focused on the physical, social, and institutional context within which collective action occurs and outcomes are realized. It presumes that no one action situation exists in isolation, and insists that policies live within a complex ecology of strategic interactions. The SES framework explicitly incorporates the nested, multi-level complexity of ecological systems, and suggests that, to be sustainable, institutional arrangements must somehow match that complexity in a productive manner. As scientific knowledge expands, so does the range of potentially policy-relevant concerns. Also, as public entrepreneurs bring to the political agenda new issues and concerns related to the environment and the sustainability of access to needed resources, then the demands on scientific knowledge expand ever further. In effect, biophysical conditions, scientific knowledge about those conditions, human communities, and the range of public policy are complexly co-evolving. Combining the IAD and SES frameworks enables researchers to better understand that increasing complexity.

Institutional analysts seeking to use the CIS framework should start their analysis by identifying a focal action situation (or at most a few of them), learn how the relevant collective and constitutional choice arenas shaped that action situation, and then think deeply about the processes through which those conditions might be changed. Applying the CIS framework to an empirical setting is all about “zooming in and out of inter-connected action situations” (Ostrom 2005) in search for places where policy interventions could be most effective, or to identify those critical junctures which can contribute the most to deepening our understanding of the relevant processes.