The interactive dynamics of transnational business governance: A challenge for transnational legal theory

Conflict, convergence, cooperation and competition among governance actors and institutions have long fascinated scholars of transnational law, yet transnational legal theorists' accounts of such interactions are for the most part tentative, incomplete and unsystematic. Having elsewhere proposed an overarching conceptual framework for the study of transnational business governance interactions (TBGI), in this article we propose criteria for middle-range theory-building. We argue that a portfolio of theoretical perspectives on transnational governance interactions should account for the multiplicity of interacting entities and scales of interaction; the co-evolution of social agency and structure; the multiple components of regulatory governance; the role of interactions as both influence and outcome; the diverse modes of interaction; the mechanisms and pathways of interaction; and the spatio-temporal dynamics of interaction. To suggest the value of these criteria, we apply them in a preliminary way to selected transnational legal scholarship and to the other articles in this special issue.


Theorising interactive dynamics: An introduction
Conflict, convergence, cooperation and competition among legal systems, institutions, actors and rules have long fascinated scholars of transnational law. Indeed, understanding the dispersion of rule-making authority and managing the resulting overlaps and interactions are central problems of transnational law. 1 These problems have increased as transnational rule systems have proliferated and international law has fragmented. In this article we focus on an important subset of the phenomenon: interactions among actors and institutions engaged in the transnational governance of business conduct, or "transnational business governance interactions" (TBGI). We propose criteria for theorising these interactive dynamics and assess the articles in this special issue and other relevant transnational legal literature against them.
Scholars of transnational law frequently recognise the importance of governance interactions and address them from diverse perspectives. Yet with notable exceptions, their accounts are tentative, incomplete and unsystematic. Philip Jessup's 1956 Storrs Lectures are illustrative. They show that interaction among legal orders has been a preoccupation of transnational law since its beginnings. 2 Jessup was concerned with the dispersion of "authority to make the rules men live by", the transnational proliferation of law-making institutions and the interplay among rule systems. 3 He saw the core problem for transnational law as determining which actors have effective authority to deal with which transnational situations and which rules prevail in cases of conflict. Jessup thus focused on rule conflict, primarily in the context of adjudication; his primary prescription was to authorise adjudicators "to choose from all of these bodies of law the rule considered to be most in conformity with reason and justice for the solution of any particular controversy". 4 Yet Jessup painted an incomplete picture. First, conflict is only one mode of interaction. Even within conflict of laws, cooperation abounds, through treaties, model laws, supranational legislation and mutual adjustment of national laws. 5 Secondly, adjudication is only one component of regulatory governance; interaction also occurs at other stages. Most importantly, Jessup was frustratingly vague about the dynamics of interaction among normative orders. And although he recognised that non-state actors and institutions may promulgate legal rules, he barely explored their interactions with other legal orders. 6 We argue for a more systematic, comprehensive analysis of the interactive dimension of transnational law. Our first step in this direction was to articulate a general analytical framework, 7 summarised at the start of Part 3. Grounded in a regulatory governance perspective, the TBGI framework enables scholars to explore interactions among transnational, state and interstate rule systems; examine the effects of interactions on governance capacity, outputs, outcomes and impacts; and investigate how interactions can be steered to foster desired effects. It provides a common frame within which to identify knowledge gaps, plan research, compare results and generate descriptive typologies.
Yet an analytical framework can take us only so far. In the remainder of Part 3, we consider theory development. Given the complexity of transnational governance interactions, no single theoretical approach can be satisfactory. Thus, we aim to foster theoretical experimentation while encouraging the application of perspectives that are well suited to produce valuable insights in this complex field. For that purpose, we propose seven broad criteria for theory building: a portfolio or toolkit of theoretical perspectives on TBGI should account for the multiplicity of interacting entities and scales of interaction; the co-evolution of social agency and structure; the multiple components of regulatory governance; the role of interactions as both influence and outcome; the diverse modes of interaction; the mechanisms and pathways of interaction; and the spatio-temporal dynamics of interaction. To suggest the value of these criteria, we apply them in a preliminary way to the articles in this special issue and other relevant works. 8

Transnational business governance interactions
Efforts to govern transnational activities have long existed in many domains of economic and social life, including weights and measures, 9 products, 10 accounting, 11 electricity, 12 and religious observance. 13 With globalisation, these efforts have proliferated. Today virtually every issue area, industry or profession that transcends national borders is the object of at least one transnational governance program. These range from straightforward technical coordination standards to the regulation of contentious production and consumption externalities. 14 With this multiplicity come increasingly complex interactions. In some domains, one governance scheme achieves dominance. 15 In others, schemes address different issues (eg measuring greenhouse gas emissions vs the effectiveness of mitigation projects) 16 or regulatory tasks (eg rule-making, enforcement and adjudication). In still others, schemes compete for "regulatory share". 17 They may compete on certain matters (eg rule stringency, transparency or verification) while cooperating or converging on others (eg general principles or design features). Competition and cooperation may lead to harmonisation 18 or to forum-shopping and persistent divergence. 19 Interactions may foster experimentation, efficiency, innovation and learning, or engender duplication, inconsistency, confusion and ennui. In short, such interactive dynamics have important but little-understood implications for the legitimacy and effectiveness of transnational law and governance. They merit systematic study. We begin by clarifying the concept of transnational business governance interactions. In transnational governance arrangements, non-state actors exercise significant authority in the performance of regulatory roles across national borders. 20 Like Gregory Shaffer and others, we focus on the transnational character of the actors and institutions that produce and apply law, 21 not the activities or events that law addresses. 22 The boundaries of the "transnational" are necessarily fuzzy, and the literature abounds with definitions. 23 We emphasise the role of non-state actors and institutions in creating and exercising regulatory authority-a role that remains under-studied in comparison with that of state and interstate organisations. We accept the continuing importance of state and interstate actors, but focus on governance arrangements in which non-state actors play decisive roles, alone or with states or international organisations.
We focus on the governance of business, which generates some of the most important challenges to (and developments in) global governance. 24 Business includes all trade and commerce, whether conducted by private or state actors. Crucially, business is a source, not merely a target, of regulatory governance. While targets always "co-produce" regulation as they implement it, 25 we focus on arrangements in which business (often together with civil society groups and other non-state actors) exercises authority to perform regulatory roles, from agenda-setting and norm development to monitoring and enforcement. Interactions may be equally relevant in domains where business is less central, such as terrorism, war, human rights, disease or education, but we leave these extensions for further research.
Within the expansive concept of governance, we focus particularly on the regulatory variety. 26 We understand regulation as: 20 Eberlein (n 7).   the sustained and focused attempt to alter the behaviour of others according to identified purposes with the intention of producing a broadly identified outcome or outcomes which may involve mechanisms of standard-setting, information-gathering and behaviour-modification. 27 So conceived, regulatory governance encompasses a wide variety of instruments, control mechanisms and actors beyond authoritative legal rules promulgated and enforced by the state. 28 It presupposes no particular institutional arrangements, organisational forms or techniques. It incorporates not only goal-driven attempts at ordering, but also their failures and unintended consequences. 29 Regulatory governance is narrower than social control, as it operates through standards, rules, goals, targets or guidelines. 30 While it encompasses various gradations of coercion, 31 it also includes the exercise of epistemic, persuasive and influential authority. 32 Such authority is important in transnational governance, where regulators "have to promote a motivational response from those whose behaviour it is they seek to change, but often without the infrastructure of the state to fall back on". 33 We do not address the jurisprudential question whether particular forms of normative ordering constitute law. Much of the transnational law literature remains preoccupied with this question. 34 As David Szablowski reminds us, law is a powerful label, reflecting collective aspirations for accountability, legitimacy, and right. 35 As a result, it is a powerful tool for legitimation and critique, and we are interested in how actors deploy the "law/not law" distinction to buttress or contest claims of authority. Interactions are mutual actions and responses of individuals, groups, institutions, structures or systems. They can be intentional or accidental, symmetrical or asymmetrical. Interactions in transnational business governance take many forms (eg competition, cooperation, meta-regulation), exploit different causal pathways (eg modeling, reciprocal adjustment, conditional rule referencing), produce different patterns (eg regulatory ensembles, orchestrations and ecosystems), and have diverse effects (eg strengthening or weakening of democratic accountability; divergence or convergence of norms; ratcheting up or down of standards). Interactions can by analysed at a micro-level (among individual actors or norms), a meso-level (among regulatory schemes or regimes), or a macro-level (among regime complexes or populations).
While some may object that our definition of interaction includes everything (and therefore explains nothing), at this early stage in theory development it is important to maintain a broad definition so as to facilitate comparison across scales and levels, contexts and theoretical perspectives. Refinement of the concept can take place at the stage of operationalisation.
To be sure, transnational (business) governance interactions are just one dimension of transnational law and governance. Our analytical framework and theoretical criteria do not purport to encompass transnational law as a whole and are not intended to displace general theories of transnational law. Rather, they are intended to extend, complement or challenge existing paradigms by insisting on the importance of interactive dynamics and by facilitating systematic theoretical attention to them.
Finally, although we focus here on criteria for explanatory and predictive theories, our analysis is informed by a normative agenda. We wish to understand the conditions under which governance interactions can advance democracy, justice, fairness, dignity, prosperity and environmental sustainability. We share with Boaventura de Sousa Santos, John Braithwaite, Peter Drahos and others a desire to understand and enhance the capacity of structurally disadvantaged interests to participate effectively in transnational governance. 36 How can governance interactions enhance popular sovereignty and ratchet up transnational standards in the interests of ordinary people and the planet? 37 When can interaction be a "weapon of the weak" 38 that empowers marginalised interests?

Theorising transnational business governance interactions
Transnational governance is a dynamic, co-regulatory and co-evolutionary process involving state, non-state and hybrid actors and institutions at multiple levels. 36 Eg Boaventura de Sousa Santos, Towards a New Legal Common Sense: Law, Globalization and Emancipation (Butterworths, 2002); Boaventura de Sousa Santos and Cesar A Rodríguez-Garavito (eds), Law and Globalization from Below: Towards a Cosmopolitan Legality (Cambridge University Press, 2005); Braithwaite and Drahos (n 15) 103. 37 Braithwaite and Drahos (n 15) 34-36. 38 Braithwaite and Drahos (n 15) 626.
These actors and institutions have varied stakes and competencies, perform varied regulatory functions and interact in diverse ways. Our TBGI analytical framework has two axes, one representing dimensions of interaction, the other components of regulatory governance. 39 The resulting matrix enables researchers to decide which aspects of TBGI to analyse for what purposes, while leaving theoretical and methodological choices largely open.
The first axis of the framework identifies six features of interaction: the nature of the interacting entities; the drivers of interaction; the mechanisms and pathways of interaction; the character of interaction; the effects of interaction; and the temporal dynamics of interaction. Each of these features can be studied at different points in regulatory governance processes, which the second axis disaggregates into six components: problem definition and agenda-setting; norm development; implementation; monitoring and information gathering; enforcement; and review and evaluation. Although these components sometimes form a policy cycle, all are not always present, they have no necessary sequence, and they may or may not include feedback loops. These are empirical questions to be determined for each case.
The two axes produce a six-by-six matrix with 36 possible combinations. No study can be expected to address all 36 cells. Most studies will address just one or a few. The matrix prompts researchers to reflect consciously on which components and dimensions are relevant and facilitates comparison of results and identification of knowledge gaps.
The development and application of this framework provide important guidance as we move to theory building on TBGI. Fundamentally, the complexity revealed by the framework makes clear that a portfolio of theories is essential. Diversity of theoretical approaches should be encouraged, but to provide valid insights, each approach must recognise the complexity of the phenomenon even if it seeks to explain only a part of it. In our view, any theory of TBGI should take into account the following features of governance interactions and justify its focus within and among them: 1. The multiplicity of interacting entities and scales of interaction; 2. The co-evolution of agency and structure; 3. The multiple components of regulatory governance; 4. The intermediate position of interaction as both influence and outcome; 5. The diverse modes of interaction; 6. The mechanisms and pathways of interaction; and 7. The development of interaction over time and space.
These criteria are derived from our analytical framework, especially from the six features of interaction described above (interacting entities, drivers, mechanisms/ pathways, character, effects, and temporal dynamics). Previously, we articulated 39 Eberlein (n 7). these features mainly in terms of descriptive taxonomies. 40 Now, we elaborate deeper criteria for theory building. These criteria reflect substantive features of TBGI, not abstract attributes of theory such as verifiability, parsimony or robustness, which are also important. We take up these criteria in turn below, applying them to the special issue articles and to selected elements of the transnational legal literature.

Multiplicity of entities and scales
Any theoretical account of transnational governance interactions must specify who or what interacts. 41 The special issue articles and legal literature identify numerous candidates, from individuals to social systems and from regulatory instruments to discourses. Choosing among these is an exercise in theory in which the researcher posits the nature of the interacting elements and specifies relevant units and levels of analysis. TBGI researchers should do this in a way that recognises the heterogeneity of interacting entities, the often fuzzy and contingent character of their boundaries, the frequent multiplicity of their regulatory roles and the multiscalar character of many of their interactions.
Many interacting entities are actors, such as individuals, governments, civil society organisations and business firms or associations. Individuals can play important roles as boundary-crossing norm entrepreneurs. Organisations are inherently complex, functioning simultaneously as actors and as arenas for interaction by individuals or other organisations. 42 Norms, instruments, and discourses also interact, as do regulatory schemes, regimes, systems and cultures. Their boundaries are even less clearly defined. Entities like these are both products of social action and social structures in which actions are embedded. Entities such as networks, epistemic communities and social movements exhibit characteristics of both actors and structures. No theory can encompass all actors, norms and social structures, but theories should recognise the heterogeneity of interacting entities and their roles.
Some illustrations from the legal literature show that there are many ways for theorists to accomplish this. Transnational legal process theory emphasises interaction among norm entrepreneurs to articulate or interpret norms that guide future interactions, ultimately reconstituting actors' interests and identities. 43 Gregory Shaffer 40 Eberlein (n 7). 41 Eberlein (n 7) 8. and Terence Halliday's transnational legal orders project examines interactions within the normative structure of transnational law (including legitimacy, clarity and coherence), within national legal systems and between the two. 44 Liberal theories posit multi-level interactions among state, interstate, sub-state and non-state actors, 45 while rational-institutionalist theories of legalisation combine interaction among actors and norms to explain the transnational spread of distinctively legal features. 46 Public international lawyers address interactions through concepts such as fragmentation, 47 "hard/soft" law interaction, 48 and "trade and . . ." conflicts. 49 Regime complexity theory examines actor strategies within regime complexes. 50 These approaches remain primarily focused on state-based governance.
Other approaches emphasise non-state actors and institutions, including interactions between interstate and non-state law. 51  Duncan Snidal locate interactions among diverse actors and institutions on a "governance triangle" formed by the state, business and civil society. 52 Abbott also applies regime complexity theory to non-state governance. 53 Theories of transnational private regulation capture interactions within deterritorialised regulatory regimes, 54 but typically overlook inter-regime interactions, with notable exceptions including accounts of transnational meta-regulation. 55 Transgovernmental approaches explore cross-border interactions among courts, legislatures, regulatory agencies, subnational governments and international secretariats, often via networks. 56 Others examine the crucial role of intermediaries such as lawyers. 57 Comparative law has developed potentially fruitful analytical frameworks that combine the interaction of norm entrepreneurs, communities, legal structures and technological systems to explain legal transplantation and norm diffusion. 58 Contemporary theories of regulation and "new governance" examine bargaining and intermediation among individuals and organisations that span the publicprivate divide. 59 60 Theories of international regulatory competition and co-opetition emphasise interactions between national legal systems 61 but bring non-state actors in to a certain degree. 62 In a similar vein, theories of multi-level governance involve horizontal, vertical and "diagonal" interactions among actors and jurisdictions at different scales. 63 Robert Ahdieh's concept of dialectical regulation focuses on inter-systemic interactions, 64 while experimentalist governance involves recursive interactions among central and decentralised units of various kinds. 65 Some transnational legal pluralists examine interactions among legal orders or cultures, 66 others among structural sites and strategic actors. 67 In Santos's conception of interlegality, hegemonic and counter-hegemonic legalities interact in  collisions among "rival normative ideas, knowledges, power forms, symbolic universes and agencies". 68 Oren Perez studies interactions among state-based, nonstate and hybrid organisations and regimes within a "complex discursive labyrinth". 69 Gralf-Peter Calliess and Peer Zumbansen's theory of "rough consensus and running code" involves interactions between actors and social structures, including norms, regimes, cultures and spaces. 70 Finally, at the structural end of the agency-structure continuum, Gunther Teubner's systems theory involves interaction among autopoietically closed social subsystems, including formal organisations and functionally differentiated societal subsystems. 71 Halliday and Shaffer contend that the boundaries between interacting legal orders are fixed by discourses, ideological frames, subjective perceptions, alliances and conflicts. 72 These elements are contestable and hard to measure, however. The fuzzy, permeable and contingent character of interacting entities can raise difficult issues of individuation, since "interaction is often more like that between waves or clouds or rivulets than between hard, stable entities like rocks or billiard balls". 73 Similar considerations lead Fleur Johns to urge scholars to focus on the "hyphen" between intersecting legal orders: the shifting spaces in which legal systems are constituted mutually by their encounters with one another. 74 Another important feature of interactions is the multiplicity of regulatory roles that actors often perform simultaneously. A business firm, for example, might implement standards and participate in standard-setting; a transnational governance scheme might set standards and also be subject to meta-rules. In short, transnational business governance involves "many kinds of actors which regulate while being regulated themselves". 75 68 Santos, New Legal Common Sense (n 36) 472.  Finally, theoretical accounts of TBGI should recognise that interaction frequently occurs at and across multiple scales. Micro-level analyses examine how individual actors (people or organisations) interact within a regulatory scheme or jurisdiction to create, implement or enforce standards. Much of the theoretical literature on regulation, 76 especially transnational private regulation, 77 follows this approach, examining interactions among rule makers, targets, interpreters, enforcers or beneficiaries. Meso-level analyses examine how regulatory jurisdictions or schemes interact, in regimes or organisational fields, and how norms, discourses, cultures and other institutions enable and constrain action. 78 Finally, macrolevel analyses explore how entire regimes or organisational fields intersect within larger complexes 79 or systems, 80 and how the latter entities interact with one another.
We argue elsewhere that meso-level analyses are initially the most fruitful, because they provide "sufficient abstraction to identify patterns and trends, without sacrificing empirical detail". 81 Yet we do not advocate an exclusive meso-level focus. Interactions often link different levels; they may occur simultaneously within a regulatory scheme (micro) and with other schemes (meso). Moreover, interactions frequently cut across geographic scales, from local to 76 Eg Robert  global. A fruitful theoretical portfolio should encompass the micro and meso levels, the meso and macro levels, or all three. Table 1 provisionally (and contestably) locates a range of theoretical approaches on the continuum of scale. Strikingly, most approaches either squarely address the meso level or straddle that level and one or both of the others. This accords with our own recommendation and provides the basis for a fruitful theoretical portfolio.
Turning to the articles in this special issue, John Biggins and Colin Scott do not theorise interacting entities explicitly, but their account of TBGI in the field of financial derivatives is squarely actor-centred. Biggins and Scott focus on interactions among the key organisations involved in over-the-counter (OTC) derivatives regulation before and after the global financial crisis, including the International Swaps and Derivatives Association (ISDA), financial regulators, courts and financial firms. They examine simultaneous micro-level interactions within ISDA and meso-level interactions between ISDA and other organisations. These interactions cut across geographic scales, between the global ISDA Master Agreement and national laws. Their case study invites further theorisation of these interactions.
Karin Buhmann's study of the UN Guiding Principles on Business and Human Rights identifies a wide range of interacting entities, including organisations (eg UN Human Rights Council), societal segments (eg civil society), individuals (eg John Ruggie, the Special Representative of the UN Secretary General on business and human rights ("SRSG")), and regulatory schemes (eg UN Guiding Principles, UN Global Compact, OECD Guidelines, and national laws). Buhmann's focus on relationships among regulatory instruments is particularly intriguing.
It is not always clear in this analysis who or what is interacting. In our terms, however, Buhmann offers a micro-level analysis of interactions at the rule formation stage, focusing on the SRSG's enrollment of private and public actors. Later, she presents a meso-level analysis of interactions between the UN Guiding Principles and other programs at the implementation stage, with a focus on mutual "piggy-backing" by these programs to enhance their legitimacy and effectiveness. Her account shows that scales of interaction and analysis can shift as activity moves from one component of the regulatory process to another. It also invites theoretical attention to the relations among actors, instruments, institutions and discourses.
Kernaghan Webb also identifies a variety of interacting entities in his study of the International Organisation for Standardisation's ISO 26000 social responsibility guidance standard. These include organisations (eg ISO and other instrument developers), individuals (eg Working Group experts), regulatory instruments (eg ISO 9000 and 14001; ILO Core Conventions), norms (eg due diligence and Plan-Do-Check-Act), issue areas (eg labour, human rights and environment) and societal sectors (public, private and civil society). While Webb paints a rich picture, it is again not always clear which of these are interacting at any specific point. Webb's account shows that interactions can occur and be analysed simultaneously at the micro, meso and macro levels. It also indicates, like Buhmann's, that the nature and scale of interaction can shift across regulatory governance components. At the rule formation stage, Webb traces a combination of micro-level interactions within ISO and meso-level interactions between ISO and competitor organisations. At the implementation stage, he describes meso-level interactions among ISO, states and business actors to solidify ISO 26000′s status as a global custom, and macro-level interactions between this custom and domestic legal systems.
Webb also shows ISO to be a multi-level, multi-scalar arena for interaction. First, ISO is a meta-organisation of national standards bodies and international organisations. As a result, it is a powerful convener; interactions between ISO and other standard-setters occur both outside and within ISO, blurring the line between micro and meso. Secondly, ISO standards are drafted by individual experts acting in a personal capacity, while standards are approved by national standards organisations. Thirdly, ISO is the pinnacle of a deep and broad standardisation infrastructure spanning subnational, national and transnational spheres. Fourthly, in developing ISO 26000, ISO organised experts into stakeholder categories, encouraging transnational collaboration within and between them. The process thus encompassed diverse cross-actor and cross-scale interactions. It would be fascinating to theorise these features, identify other organisations where they appear and investigate their role in TBGI.
David Doorey is alone among the authors in this special issue in devoting substantial attention to theory development. He identifies two categories of interacting entities in the development of a living wage norm in transnational codes of conduct for supply chains: organisations and social subsystems. He first examines interactions among diverse NGOs, labour unions, firms, trade associations, governments, and standards schemes within a TBG social subsystem. Doorey emphasises heterogeneity within as well as between categories and even individual actors. He also examines interactions between the TBG subsystem and other subsystems in its environment. Doorey develops a model of TBGI grounded in the theory of open systems. His model links the micro-level of interaction (among individuals, NGOs and firms) to the meso-level (among transnational regulatory schemes) and the macro-level (between societal subsystems). Here again, the scale of interaction shifts over time: from micro-level, conflictual interactions between NGOs and business; to meso-level competition between TBG schemes; and to multi-level interactions that facilitated the spread of the living wage norm from NGO-led to business-led schemes. Macro-level interactions among social subsystems conditioned all these interactions.
Doorey's individuation of the TBG subsystem raises some questions. Treating it as a single, functionally autonomous system might obscure both its in-between-ness 82 and its internal heterogeneity. 83 This limitation notwithstanding, Doorey's model provides one possible answer to the theoretical challenge of capturing the multi-scalar and multi-level character of governance interactions.
3.2 Co-evolution of agency and structure Theoretical frameworks that focus only on structural features such as institutions, discourses, cultures, technological systems or social systems without attending to actors and their interests and resources, or that analyse strategic action in isolation from the social structures in which it occurs, will fail to grasp all the dynamics of governance interactions. Structural elements help constitute actors' identities and possibilities for action, and are in turn constituted, reproduced and transformed by actors' actions. Whether a theoretical perspective emphasises agency or structure, therefore, it should recognise this mutual constitution and co-evolution.
Theories that emphasise reflexivity, co-evolution and mutual interaction between agency and structure will have particularly strong purchase. This includes most institutionalist, experimentalist, socio-legal, pluralist and legal-process approaches described above. 84 It also includes constructivist theories of international law, which assume the mutual construction of social structures and actors' identities and choices. 85 In Jutta Brunnée and Stephen Toope's interactional theory, for example, normative structures are constructed, reproduced and transformed via interaction between intersubjective norm-creation and actors' ongoing practices of legality. 86 Other approaches to the agency-structure challenge can also be valuable. Consider the two theoretical perspectives in the bottom row of Table 1; both engage micro-, meso-and macro-level interactions simultaneously, yet offer contrasting perspectives on agency-structure interaction.
Braithwaite and Drahos seek to explain how micro-level actions "constitute structural change, just as those micro processes are constituted and constrained" by macro structures. 87 They theorise micro-macro sequences of regulatory globalisation in which entrepreneurs promote a regulatory innovation and enrol 82 See Johns (n 74). organisational power via webs of dialogue. The innovation spreads through modeling, eventually becoming a standard. 88 Braithwaite and Drahos also theorise a forum-shifting sequence. 89 In our terms, this is a meso-level phenomenon in which an actor moves a regulatory agenda from one forum to another, abandons a forum, pursues an agenda in multiple fora or blocks a forum from acting. 90 Forum-shifting is used by a range of state and non-state actors. 91 Business groups' creation of their own forestry certification schemes to divert the forest certification agenda away from the multi-stakeholder Forest Stewardship Council (FSC) is an example. 92 More recently, the FSC and its main industry-led competitor failed in a joint bid to block ISO from taking up the sustainable forestry standardisation agenda, 93 while consumer and business interests succeeded in making ISO an alternative to the ILO as a forum for developing global labour-related standards. 94 A closely related strategy of institutional bypass has been used by nonstate actors to get around intergovernmental fora. 95 Forum-shifting and institutional bypass are promising strategies by which micro-or meso-level interaction can produce macro-level change. 88 Braithwaite and Drahos (n 15) 33, 551, 559-562. 89 Braithwaite and Drahos (n 15) 564, 569-571. 90 Braithwaite and Drahos (n 15) 29, 564. 91 Eg Laurence R Helfer, "Forum Shopping for Human Rights" (1999) 148 University of Pennsylvania Law Review 285. 92 The main industry-led umbrella scheme, the Programme for the Endorsement of Forest Certification (PEFC), boasts a much larger total certified area despite having been created six years after the FSC and despite strong endorsement of the FSC by many civil society groups and academics (though the FSC's total certified area is growing at a quicker rate). See generally Benjamin Cashore, Graeme Auld and Deanna Newsom, Governing Through Markets: Forest Certification and the Emergence of Non-State Authority While Braithwaite and Drahos recognise the co-evolution of agency and structure, their account is fundamentally actor-centred. 96 They insist that "explanation is not indissolubly linked to social systems", and they "take seriously the intentionality of agents and the ways in which their beliefs and desires can lead them to change their regulatory worlds". 97 This has the advantage of clearly identifying the drivers of regulatory interaction (goal-seeking agency). Yet there are also other potentially fruitful avenues for investigation, including interactions among norms, discourses or social systems; logics of contradiction, juxtaposition and contingency; and uncalculated interactions such as institutional isomorphism. 98 Teubner's theory of autopoietically closed systems offers a contrasting perspective on the agency-structure relationship. 99 Although this theory might appear to offer a purely macro-level analysis, it actually spans all three levels. Interaction takes place between societal subsystems (macro) and between organisations (micro and meso). Every attempt to regulate an organisation involves inter-systemic interaction; all inter-systemic interactions must pass through the "eye of the needle" of individual communicative events. 100 Teubner's interactive mechanisms of blind co-evolution, information and interference 101 operate simultaneously at all three levels. But autonomous human agents take a back seat. Social systems are constituted by communications, not by human beings. 102 The theory of autopoietic systems thus offers a radically limited account of the co-evolution of agency and structure.
Doorey's use of open systems theory offers an alternative. It combines macrolevel interactions among social subsystems with micro-level interactions among individuals and organisations. Within the TBG subsystem, actors interact via varied pathways and mechanisms to produce regulatory rules and behavioural change. Internal feedback loops link these outputs back to actors, while external feedback loops link them to economic, legal and other subsystems, which in turn provide inputs to the TBG subsystem. 103 The co-evolution of agency and structure is central to Doorey's model. Alignment or conflict among actors' goals, values and risk perceptions is a powerful driver of interaction, but actors' characteristics-and the choices available to them-evolve in response to 96 Braithwaite and Drahos (n 15) 9, 16, 528-530. 97 Braithwaite and Drahos (n 15) 17.  104 Calliess and Zumbansen's Rough Consensus and Running Code (RCRC) offers another systems theory-inspired perspective. On one hand, the fragmentation of society into autonomous, functionally differentiated, law-creating subsystems with competing rationalities creates the structures that shape actors' identities and possibilities for action (macro). 105 On the other hand, actors' interactions generate the rough consensus that congeals into a subsystem's running code and drives its continual evolution (micro). 106 The RCRC model provides a promising theoretical frame for analysing interactions among actors, norms and social structures.
Turning to other articles in this special issue, Biggins and Scott hint at a subtle relationship between the ISDA and market-wide institutional structures. For example, the Master Agreement and close-out netting procedures devised by ISDA were adopted so widely that they became part of the institutional structure of the derivatives market. As such, they helped moderate the global financial crisis in that market, leading powerful states to enhance ISDA's agency further.
Buhmann hints at another agency-structure relationship. By combining an actor-centred negotiation history with discourse analysis, she illustrates the mutual construction of small-scale agency and large-scale discursive structures, including the strategic deployment of discourse to influence action. 107 Theorising these agency/structure interactions would be a welcome next step.

The components of regulatory governance
Our analytical framework identifies six components of regulatory governance: agenda-setting, norm formation, implementation, monitoring, enforcement and review. 108 Whatever theoretical frameworks they apply, researchers should recognise that regulatory governance has such components, and that interactions can occur within any of them. This divisibility also invites theory development. Each component "demands a different portfolio of resources or capacities, including financial resources, organisational capacity, expertise, legitimacy, and strategic position". 109 Each can, moreover, involve different actors and display distinct interactive dynamics, as Buhmann and Webb show. 110 104 Doorey, ibid, 464. 105 Calliess and Zumbansen (n 34) 44-50. 106 Calliess and Zumbansen (n 34) 135-136. We nonetheless list this theory under "Micro and meso combined" in Table 1 because the authors focus almost entirely on these levels. 107 Buhmann, this issue, 414, 422, 424-5. 108 Eberlein (n 7) 6. The regulatory governance process can also be disaggregated in other ways. 109 Eberlein (n 7) 7. 110 See above, Part 2.1.
Much of the legal literature examines only one or two components, most commonly norm formation 111 and enforcement (adjudication). 112 Margaret Young, Robert Howse and Ruti Teitel are among the scholars calling for attention to a broader range of regulatory processes. 113 Analysing interactive dynamics within and among different components will be a valuable area of theory building.
The articles in this issue illustrate the opportunities. Scott has elsewhere emphasised the importance of looking beyond standard-setting in transnational private regulation, 114 and he and Biggins do so in their article. They trace how interactions expanded from agenda-setting and norm formation to implementation and enforcement. Interactions in agenda-setting and rule formation included micro-level bargaining among dealer banks to produce the ISDA Master Agreement and mesolevel interaction in which ISDA drafted and governments enacted OTC derivatives-friendly legislation. Here the outputs of micro-level interactions (boilerplate contracts) were inputs to meso-level interactions, reinforcing ISDA's lobbying for self-regulation. Once the ISDA self-regulatory regime achieved global dominance, governments implicitly endowed it with additional interpretative and adjudicative responsibilities, exercised mainly by newly formed credit derivatives determination committees (DCs). DCs have unusual significance because of their impact on the financial positions of states and their creditors.
Buhmann discusses all six components of regulatory governance, showing how the nature and scale of interaction change as activities shift from one component to another. During agenda-setting, micro-level interactions within the UN human rights apparatus evolved from competition to coordination. During rule formation, micro and meso interactions blended together, shifting to cooptation and meta-regulation as the UN Guiding Principles achieved dominance. During implementation, the focus shifted again, to meso-level mutual "piggybacking" by the UN Guiding Principles and other programs.
Doorey's article illustrates how conflict during rule formation (over defining a living wage) can be diverted into cooperation at the implementation stage (where experimentation with implementation strategies led to agreement on a "wage ladder" approach). It also indicates, however, that the result may be merely to defer rather than to resolve conflict.
Webb's study invites theorisation on two points. First, the same interactions may affect multiple components of regulation simultaneously. For example, interactions over implementation of ISO 26000 occur partly within state and interstate norm development processes. One scheme's implementation may thus be another's norm formation, involving similar actors and interactions. Secondly, the article 111 Eg Calliess and Zumbansen (n 34). suggests that organisations' pre-existing legitimacy endowments influence interactions across multiple components. ISO's legitimacy as a global standards-setter facilitated its assertion of authority over social responsibility (SR) standards at the agenda-setting stage. During norm development, it allowed ISO to steer competition with rival SR standard-setters into co-opetition within ISO. During implementation, the widespread acceptance of the ISO management systems standards on which ISO 26000 was modelled facilitated business uptake of the new standard.
In sum, scholars of TBGI should turn their attention to theorising interactions within and across the components of regulatory governance. Important questions for theory development include whether particular components have peculiar interaction dynamics; whether those dynamics change when interactions relate simultaneously to multiple components; how interactions in one component influence those in another; how linkages between components can be exploited to manage regulatory interactions; where feedback loops are likely to emerge; and how organisational legitimacy endowments influence interactions throughout the regulatory governance process.

The intermediate position of interaction
Theories of transnational governance interaction must recognise that interactions occupy a dual position. On one hand, they are the results of conditions, forces and actions that influence who or what will interact, when and how. On the other hand, interactions affect the design of regulatory institutions, the nature and content of regulation, the perceptions and behaviour of regulated actors, and other downstream conditions.
A range of factors drives interaction. 115 Actor-level drivers include alignment or conflict of goals, perceptions and interests, and overlapping scope of activity. 116 System-level drivers include functional differentiation of society into subsystems with different logics; 117 duplication or inconsistency of rules; 118 and governance gaps created by globalisation and the transformation of the state. 119 Other relevant factors include problem structure, 120 industry structure, 121 technical systems, 122 discursive structures and cultures. 115 Eberlein (n 7) 9-10. 116 Eg Doorey, this issue, 441-43. 117 Eg Gunther Teubner, "'Global Bukowina': Legal Pluralism in the World Society", in Gunter Teubner (ed), Global Law Without a State (Dartmouth, 1997) 3; Schepel (n 54). 118 Raustiala, "Institutional Proliferation" (n 50) 309, argues that fragmentation and conflict, not proliferation and density per se, drive interaction. 119 Eg Calliess and Zumbansen (n 34). 120 Eg Braithwaite and Drahos (n 15) 58; Abbott and Snidal, "Law, Legalization and Politics" (n 46) 50. 121 Eg Cashore, Auld and Newsom (n 92). 122  Turning to effects, the ultimate goal of our research agenda is to explain and predict the impacts of governance interactions on social, political and environmental conditions. 123 However, such impacts are very difficult to determine, given the complexity of causal and other relations and the multiplicity of contributing factors beyond regulation (let alone regulatory interactions). These points have important implications for TBGI theory building. We focus here on two: choosing the object of inquiry and integrating drivers and effects.
First, if it is not practicable to assess directly the social impacts of governance interactions, what is the most fruitful proxy? The most common alternatives are outputs and outcomes. 124 Outputs of interaction include regulatory standards and institutional design features. Outcomes include changes in behaviour, values or perceptions. Ideally, scholars should seek to theorise how interactions affect the outputs and outcomes not only of single governance schemes but of entire governance regimes.
In studying outputs, theories of institutional isomorphism, 125 norm diffusion 126 and regulatory races 127 are relevant. Of particular interest are theories explaining how races become ratchets. Braithwaite and Drahos argue that certain principles can combine to prevent reversal of a regulatory race's direction, transforming a race into a ratchet. 128 They theorise the conditions for this to occur and give numerous illustrations. Multiple ratchets are particularly powerful. In forestry, for example, principles of rule compliance and continuous improvement in national law, ISO 14001 and the FSC create multiple ratchets, each driving regulatory standards up and all of them together preventing backsliding. 129 In addition to behavioural change, outcomes can include dialogue, information sharing, shared understandings, policy learning, and changed values, perceptions or expectations. They may also include heightened hostility or mistrust. As Doorey notes, such outcomes can emerge even without formal outputs such as rules. 130 We propose focusing on one outcome in particular: changes in the capacity of regulatory governance actors, schemes and systems. 131 Scholars should investigate how interactions affect the mobilisation of motivations, competences and resources along governance chains, and how capacities affect the performance of regulatory systems-in the narrow sense of making rules and shaping practices. The motivations, competences and resources that must be enrolled are controlled by a variety of actors, who may be unwilling or unable to contribute, leaving the regulatory system without essential capacities and jeopardising its performance.
Do interacting schemes reinforce one another and promote diverse solutions 132 through innovation, experimentation, learning and adaptability? 133 Or do they work at cross-purposes, facilitating forum-shopping for the least stringent rules? 134 What conditions-eg, shared interests, goals and cognitive understandings-are associated with successful mobilisation and orchestration of actors and resources? Bruno Latour's concept of enrolment (the interactive process of mobilising other actors and their resources in support of an actor's regulatory goals) should be fruitful here. 135 Yet questions remain, including which indicators of capacity to adopt, how to isolate the effects of interaction from other factors, and how to assess the independent effect of capacity constraints on regulatory performance. There is also a need to theorise the effects of interaction on governance capacity in deeper senses, including democracy, legitimacy, openness, transparency and accountability. 136 A second challenge for theory is to integrate a backward-looking focus on the drivers of interaction with a forward-looking analysis of its effects. The phenomena they capture are closely intertwined in the regulatory process. Problem structure, for example, is both a background condition that shapes interaction and at least partly the product of discursive interaction. 137 This challenge can be addressed in many ways, including applying a single theoretical frame to drivers and effects, and drawing on multiple theoretical tools to wrestle with drivers, effects or both.
In this special issue, Biggins and Scott offer a functional account of the drivers of governance interactions in derivatives markets. Increasingly complex financial transactions and the liberalisation of state regulation created a governance gap, prompting demand from market actors for coordination and sporadic pressure from states for industry self-regulation. The resulting interactions between state and industry actors had three main effects: the emergence of a hybrid public/ private governance regime; ISDA's consolidation as the dominant non-state regulator; and further enhancement of ISDA's regulatory role following the global financial crisis.
Drawing on Tim Büthe's study of the International Electro-technical Commission, 138 the authors explain ISDA's dominance as a result of state support and ISDA's neutralisation of competitors, both explicitly interactive processes. They explain ISDA's remarkable resilience as a result of industry structure. They endorse Daniel Mügge's theory that industries dominated by highly organised producers can, despite exogenous shocks, define regulatory agendas and resist state intervention in complex areas by offering to tighten self-regulation in exchange for public oversight. 139 Governance interactions in the derivatives regime enhanced the regulatory capacities of both ISDA and states. ISDA was able to enroll states' legislative and enforcement capacities in support of its regulatory agenda. State endorsement conferred legitimacy on ISDA and facilitated its dominance. For their part, states enrolled ISDA's rulemaking and communication capacity to compensate for their limited expertise and legitimacy. The results are unstable, however. ISDA must walk a political tightrope between maximising regulatory share and angering state policy makers, especially in systemically important areas like sovereign debt.
Buhmann also offers a functionalist account of drivers, but points additionally to actors' interests, values and perceptions. These can be manipulated, as with the SRSG's discursive strategy of articulating stakeholders' interests in cooperation, which amplified their drivers to interact. Buhmann also cites differences in geopolitics and in institutional locus and mandates to explain why interaction was more collaborative under the SRSG than previously.
The effects of interaction at the agenda-setting stage were a shared understanding of the need for a transnational governance instrument on business and human rights; at the norm-formation stage, the result was the UN Guiding Principles themselves; and at the implementation stage, the effects were normative homogenisation and mutually enhanced regulatory capacity. In the face of member states' reluctance to regulate business directly, the regulatory capacity of UN human rights organs was enhanced by enrolling non-state actors in rulemaking and by piggy-backing on other TBG schemes to complement the Principles' limited implementation and enforcement modalities. The piggy-backing also worked in the other direction: the UN Human Rights Council's endorsement of the Principles allowed other TBG initiatives that incorporated the Principles to expand their membership, audiences and regulatory space.
In contrast to these functionalist accounts, the drivers of interaction in Doorey's open systems model are actors' goals, values, risk perceptions, power and capacity, which evolve in response to internal and external forces. Doorey's main focus, however, is on the effects of interaction: the outputs and outcomes of the TBG subsystem. The main output was the transformation of the living wage norm from "an outlier . . . to a norm found in most leading TBG schemes targeting global supply chain labour practices". 140  agreement on rules (outputs) is sometimes possible only because they require little behavioural change (outcomes). Finally, Doorey emphasises the difficulty of measuring effects other than regulatory outputs, of isolating interactions from other causal variables, and of predicting whether interaction will have salutary or deleterious effects.
Webb's main contribution here is his examination, discussed above, of the effect of legitimacy endowments on the character and trajectory of interactions. This highlights the intertwined nature of drivers and effects. ISO's legitimacy endowment enabled it to attract unprecedented participation by intergovernmental organs, which in turn conferred legitimacy on ISO 26000. Webb also emphasises the importance of institutional design. He argues that ISO's publicly accessible, transparent, multi-stakeholder consensus process mitigated dominance by any one actor category, enhanced ISO 26000′s legitimacy, ensured that it genuinely approximated a global societal consensus, and facilitated widespread acceptance.

Modes of interaction
While some legal scholars, like Jessup, focus on a narrow range of interactions such as conflict, harmonisation, or cooperation, 141 many identify greater variety. 142 Theoretical approaches should embrace this diversity, treating the character of interaction as an empirical question, not an assumption. While descriptive typologies have value, 143 they do not capture all the observed complexity. The challenges for theory include explaining why interaction takes particular forms, how certain forms overlap or transform into others, and how forms vary across scales or levels of interaction. In the latter connection, theories of meso-level, inter-scheme interactions such as orchestration 144 and meta-regulation 145 are likely to be particularly fruitful.
In this special issue, Biggins and Scott show the complexity of cooperative interactions. Interactions between ISDA and state actors included ISDA's enrolment of state regulatory capacity; its oversight of governments (through model laws, lobbying and strategic litigation); state recognition of ISDA self-regulation; implicit state delegation of authority to ISDA DCs; and state cooptation of potentially subversive non-state regulation through selective incorporation. Here the drivers of conflict and cooperation produced a delicate, continually renegotiated balance between state and non-state regulation.
Buhmann demonstrates how the character of interaction can change over time, across institutional settings, and among components of regulatory governance. Interactions shifted from conflict to cooperation at the agenda-setting stage and to cooptation and meta-regulation at the rule formation stage; cooptation largely continued during implementation. The earlier-described mutual enrolment and piggy-backing by the UN Guiding Principles and other TBG schemes suggests a form of symbiosis that may not be captured by conventional concepts of cooperation. These increasingly integrative interactions produced norm convergence, consolidation of the Guiding Principles' dominance and the resurgence of state authority in this field.
Doorey emphasises the simultaneity of competition and cooperation. From the time NGOs formed the Workers' Rights Consortium as an alternative to industry codes of conduct, inter-scheme competition combined with dialogue and cooperation. Co-opetition continued even after the leading schemes defined common standards for pilot testing. The six participating schemes competed to have their standards adopted, yet also agreed to favour the highest standards within the six schemes-including a living wage clause.
Co-opetition also plays a key role in Webb's study. Co-opetition occurs when "rule instrument developers that are in other respects in competition with each other, may for strategic reasons recognise an opportunity for collaboration for their mutual benefit". 146 Webb investigates the drivers and effects of this mode of interaction. Drivers include competitors' desires to align their rule instruments, advance their own agendas through competitors' schemes, and enjoy the legitimacy-enhancing benefit of association with other schemes. In this case, co-opetition led to endorsement of ISO 26000 by intergovernmental organisations, governments, TBG schemes, transnational corporations and NGOs. It also produced regulatory synergies, with competitor schemes articulating general principles and ISO 26000 providing detailed standards and guidance.
The processes of normative convergence and diffusion described by Buhmann, Doorey and Webb deserve greater attention. Which norms spread to which governance schemes, and why? If norms do not spread, why not? Is co-opetition a special case, or a broader phenomenon? Is it more likely to strengthen or weaken regulatory standards, and to enhance or degrade regulatory capacity?

Mechanisms and pathways of interaction
Specifying and theorising the mechanisms and pathways that link drivers, interactions and effects holds great promise for middle-level theory building. An understanding of mechanisms and pathways offers substantial purchase on interactive dynamics. Such understanding also helps identify the opportunities for structurally weak parties to remake their regulatory worlds. 147 At the same time, Abbott and Snidal warn that mechanisms are "challenging analytical tools": They are often difficult to identify and analyze in practice. They require the development of "stylized facts", so that detailed descriptions do not obscure analytical insights. They may be complex . . . multiple mechanisms may be at work. 148 reciprocal adjustment. 157 There is no master mechanism, however; the globalisation of regulation always involves multiple mechanisms. 158 A question for theory is to clarify the relationships among modeling, mimicry and isomorphism. Many scholars equate modeling and mimicry, but Braithwaite and Drahos distinguish them. For them, modeling entails observational learning, in which actors display, symbolically interpret and emulate regulatory models; imitation merely involves one actor matching another's actions, evidently without reflection. 159 Imitation, then, is restricted to unconscious isomorphism, where isomorphic pressures emerge from systems, cultures or mentalities. 160 Braithwaite and Drahos offer a detailed theory of modeling in which model missionaries promote a particular model because they believe in it; model mercenaries exploit the markets model missionaries open up; model mongers, lacking the resources to develop their own models, experimentally float numerous models in hopes of finding one that catches on; model misers adopt pre-packaged models (often floated by model mongers) to economise on the costs of model development; and model modernisers in the periphery adopt models from the centre to harness the legitimating power of a modern, progressive identity. 161 Modeling is important because of bounded rationality: decision-makers are unable to address fully the limitless range of issues they face, so it often makes sense to adopt pre-packaged solutions that are good enough, rather than search for optimal solutions. 162 Decision-makers frequently "dither in a confusion of complexity they cannot grasp, which is why they can be led by entrepreneurs who encourage them onto a plausibly interest-enhancing path". 163 Modeling is also important because it is one of the few mechanisms available to the weak. Structurally weak model mongers can devote their scarce resources to floating regulatory models until they land on one that catches a powerful adversary off balance; then they can pour all their resources "into a feat of political ju-jitsu that flips the off-balance adversary". 164 Yet Braithwaite and Drahos's conclusions about model mongering are not entirely persuasive. Most of their case studies suggest that the key actors are either model missionaries, who promote preferred models out of belief, or model mercenaries, who profit by promoting a particular model, rather than experimentalist model mongers.
Other significant mechanisms in TBGI include conditional rule referencing, 165 overlapping memberships and contracting. 166 Benjamin Cashore, Michael Stone and Lars Gulbrandsen show that rule referencing in forest certification creates interdependencies that enhance the capacities of state and non-state regulators: non-state schemes benefit from states' enforcement capacity, states gain nonstate schemes' norm-generating capacity and each gains symbolic resources from the other. 167 Theories of transnational governance interaction should also acknowledge the multiplicity and interlinking of mechanisms, in line with Braithwaite and Drahos's observations. 168 More empirical and theoretical work on mechanisms is also needed. Even Braithwaite and Drahos's analysis of mechanisms is generic and synthetic. It is often unclear which mechanisms are employed, by which actors for what purposes, or how they operate.
Finally, scholars should investigate the structural factors that affect mechanisms and pathways. Santos identifies six structural places in society, each with its own interactive logic. The coexistence and overlap of structural places mean that interactions "are often informed by different and mutually incongruent logics". 169 Alternatively, pathways of interaction may coincide with or exploit the pathways of economic production and consumption. 170 Markets, hierarchies, networks and communities may have different logics that affect the availability or operation of mechanisms. For example, coercion may be more common and effective in a hierarchy than in a network. 171 Braithwaite and Drahos observe that webs of dialogue are more common and more often effective than webs of reward and coercion, 172 and that strategic wisdom lies in knowing "which strand(s) to seek to tighten at which moment in order to tauten a web that floats in time and space". 173 Networks themselves may be conceived as pathways of interaction or structures conditioning interaction. Variation in network characteristics may also influence the character of interaction between regulatory networks and conventional regimes. 174 In this special issue, Biggins and Scott identify five mechanisms of interaction in the case of derivatives: ISDA standard-setting; its lobbying and other activities designed to shape the regulatory regime for derivatives; its intervention in important court cases; its provision of advice on systemically important events; and its central role in reforming credit default swap settlement mechanisms after the global financial crisis. These are not mechanisms in our sense, however; they are case-specific strategies. Yet it may be possible to abstract more general mechanisms from these examples. For example, ISDA's standard-setting and lobbying, and states' passage of ISDA's model laws, may be instances of modeling, norm entrepreneurship and rule incorporation.
This point highlights the need to clarify what constitutes a mechanism or pathway, and how they are distinct from strategies, modes, drivers and shapers of interaction. For analytical purposes, mechanisms must be abstracted from specific actions, but concrete enough to be recognisable. Braithwaite and Drahos, for example, focus not on higher-order, abstract mechanisms like reinforcement, but on lower-order, concrete mechanisms like reward and coercion.
Buhmann's article poses similar issues. She describes networks, discursive strategies and even TBG schemes as mechanisms of interaction. Yet her examples suggest the mechanisms of modeling, mimicry, rule incorporation, learning and other cognitive interaction mechanisms. 175 Buhman also suggests that mechanisms at the evaluation and review stage include cross-scheme comparison of experiences, investigation of complaints, and governmental reporting. These resemble generalisable mechanisms of peer review, benchmarking, learning and experimentalism.
Doorey proposes a typology of four actor-level and three scheme-level mechanisms and pathways. In the first four, the pathways are actors who participate in multiple TBG schemes. These mechanisms include dialogue and norm entrepreneurship, with actors "like honey bees pollinating TBG schemes with ideas and norms"; 176 information distribution and knowledge sharing across schemes; supply chain coordination, with suppliers selecting or avoiding more stringent codes; and policy learning and diffusion, through education, research, and social networks. The relationship between dialogue and norm entrepreneurship, on one hand, and policy learning and diffusion, on the other, is not entirely clear: both involve actors spreading the "pollen" of norms and knowledge. The first scheme-level mechanism is mimicry or mimetic isomorphism, in which a TBG scheme adopts successful designs or norms; Doorey seems to treat this as synonymous with modeling, raising the definitional issue discussed above. The final two mechanisms are conditional rule referencing and metaregulation.
Webb's description of ISO 26000 as a bridge suggests a master metaphor for several mechanisms of interaction. Webb identifies several bridging functions: top-down transposition of concepts from intergovernmental to non-state instruments; bottom-up transposition of concepts from narrow non-state standards to broader social responsibility activities; uniting public, private and civil society sectors in standards development and implementation; and the emergence of a global custom. However, the main mechanisms at work in the case are modeling, institutional isomorphism and conditional rule referencing.
Webb's study indicates that these mechanisms may work top-down or bottomup, and may amplify the source standard, not just duplicate it. Thus ISO 26000 extended the due diligence and Plan-Do-Check-Act approaches to all social responsibility activities, from human rights and management systems respectively. The case also highlights meta-regulation and contracting as coordination mechanisms. Contracting through Memoranda of Agreement between ISO and other instrument developers facilitated top-town transposition and normative convergence. Finally, the rich normative conversation achieved in the ISO 26000 drafting process appears to embody Doorey's mechanism of dialogue.

The development of interactions over time and space
"It is easier to call for a dynamic theory than to produce one", Abbott and Snidal remind us. 177 Yet theories of TBGI must seek to capture spatio-temporal dynamics. 178 Over time, institutional structures and processes converge and diverge; some norms and practices spread, others do not; standards are ratcheted up or down; states enter and retreat from governance fields; non-state regulatory authority waxes and wanes; conflicts transform into cooperation or vice versa; some TBG schemes wither away and others dominate. Change over time is a key dimension of our TBGI analytical framework, but the spatial dimension also merits more attention. Transnational business governance has an uneven geography that implicates disparities between legal centres and peripheries. 179 Theoretical approaches that integrate spatial and temporal dynamics into their conceptual apparatus should be especially fruitful-examples include the recursive dynamics in theories of experimentalism, transnational legal orders and open systems. Shaffer and Halliday theorise transnational legal ordering as a dynamic, recursive process of interaction among transnational and national law, characterised by mutual influences, temporary and contingent settlements, and periodic destabilisations triggering further recursive cycles. 180 Experimentalist governance involves recursive interactions between central and local regulating units. 181 Doorey's open systems theory integrates change within its processual model of inputs, outputs and feedback loops. Calliess and Zumbansen's RCRC model emphasises ongoing processes of normative experimentation and recombination. 182 When addressing spatio-temporal dynamics, theories should seek to explain instances of governance failure, not just of success. 183 Meinhard Doelle and others, for example, apply David and Louise Trubek's concepts of complementarity, rivalry and transformation 184 to the failure of the Forest Carbon Standards Committee. 185 It would be instructive to examine whether interactions with other governance actors and schemes contributed to this failure.
All the articles in this special issue describe fascinating changes over time, including changes in the scale, character, mechanisms and effects of interaction over time and across institutional settings and components of regulatory governance. Doorey highlights another important aspect: tension between change at the level of norms and persistence at the level of on-the-ground practices, which may give rise to further rounds of interaction and the transformation of governance schemes. All these articles invite further theoretical elaboration of temporal and spatial dynamics.

Conclusion
Interactions among legal actors, institutions, norms and orders are central features of transnational law, yet transnational legal theory casts only fleeting or narrow glances at them. The time is ripe to make the interactive dynamics of transnational law the subject of sustained empirical and theoretical attention. Efforts to govern global business provide a useful context in which to do so. A rich body of interdisciplinary empirical research on this subject is emerging, but more theory building is needed to advance scholarly understanding of the dynamics and effects of transnational business governance interactions.
The relative lack of attention to theory in the special issue articles (with the notable exception of Doorey) is understandable at this stage in the TBGI research program, when the focus remains on rich empirical description. The articles demonstrate that our TBGI analytical framework provides a useful common language for conducting and comparing such research. But the framework's relative agnosticism as to theoretical and methodological approaches should not be mistaken as invitation to ignore theory. Quite the opposite: we hope to spur researchers to develop, test and refine theories.
The most fruitful theoretical approaches, we believe, will recognise the heterogeneity of interacting entities; emphasise meso-level interactions among regulatory authorities and schemes; explore the links between interactions at different scales, between agency and structure, among components of the regulatory process, between drivers and effects and among modes of interaction. They will also examine the impacts of interactions on the capacity and performance of regulatory systems, specify concrete interaction mechanisms and strive to capture spatio-temporal dynamics.
No single theory can accomplish all this. A variety of theoretical perspectives will be needed to illuminate this complex phenomenon. Theory building, then, can focus on middle-range generalisations about specific features of TBGI, with the aim of building up "a 'toolbox' of explanations that can be widely applied, gradually increasing the field's stock of understanding". 186 There are many possible candidates, including problem structures; interaction mechanisms such as modeling, conditional rule referencing and supply chain coordination; interlinkages among mechanisms; regulatory ratchets; indicators and determinants of regulatory capacity; meso-level processes like meta-regulation and orchestration; cross-scalar strategies like forum-shifting and norm entrepreneurship; and structural logics of action. Theoretical "modules" like these are the building blocks of middle-range theory. 187 Thick descriptive research and inductivist middle-range theory building will allow us to understand better the impacts of governance interactions on regulatory outputs, outcomes and capacities and, ultimately, on what really mattersthe social and economic performance of transnational business and the lives of ordinary people.

Disclosure statement
No potential conflict of interest was reported by the authors.