Member-dominated international organizations as actors: a bottom-up theory of corporate agency

Abstract

This article introduces an innovative theoretical conception of the corporate agency of international organizations (IOs). Existing rationalist and constructivist accounts attribute IO agency to the influence of intra-organizational agents. Drawing on general conceptions of corporate agency in International Relations, sociology, and philosophy, we elucidate how IOs can develop corporate agency, even if the member states prepare and adopt all organizational decisions themselves. In line with recent studies on international political authority, we replace the IO-as-bureaucracy model with the more comprehensive concept of IOs-as-governors. To establish the micro-foundations of IO agency, we adopt a bottom-up perspective and outline how, and under which conditions, IO agency arises from the interaction of constituent actors. Irrespective of any specific institutional design, IOs become actors in their own right whenever they gain action capability and autonomy. They acquire action capability whenever their members pool governance resources like the right to regulate certain activities or to manage common funds and authorize IOs to deploy these resources. IOs gain autonomy whenever they affect organizational decisions. Both dimensions of IO agency are variable and open to empirical enquiry. To illustrate our argument, we refer to the United Nations Security Council and other IOs with member-driven decision processes.

Introduction

While international organizations (IOs) with extensive bureaucracies are widely recognized as important actors in contemporary world politics, the agency of member-dominated IOs has largely remained uncharted territory. By IO agency, we refer to an international institution's quality of being an actor with the capability to influence world politics in its own right and some autonomy, that is, the ability to determine its action according to its own logic. It is among the most intriguing questions for International Relations (IR) scholars whether international institutions are actors in their own right, and how they might acquire agency.¹ Constructivists, rational institutionalists, and international public administration scholars have shown that international bureaucracies and other intra-organizational agents may transform IOs into actors that actively influence world politics.² However, IR theory lacks a conceptual understanding of whether, how, and under which conditions member-dominated IOs like the United Nations Security Council (UNSC) and informal organizations³ like the Basel Committee on Banking Supervision or the G-7/8 World Summits may gain agency. Do these institutions merely provide arenas for activities of their members or do they become entities with the capability to act in their own right, distinct from the action of their members? If they acquire separate agency, we should identify the sources and nature of their action capability and understand the ways in which they can influence world politics.

IOs are frequently conceived of as separate actors, even if their decision processes are strongly dominated by their member states. Yet, it is conceptually entirely unclear whether this is more than metaphorical language and whether member-dominated IOs are more than arenas for interaction of their member states. For instance, UNSC decisions are made by Council member states. Still, Abbott and Snidal note that this institution has a unique ‘power of action’ and acts as a ‘community representative’ of the United Nations' constituency.⁴ Likewise, the World Trade Organization (WTO) is under tight control of its membership and has a weak secretariat. However, Lake presents it as a governor authorized to make binding international rules far beyond its centralized dispute settlement system.⁵ Multilateral environmental institutions like those related to the Kyoto Protocol and the Paris Agreement on climate change⁶ comprise extensive decision-making apparatuses organized around conferences of the parties.⁷ They have law-making powers and compliance mechanisms⁸ and may even acquire international legal personality.⁹ Even the largely informal G-20 is identified as ‘a locus of decision-making authority which has the power to direct specific IOs to “follow up” G20 determinations’.¹⁰ The widespread practice of treating member-dominated IOs as actors suggests that at least some of them may indeed have agency. Yet, this hunch raises the question of whether and how organizational agency can arise from institutionalized interaction among member states.

We develop a theory of IO agency with strong micro-foundations that expands existing IR institutional theory. It explains how institutionalized interaction among member states may generate IO agency and what this implies for world politics. We challenge the prevailing view that IO agency arises predominantly, or exclusively, from the activities of secretariats and other intra-organizational agents and argue that member-dominated decision processes might also generate IO agency. To locate the sources of organizational agency, we employ an ideal-typical dichotomy. ‘Member-dominated IOs’ are institutions in which the member states enjoy exclusive decision-making authority. In contrast, in ‘bureaucratic IOs’ decisions are made by agents like courts or secretariats. In reality, of course, most IOs comprise, in various combinations, both activities of intra-organizational agents and of member state bodies.

Established IO theories have largely sidestepped the question of whether and how member-dominated IOs acquire agency. Constructivists emphasize the autonomy of IOs as actors, but attribute agency almost exclusively to intra-organizational agents, particularly to secretariats¹¹ and international courts.¹² Rational institutionalists conceptualize IOs as arrangements established by member states in principal–agent relationships and attribute IO autonomy to undesired action of these agents (‘agency slack’).¹³ Political authority scholars conceptualize IOs as governors of areas of world politics¹⁴ that adopt organizational decisions by pooling of member state votes.¹⁵ None of these contributions examines the possible sources and consequences of agency of member-dominated IOs.

To theorize the intricate relationship between organizational agency and the activities of constituent actors, we draw on general scholarship on corporate agency. A small body of political science research discusses issues of corporate agency in broader terms with a particular focus on states¹⁶ and non-state actors in world politics.¹⁷ These contributions adopt a general perspective on collective agency beyond IO secretariats and other institutional arrangements. To acquire agency, member-dominated IOs must be established as separate entities of world politics equipped with action capability and autonomy.¹⁸ Contributions from sociology and philosophy explore the micro-foundations of collective agency in general terms. A sociological theory of corporate agency shows how groups of actors form organizations that gain action capability.¹⁹ Likewise, a lively debate in analytical philosophy examines how group actors obtain autonomy,²⁰ develop ‘minds of their own’,²¹ and acquire moral responsibility.²²

Against this backdrop, we replace the IO-as-bureaucracy model with an IO-as-governor perspective and adopt a bottom-up perspective to investigate the micro-foundations of organizational agency. We conceptualize IOs as decision-making institutions authorized to manage cooperation projects by adopting binding decisions. In line with Coleman's sociological theory of corporate agency²³ and the IR conception of IOs as international governors,²⁴ we argue that an IO gains action capability, whenever its members pool governance resources, such as competencies to regulate areas of IR or manage common funds, under its authority. IOs act by making organizational decisions on the use or non-use of such resources; and this action cannot be attributed to their member states or any other actor. In line with philosophical conceptions of group agency,²⁵ we argue that IOs gain autonomy, whenever organizational processes and structures influence the course of organizational action and produce specific organizational rationales. Consequently, IOs acquire agency even in the absence of intra-organizational agents, if they are authorized to adopt governance decisions that are potentially relevant for world politics and if these decisions are subject to a minimum of organizational autonomy.

Our theory of IO corporate agency yields two important insights for the study of world politics. (1) Member-dominated IOs become potentially powerful actors, if their members authorize them to make governance decisions according to organizational procedures. The power of IOs arises from the nature and extent of governance resources under corporate control. The mere coordination of collective member state action does not create organizational agency. (2) Member-dominated IOs may gain autonomy and become subject to processes of gradual autonomization even if all decisions are adopted by their members. Procedural and substantive decision rules shape organizational action and provide criteria for organizationally desired decisions. Organizational rationales dissociate IO decisions, more or less extensively, from aggregated member state preferences. As a result, the question of whether or not IOs acquire agency is empirical, not conceptual. Both, the type and amount of resources under corporate control and the sources and effects of autonomy vary across IOs and can be investigated empirically.

Our conceptual analysis provides the foundation for an empirically fruitful research programme on IO agency and a new perspective on related normative implications. By examining the nature, scope, and limits of IO power to shape world politics, we may distinguish between IOs with minimal action capability, such as the G7/8, and those with extensive action capability, such as the UNSC. By focusing on the sources and effects of organizational autonomy we may better understand patterns of emerging organizational identities and cultures. For instance, we might account for the different perspectives on the desirability of environmentally motivated trade restrictions developed by the WTO and IOs managing multilateral environmental treaties.²⁶ Finally, we may examine the normative implications of IO agency, which are presumably closely related to the sources and effects of organizational autonomy.²⁷ A reasonable conception of member-dominated IOs as actors in their own right is a precondition for examining all these subjects for heavily member-driven IOs.

This article contributes to three current IR debates. First, the corporate agency perspective broadens the scope of the discussion on IOs as actors in world politics. Although established conceptions attribute IO agency virtually exclusively to intra-organizational agents,²⁸ we demonstrate that member-dominated IOs can also acquire agency. The latter are likely to matter as actors in their own right, if they have action capability and autonomy. To ignore their agency is tantamount to conceiving them as mere arenas for member state coordination. This, however, grossly underestimates pivotal organizational implications of IOs that lack extensive secretariats. Second, our conception fills a conceptual gap in existing scholarship on international political authority. Thus far, member-dominated IOs are only conceived of as international governors that may exercise political authority but not as collective actors – because a suitable concept of agency is lacking.²⁹ Third, we contribute to the fragmented debate on general issues of corporate agency in IR.³⁰ Drawing on little-used conceptions of corporate agency in sociology³¹ and analytical philosophy,³² we provide a micro-foundation for the emergence of IO agency from interaction among an IO's constituency. Our conception recognizes the existence and relevance of IO agency as a social fact but submits it to a ‘no mystery’ constraint’.³³ Thus, it bridges the opposing camps of individualists denying corporate agency altogether and holists attributing agency to organizations without enquiring into their origins and limits. Our conception is sufficiently general to be applicable to other types of international actors, such as non-governmental organizations.

The next section lays the ground for our theoretical argument. We explore existing IR perspectives on IO agency and international political authority, identify the need for a re-conceptualization of IO agency, and briefly discuss some fundamental issues of corporate agency. In the following section, we develop a bottom-up theory of IO agency in four steps. We start with a brief sketch of James Coleman's combining resources model, discuss the basic nature of IOs as decision-making systems, and demonstrate how member-dominated IOs can gain action capability and autonomy. We illustrate our theoretical argument with evidence from a variety of different member-dominated IOs to show that corporate agency is an empirically widespread phenomenon.

IOs as actors in IR and the puzzle of corporate agency

State-dominated IOs like the UNSC raise the thorny issue of corporate agency in a strong version. Do they act in their own right, or is attributed action merely a shortcut for the actions of the member states and their representatives?³⁴ What does it mean that the UNSC acts, given the fact that it is an organization under strict control of its 15 members, in particular the Permanent Five? Scholars do not hesitate to treat the UNSC as an actor in its own right³⁵ and the Council considers itself as an actor too. For example, Resolution 1973 (2011) states that ‘the Security Council …, acting under Chapter VII of the Charter of the United Nations … authorizes Member States … to take all necessary measures … to protect civilians and civilian populated areas under threat of attack in the Libyan Arab Jamahiriya; … [and] decides to establish a ban on all flights in the airspace of the Libyan Arab Jamahiriya …’.

Theory of international institutions: in need of a conception of organizational agency

IOs are today widely recognized as actors that perform a wide variety of functions for their membership.³⁶ They centralize collective activities, including the production of goods that cannot be produced by any state alone. IO scholars hold that IOs pursue their own organizational interests and have some degree of autonomy.³⁷ Without a minimum of autonomy, they would merely be transmission belts for the activities and preferences of their members. IOs are distinct from other international institutions like regimes, rules, and conventions, which may influence the behaviour of other actors, but reflect sectoral orders or international norms. The distinction between organizations-as-actors and rules of the game is well established in both rational institutionalism and constructivism. However, current approaches cannot account for IO agency arising from interaction among member states and, consequently, treat member-dominated IOs as mere forums for coordination of member state activities.

Constructivist approaches emphasize bureaucracies and other non-state entities as sources of organizational agency. However, constructivists tend to marginalize the role of member states and member-driven decision processes. Barnett and Finnemore seek to provide ‘a theoretical basis for treating IOs as autonomous actors in world politics’ and thus challenge ‘the statist ontology prevailing in international relations theories’.³⁸ Drawing on Weberian bureaucracy theory, they conceptualize ‘international organizations as bureaucracies’.³⁹ Accordingly, they attribute IO autonomy to characteristics of bureaucracy, such as internal rules, legitimacy, and control of reliable knowledge as well as organizational cultures. Numerous authors follow this path.⁴⁰ Other scholars examine the role of international courts.⁴¹ While this branch of literature elucidates the autonomy of bureaucratic and judicial agents, it largely, if not completely, ignores the member states and understates the role of state-dominated decision processes, which are highly important in most IOs.

Rational institutionalists draw on the principal–agent approach to grasp IO agency. However, their seemingly more comprehensive analytical perspective also attributes IO agency largely to intra-organizational agents. In their seminal volume on IOs and the principal–agent approach, Hawkins et al. claim to ‘treat IOs as actors in their own right’.⁴² The member states delegate, as principals, certain tasks to an agent that is empowered to act on their behalf. The agent's ability to defy control and to pursue its own interests despite oversight arrangements (‘agency slack’) is a source of autonomous organizational agency.⁴³ Hence, from a principal–agent perspective, the member states are part of the analysis, but not drivers of IO agency.

The current debate on international political authority in global governance reinforces the demand for a new theoretical conception of IO agency. Scholars observe that states increasingly yield sovereignty to IOs and submit to their rulings.⁴⁴ Political authority denotes rightful legitimate rule.⁴⁵ International institutions have political authority ‘when states recognize, in principle or practice, their ability to make … binding decisions on matters relating to a state's domestic jurisdiction’.⁴⁶ This does not preclude that addressees contest and politicize individual orders, as long as they generally accept the exercise of political authority as legitimate.⁴⁷

The exercise of international political authority presupposes the existence of governors that actively execute authority functions. Political ‘authority is always wielded by someone for some purpose’⁴⁸ and ‘governance is the exercise of authority by an actor over some limited community’.⁴⁹ The exercise of international political authority through IOs is distinct from other forms of international governance. States can govern areas of IR by concluding international treaties that do not envisage the transfer of political authority to an IO. Such regimes may become causally powerful without gaining agency on their own.⁵⁰ Likewise, fundamental international norms, such as state sovereignty, may gain authority, understood as being accepted by actors and guiding their behaviour,⁵¹ without becoming actors.

The political authority perspective includes member-driven IOs and thus transcends the narrow focus on IOs-as-bureaucracies. Scholars distinguish between two ideal-typical forms of intra-institutional decision-making, namely delegation to agents and pooling of votes.⁵² Member states can choose among these two forms of IO decision-making to exercise international political authority.⁵³ Accordingly, state-dominated IOs may become governors of areas of IR, even if decisions are predominantly or entirely made through pooling. Bradley and Kelley identify member state bodies of IOs as entities that may receive and hold authority.⁵⁴ If such bodies are vested with the authority to adopt binding decisions, they differ from multilateral treaty conferences to which states do not grant authority. Likewise, Green and Colgan examine patterns of authority transfers to treaty bodies of multilateral environmental agreements, which typically lack powerful bureaucracies.⁵⁵ This raises the question, whether member-dominated IOs can become actors in their own right, or what else it might mean that they operate as ‘governors’ of areas of world politics.

Although the contributors to this debate have not yet addressed the intricate issues of corporate agency, the conception of political authority opens a promising road for the theoretical exploration of IO agency. Political authority is always based on a relational contract. Since IOs ought to overcome collective action problems, their authority results from ‘a social contract in which a governor provides a political order of value to a community in exchange for compliance by the governed with the rules necessary to produce that order’.⁵⁶ The obligation to comply with IO decisions does not follow from the commands of the ruler, nor from the inherent properties of the IO, but from state interest in and satisfaction with the social order so produced. Zürn et al. emphasize that ‘international institutions exercise authority in that they successfully claim the right to perform regulatory functions like the formulation of rules and rule monitoring or enforcement’.⁵⁷

By replacing the IO-as-bureaucracy perspective with the IO-as-governor perspective, the analytical focus shifts from institutional properties to governance functions. Instead of locating IO agency in substantial institutional characteristics, we should focus on relations between member states and the IO.⁵⁸ IOs acquire agency because they are empowered to exercise political authority, not because they have bureaucracies or other institutional features. However, bureaucracies might reinforce the sincere and effective exercise of governance functions.

The puzzle of corporate agency

Despite a general recognition of IOs as actors, IR theory has almost totally sidestepped some fundamental issues of corporate agency raised by virtually all relevant international actors, including IOs, states, and NGOs. Historically, state agency has not been a subject of thorough enquiry, because states are constituent for the very discipline of IR, which traditionally conceives of international relations as relations among states.⁵⁹ More recently, studies of global governance have enlarged the set of relevant actors, especially to private actors and IO secretariats,⁶⁰ however largely without enquiring into the foundations of their agency.

All composite actors raise the major conundrum arising from corporate action, as pinpointed by Verba: ‘It is a truism that all action within the international system can be reduced to the action of individuals. It is also true, however, that international relations cannot be adequately understood in terms of individual attitudes and behaviours’.⁶¹ The fundamental issue is whether we should consider corporate entities as actors in their own right; or whether treating them as actors is merely a shortcut for the actions and activities of their constituency.⁶² At stake is whether our common expressions (e.g. ‘the Security Council acts under Chapter VII’) are more than useful (or even misleading) metaphors. There are two fundamentally different positions on this issue.

Individualists refer to corporate agency merely as a metaphorical shorthand for the actions of involved individuals. For example, Gilpin holds ‘that the state does not really exist; … neither do … interest groups, nor even transnational actors … Only individuals really exist’.⁶³ Lomas emphasizes ‘that it is never right to speak of states acting’ and that the common tendency to do so is ‘a misleading habit’.⁶⁴ This position is not only (often implicitly) held by virtually all methodological individualists, but also by constructivists. Both narrativists⁶⁵ and structuralists⁶⁶ reject the claim that states (and other corporate actors) are real, although they may be like persons in some important respects.⁶⁷ In the IR debate on IOs, the individualist perspective is mirrored by the claim that IOs cannot become actors in their own right because they are created and dominated by their member states, as realists tend to assume, or because they are ‘meta-organizations’ composed of members, which are themselves corporate actors.⁶⁸

In contrast, holists hold that at least some corporate actors are real and can act on their own. In IR, Alexander Wendt advocates this position most vigorously. He argues that states are ‘persons’ that act purposively and intentionally.⁶⁹ He conceptualizes corporate actors as persistent social structures that are primarily defined by an ‘idea’ of corporate agency and a separate institutionalized decision structure.⁷⁰ Likewise, organization studies generally assume that organizations are actors without discussing the foundations of organizational agency.⁷¹ In the IR debate, the holist position is associated with the claim that IOs become actors in their own right because they comprise intra-organizational agents that are capable of acting autonomously.⁷²

To investigate how the agency of member-dominated IOs emerges from interaction among member states, we adopt a bottom-up approach that sidesteps this metaphysical conundrum. An extended discussion of the pros and cons of the individualistic and the holistic positions of collective agency is beyond the scope of this article, and it is unlikely to settle the matter. A bottom-up approach promises to elucidate the micro–macro link between activities of the member states at the unit-level and effects of organizational agency at the collective level.⁷³ It demonstrates how higher-level actors and their properties emerge from interaction among lower-level actors, although the properties of the former may not be easily reducible to the properties and actions of the latter. Think of the physical world, in which water emerges from interaction of oxygen and hydrogen, but has qualities that differ profoundly from those of these two chemicals. Generally, new qualities of higher-level actors are emergent, if they are not merely the result of aggregation, but of social organization.⁷⁴

The philosophical concept of supervenience provides an abstract idea of how interaction among member states might produce the collective effect of IO agency. It reflects an individualistic concept of the emergence of organizational agency and comprises a ‘no mystery’ constraint on social facts.⁷⁵ Supervenience implies that macro phenomena (of organizational agency) arise from (‘are supervenient on’) events at the micro level. Accordingly, any change at the macro level necessitates a change at the micro level. Yet, not all social phenomena are clearly related (i.e. fully reducible) to one particular configuration of individual actions.⁷⁶ List and Spiekermann identify two conditions under which a social phenomenon does not clearly reflect a particular configuration of individual actions.⁷⁷ First, a particular higher-level property may be caused in different ways (multiple realizability). For instance, a UNSC resolution which is duly adopted by a majority of UNSC members certainly arises from actions of lower-level actors. However, the exact configuration of members forming the necessary quorum is contingent and irrelevant for the validity of the resolution. Second, organizational decisions may be robust against changes at the micro level (microrealization–robust causal relations). For example, the entry of a new member state or representative does not affect the validity and implications of existing UNSC resolutions.

The general debate on corporate agency in philosophy and sociology tackles some general issues that are highly relevant for the theoretical analysis of IO agency. An important strand of scholarship addresses the question of whether states can act in their own right, or whether their actions are generally made by state officials and citizens on their behalf.⁷⁸ From a bottom-up perspective, sociologist James Coleman provides an abstract idea of how interaction among member states may produce organizational action capability.⁷⁹ It is compatible with the conception of IOs as governors exercising political authority and has been tentatively employed to examine the external action capability of the European Union.⁸⁰ Another strand of scholarship, especially from philosophy, examines the autonomy of group actors. Groups may develop distinct ‘beliefs’ that depict how things are in the world, ‘desires’ that specify how things should be modified through group action, and the capacity to process group beliefs and desires to determine group action.⁸¹ Important contributions provide bottom-up ideas on how group actors may develop ‘a mind of their own’⁸² from interaction of their members and how IOs might even acquire normative responsibility for their actions and the capacity for adaptive learning.⁸³

This debate yields a general conception of core properties of corporate actors. Actors must feature the three characteristics of corporeality, action capability, and autonomy.⁸⁴ Corporeality identifies an entity as ‘anchored and located in time and space’.⁸⁵ While individuals are identifiable through their physical bodies, corporate actors do not have a physical existence. Still, we need to distinguish them clearly from the rest of the world. Action capability refers to an entity's ability to act in ways that make a difference beyond its own confines. Without this ability, an IO would be irrelevant for world politics. Autonomy implies the existence of some capacity for autonomous decision-making, sometimes called intentionality, without which an actor could not meaningfully choose among available options. It captures the ‘commonsense notion of purposive action’.⁸⁶ Similarly, Scharpf argues that actors are generally ‘characterized by their orientations (perceptions and preferences) and by their capabilities’.⁸⁷ To theorize the agency of member-dominated IOs, we must demonstrate how corporeality, action capability, and autonomy arise from institutionalized interaction of member states.

Towards a bottom-up theory of IO corporate agency

To base the conception of IO agency on a strong micro-foundation, we explore generative mechanisms that elucidate how organizational agency arises from interaction of constituent actors. For this purpose, we adopt a relational perspective and examine how interaction of the member states creates and sustains IO agency. We assume that actors behave strategically according to their preferences and evaluate alternatives regarding their consequences. The rational actor assumption raises the burden of proof, because we cannot attribute the emergence of organizational agency to the preparedness of actors to sacrifice the pursuit of their own (individual) interests for the sake of community well-being, to socialization, or to unreflected internalization of norms and community demands. The rational actor assumption also ensures compatibility with mainstream IO theory.

The theoretical baseline: Coleman's combining resources model

Coleman's ‘combining resources model of corporate actors’⁸⁸ provides a general idea of the emergence of corporate agency and the general purpose of organizations. Coleman seeks to explain the macro-effect of corporate agency from interaction of strategically acting members at the micro level.⁸⁹ Although Coleman's model is highly compatible with modern institutional theory, it has not gained much attention in IR theory.

The foundation of organizations is closely associated with the institutionalization of mutually desired cooperation projects. Since IOs are costly and may constrain national policy-making, strategic actors will only establish them if they expect cooperation gains in return. Although organizations may partly eschew control and gradually develop ‘a life of their own’, they are initially established in line with the interests of their constituency to realize mutually beneficial cooperation projects.

Whenever a group of actors pools governance resources and submits them to organized collective management, these actors create an organization in the sociological sense.⁹⁰ Pooling of a resource, like development aid funding, means that the member states assign to the organization the right (competence) to use this particular resource and thus sacrifice their previous right to deploy it unilaterally. While pooling is frequently defined as a specific form of (majoritarian) decision-making,⁹¹ we conceive of pooling as a distinct feature of institutionalized cooperation.⁹² In essence, the very purpose of organizations is to decide on the use of centralized governance resources to realize collectively agreed goals.⁹³

Whenever states pool governance resources under corporate control, they must adopt rules and procedures according to which organizational decisions are made. These rules and procedures form the ‘constitution’ of an organization⁹⁴ and establish the organizational decision-making system. Henceforth, organizational decisions on the use of transferred resources are adopted according to this constitution. Organizational rules and procedures may provide that decisions are adopted by the constituent members, by intra-organizational agents, or by specialized decision-making bodies, and they may grant selective participation rights to other actors.

Coleman's combining resources model elucidates that corporate agency is based on a specific form of managing cooperation projects and not on a specific institutional design. The transfer of the right to control governance resources from the constituent actors to the organization is pivotal. Without such transfer, the new organization would be largely meaningless or merely symbolic. Essential is also the establishment of an organizational decision-making system, without which organizational decisions on the use of pooled governance resources could not be made. Everything else is variable, including the nature and extent of pooled resources and the specific procedures according to which organizational decisions are made. The corporate agency perspective is thus compatible with all forms of organizational decision-making. Accordingly, assigning the competence to use pooled resources is an indispensable first step, while delegation of decision-making authority to intra-organizational agents is an optional second step.⁹⁵

Coleman's conception provides a first orientation for the enquiry of the three predispositions of organizational agency identified above, namely corporeality, action capability, and autonomy. In the following sub-sections, we examine them separately.

The precondition: IOs as decision-making institutions

To become actors in their own right, IOs must be identifiable as separate institutions of world politics. Only entities that are clearly anchored and located in time and space can exercise agency.⁹⁶ This does not necessarily imply physical characteristics. What matters is the institutionalization of IOs and their clear separation from the outside world.

The separate existence of IOs arises from three distinct characteristics. First, the actors that establish an IO form a clearly identifiable group. This group defines membership and associated rights and obligations. IOs are clubs with clear boundaries between members and non-members. Second, the members define, more or less specifically, areas of joint activity. Issues falling within the realm of organizational activity can thus be distinguished from issues and problems located outside this realm. Third, the definition of decision-making procedures separates organizational decisions from all other decisions and activities. Accordingly, IOs may be defined as durable institutions established by states with a clearly delimited membership, a specification of the tasks of cooperation and procedures for making organizational decisions. In contrast to many definitions of IOs,⁹⁷ which presuppose a formal treaty or existence of a secretariat, this definition includes both member-dominated and even informal IOs, such as the G-7/8,⁹⁸ and highly bureaucratic ones, such as the International Monetary Fund (IMF). However, it distinguishes IOs from organizations established by non-state actors and from loose groups, such as ad hoc coalitions or social movements.

Just like any other organization, IOs are ‘goal-directed, boundary-maintaining, activity systems’.⁹⁹ IO boundaries are contingent but relatively stable institutional structures. They are based on agreement among members and allow a limited number of actors to focus their attention on a subset of relevant issues pending among them.¹⁰⁰ Any meaningful cooperation project needs some limitation of the range of issues and participating actors. Accordingly, all actors interested in cooperation prefer maintaining some boundaries, although they may advocate different ones. Attempts to pursue issues located outside an IO mandate are likely to prompt resistance of some members. Consequently, once established, IO boundaries remain stable unless actors can agree on their change.

Interaction among relevant actors inside organizational boundaries is shaped by organizational decision procedures¹⁰¹ and directed at influencing organizational decisions.¹⁰² Formal procedures may be complemented by informal rules and practices.¹⁰³ Hence, IOs constrain the ways in which states can advocate their preferences. They provide additional spheres of interaction, clearly separated from the rest of world politics. Constituent actors can act both within and outside an IO.

Nothing in this conception is mysterious. An IO emerges as a separate entity of IR entirely from interaction among actors at a lower level of social organization, primarily the member states. These actors establish the IO to realize joint cooperation interests; they define membership, the realm of organizational activity, and decision-making procedures. The IO is identifiable as a separate institution of world politics even if its decision procedures provide that a member state body adopts all decisions, and irrespective of whether this body decides by unanimity or majority. While this basic conception of IOs as decision-making institutions is unlikely to raise major disagreement, it differs profoundly from IO conceptions that emphasize the constitutive role of IO secretariats, on which existing approaches of IO agency are based.

Action capability: gaining efficacy in world politics

To become actors in their own right, IOs must acquire action capability, or what Hofferberth calls ‘efficacy’.¹⁰⁴ IOs have action capability, if, and to the degree that, they can influence world politics through their own actions. An entity does not qualify as an actor if it is incapable of performing activities that ‘make a difference’ beyond its confines.¹⁰⁵ All current IR actor conceptions presume, at least implicitly, that actors can influence their environment through their own actions. This does not imply that action is successful. What matters is the ability to influence world politics. Actors have agency, even if their action fails or if they choose not to act in a particular situation.

Following Coleman's combining resources model, IOs are equipped with action capability, whenever their members authorize them to make decisions that are intended to be binding¹⁰⁶ or to create effects beyond their confines. Accordingly, both member-dominated IOs and those with extensive intra-organizational agents can gain action capability. IO action capability follows from the right to decide on the use of pooled governance resources, not from the IO's actual possession or ownership of such resources.¹⁰⁷ Governance resources may include the right to create and amend legal rules, to regulate specific policy areas, to implement joint policies by deploying pooled financial and administrative assets, to adjudicate disputes, to monitor and enforce compliance with international commitments, or to issue recommendations, opinions and interpretations intended to create external effects.¹⁰⁸ For example, the UNSC has the right to authorize the legitimate use of force to restore international peace and security, to impose binding international sanctions, and to deploy blue helmet troops in crisis situations.¹⁰⁹

Authorizing an IO to adopt decisions on the employment of pooled resources inevitably involves a loss of member state sovereignty.¹¹⁰ It changes the default condition, that is, the outcome that occurs if actors cannot reach an organizational decision.¹¹¹ As long as states retain control of governance resources, they can employ these resources unilaterally if coordination fails. In contrast, pooled resources cannot be used at all if organizational decision-making is blocked.¹¹² Although institutional theory and political authority scholars¹¹³ typically assume that in the absence of an agreed decision members remain free to pursue their own goals with their own means, this does not hold for organizations. For example, development aid funds under national control can be used unilaterally if coordination fails, while funds pooled in an IO can only be allocated by organizational decision and are blocked if no decision can be achieved. In both cases, funds originate from member states, but the default condition differs.

Accordingly, IO action capability arises from a permanent authority relationship between the constituent members and the organization,¹¹⁴ known from the debate on international political authority.¹¹⁵ The member states authorize the IO to act in certain ways to govern an area of IR and generally accept the legitimacy of organizational decisions. Hence, IO corporate agency requires that the member states continuously support the organization and accept its right to exercise political authority, even if they contest particular decisions or occasionally behave non-compliant.¹¹⁶

This raises the question of whether the use of governance resources by an IO reflects genuine IO action, or whether talk about IO action is merely a shortcut for action of its members or relevant individuals like state representatives or international civil servants. Like any other organization, IOs are non-physical entities, which cannot execute physical action by themselves. They cannot negotiate or sign agreements because these activities require physical capabilities. Therefore, Gilpin argues from a methodologically individualist perspective: ‘Only individuals act, even though they may act on behalf of one of these collective social entities’.¹¹⁷ Drawing on Hobbes' conception of the Leviathan as an ‘artificial person’, Fleming suggests conceiving of corporate actors as ‘the “owners” of actions – the entities that are responsible for them – rather than the agents that perform the actions’.¹¹⁸

We argue that IOs can indeed act in their own right without the assistance of agents, if their action does not require physical activity. A closer look at the authority relationship between an IO and its constituency helps to clarify the nature and limits of IO action capability. Take Resolution 1973 (2011), by which the UNSC established a no-fly zone over Libya and authorized member states to take all necessary measures to enforce it. The resolution was based on UNSC authority under Chapter VII of the UN Charter and triggered particular consequences for states and non-state actors.¹¹⁹ It emerged as a joint product of activities of member states and their representatives in the Council, negotiating the resolution text and casting votes. Yet, it is difficult to attribute Council action to the member states. The Council is authorized to adopt legally binding decisions – not a particular group of states or their representatives. Had the same 15 states or their representatives gathered outside the Council framework and reached the same decision, this agreement, although identical in substance, would have had a different effect. As a political statement of a group of countries, it would not have resulted in legitimizing the use of force and would thus not have changed international legal obligations for states and non-state actors around the world. Moreover, the Council member states did not act on behalf of the Council. Their votes represent national foreign policy preferences, possibly of larger groups of UN member states, but not Council interests. Likewise, we cannot attribute Council action to the diplomats, which physically negotiated the resolution text and cast the votes. Certainly, these individuals constitute an important underpinning of Council action.¹²⁰ However, ambassadors represent their member states in the Council; they do not act on behalf of the Council.

A clear distinction between genuine IO action and action by some other actor on behalf of the IO is necessary. As non-physical entities, IOs cannot exercise physical activities on their own. For such action, they need the physical resources of individuals that act on their behalf. Here applies Fleming's distinction between IOs as the owners of action and agents executing such action on their behalf.¹²¹ Hence, the UNSC president, that is, an individual, who happens to be the representative of the member state in charge of the presidency in a given month, occasionally issues statements ‘on behalf of the Council’.¹²² Thus, s/he highlights to act not in a personal role, nor as a representative of his/her own country, but as a Council representative. Likewise, the Council may establish a UN military mission (‘blue helmets’), vest it with the authority to use military force (‘a robust mandate’) and request the UN Secretary General to organize the mission, as it did when establishing the United Nations Mission in Sudan (UNMIS) by Resolution 1590 (2006). However, the Council cannot conduct military action on the ground. It delegates this task to soldiers assigned to the mission by their states. When exercising military power, these individuals act on behalf of the Security Council (or the UN), not in their own name, nor on behalf of their states, nor on behalf of 15 states and their representatives that were members of the Council at the time of decision-making. By delegating the execution of action to these individuals, the Council becomes a principal and establishes a new authority relationship. The delegation contract firmly embeds the soldiers in a tight hierarchical command structure. Therefore, we can distinguish between military action of these individuals as agents acting on behalf of the Council and their individual action, possibly disobeying orders.¹²³ This is also true for IO staff. Had the UN member states authorized the Council to establish a standing army directly employed by the UN, it would not have to ask member states to assign troops for military missions. However, keeping peace on the ground in a crisis area would still require activity of individuals acting on behalf of the Council.

While many IOs, whether bureaucratic or member-dominated, are empowered to make binding decisions, their action capability varies greatly. To assess their action capability, we must investigate their authorization to make decisions intended to create external effects. For instance, the tremendous power of the World Bank and the IMF originates rather from their huge financial assets, which they can hand out to finance development projects or to provide financial assistance,¹²⁴ than from their extensive secretariats. The WTO is inter alia authorized to decide on new members and on temporary waivers of obligations of countries in crisis,¹²⁵ thus changing the implications of world trade rules. Some treaty management organizations established under multilateral environmental agreements can make far-reaching decisions, such as changing treaty obligations or establishing funding schemes.¹²⁶ Even the World Summits (G-7/8 and G-20) may gain selective action capability, when member states use them to direct activities of other IOs.¹²⁷

Accordingly, member-dominated IOs can gain the capability to act in their own right. This capability arises from their authority to adopt decisions that commit other international actors and create effects on world politics. Certainly, IO action capability is based (‘supervenes’) on the actions of lower level actors empowering an IO to make relevant decisions. Moreover, IO decisions arise from activities of lower-level actors, for example, voting of member states. However, they have an emergent quality. For instance, UNSC decisions effectively change the state of international law. This new quality cannot immediately be derived from the negotiating and voting activities of the member states, and it is certainly not an aggregate of these activities. While IO action capability can emerge without the presence or activity of a secretariat, IO secretariats may contribute to all stages of IO action. They may assist, and possibly influence, the preparation of IO decisions in their own name. Secretariats and their staff may implement IO decisions. They may also be authorized to adopt secondary decisions on behalf of the IO.

Instead of locating IO action capability in the activities of intra-organizational agents, we locate it in an IO's authority to adopt decisions. Our concept demonstrates how even member-dominated IOs can acquire action capability and act as governors in IR. Thus, our conception differs fundamentally from the IO-as-bureaucracy perspective and from the principal–agent model and fills part of the conceptual gap identified above.

Autonomy: towards a specific organizational rationale

To gain agency, IOs require autonomy. Purposive action presumes the ability to adopt decisions according to one's own logic.¹²⁸ We argue that even member-dominated IOs can gain a considerable amount of autonomy and develop specific rationales that are typical for organizations.¹²⁹

IOs are autonomous if, and to the degree that, their policies cannot be explained simply as a compromise of their member states.¹³⁰ This definition reflects the widely shared understanding of organizational autonomy in political science. Autonomy traditionally denotes the distance of an actor from immediate control by other actors. IO scholars conceive of the autonomy of international bureaucracies and other intra-organizational agents in terms of their ability to act independently from member state governments¹³¹ or to create undesired agency slack.¹³² Administrative science scholars argue that organizational autonomy involves a shift of decision-making capacity from external actors to a corporate body, which reduces the extent of ex ante instructions and regulations.¹³³ Hence, organizational autonomy reflects the ability of IOs to develop distinct organizational rationales that shape and affect organizational decisions.

Although organizational autonomy may arise from IO secretariats or other intra-organizational agents, it can also originate from institutional rules and procedures that govern organizational decision processes. These opportunity structures shape and constrain action within intra-organizational decision processes and may give rise to distinct organizational rationales, which affect IO decision-making and resulting IO decisions.

Simple formal decision procedures affect the aggregation of member state preferences. Member-dominated IOs adopt decisions either by majority voting or by consensus/unanimity. Even the most basic decision rules create some organizational influence on collective decisions because no decision rule can fully represent the constellation of power and interests prevailing outside the IO. When determining decision rules, members face the ‘problem of organization’,¹³⁴ that is, an awkward trade-off between control and flexibility.¹³⁵ Unanimity protects them from undesired organizational decisions, but risks that pooled resources remain unused. Majority voting or delegating decisions to intra-organizational agents flexibilize organizational activities, but involve the risk that pooled resources are used in undesired ways.

Substantive rules and complex procedures introduce systematic bias into organizational decision-making and create distinct organizational rationales. At least three sources are relevant for member-dominated IOs. First, IOs are goal-oriented organizations.¹³⁶ Member states are unlikely to vest them with action capability without some specification of purpose.¹³⁷ Accordingly, IOs embody a clear idea of how world affairs should be changed, which is a prerequisite for purposive action.¹³⁸ Purpose is reflected in deeply engrained fundamental organizational norms that introduce a selection bias into organizational decision processes.¹³⁹ For instance, the WTO has the purpose of promoting a liberal world trade system; and the climate change regime aims at stabilizing the global climate. IOs privilege organizational action promoting these goals and discourage adverse action. Second, IOs are heavily path-dependent.¹⁴⁰ Organizational path dependence occurs if preceding organizational decisions shape subsequent ones. Past decisions lock in previous agreement and stabilize collective choice in indeterminate decision situations with multiple equilibria.¹⁴¹ Specified criteria guide subsequent organizational decision processes and mitigate the risk of arbitrary decisions. Hence, IOs ‘generate a history of judgements that is on record’¹⁴², which shapes future decisions. Third, IO decisions frequently arise from multi-stage decision processes, in which two or more IO bodies collaborate.¹⁴³ Most IOs comprise specialized committees that produce scientific, technical, or legal advice and prepare decision proposals.¹⁴⁴ These committees introduce expertise into the decision process and confront final decision-making bodies with expert-based recommendations. The more organizational decisions are systematically influenced by these factors, the more they deviate from aggregated member state preferences.

A stylized example demonstrates that decision criteria might dissociate organizational decisions from member state preferences, thus reflecting a distinct organizational rationale. It plays an important role in philosophical thinking about group agency.¹⁴⁵ Suppose that an IO body with three members deciding on a new policy measure on climate change shall adopt the measure if it finds, first, that the measure is technically feasible and, second, that it is likely to be effective. Now imagine that the three members conceive of these issues as follows:

Two of the three members reject either technical feasibility or effectiveness and therefore prefer not to adopt the proposed measure. Accordingly, the group would not adopt the measure if the organizational decision were based on preference aggregation (right column). However, two of the three members appraise the measure as technically feasible and two of them consider it effective. Consequently, the group would adopt the measure if the decision were based on the two premises (bottom row). Accordingly, criteria-based decisions may be fully dissociated from members' individual preferences, although they originate entirely from their input. So, ‘knowing what the group members individually think about some proposition does not generally tell us how the group as a whole adjudicates that proposition’.¹⁴⁶

Why should rational actors accept such potentially far-reaching IO autonomy? After all, member states are likely to establish member-dominated IOs and member-dominated decision processes to retain control of IO activities and avoid sovereignty losses arising from delegation to intra-organizational agents. Partly, IO autonomy emerges as an inevitable side effect of the collective choice to establish an IO to manage a desired cooperation project. Whether organizational decisions are made by unanimity, majority, or delegation, actors always create some organizational autonomy. Likewise, the lasting nature of IOs inevitably creates some path-dependence, so that previous organizational decisions shape subsequent ones. Yet, IO autonomy may also be deliberately established. It may strengthen IO capacity to overcome stalemate among member states, reinforce mutual commitment to agreement, avoid arbitrary decisions, and introduce expertise to the decision process. Generally, rational actors can accept IO autonomy, as long as they consider IO membership as more beneficial than non-membership. This does not mean that IO members appreciate all IO decisions alike, nor that they comply with all IO decisions.

While most IOs, whether bureaucratic or member-dominated, gain some autonomy, the nature, extend, and effects of their autonomy varies greatly. To assess IO autonomy, we must investigate the organizational influence on collective decisions. For example, the UNSC has developed practices of humanitarian intervention because ‘every Council resolution that invokes a “threat to international peace and security” helps to define that phrase’ and ‘acts as a kind of informal precedent’¹⁴⁷ for future cases. UNSC sanctions committees tend to develop consistent decisions based on precedents and on the general rules of UNSC sanctions resolutions.¹⁴⁸ Likewise, many multilateral environmental institutions, such as those managing the Convention on International Trade in Endangered Species of Flora and Fauna¹⁴⁹ and the Montreal Protocol¹⁵⁰ comprise criteria-based decision processes that avoid straightforward aggregation of member state preferences and create distinct organizational rationales. Even the G-20 Economic Summit has established indicative guidelines to direct the ‘mutual assessment process’ identifying economic imbalances that constrain member states in subsequent decision situations.¹⁵¹

As a result, member-dominated IOs can gain autonomy that influences organizational decisions and creates a specific organizational rationale. Organizational autonomy arises from rules and procedures and thus ‘supervenes’ on member states' (past) decisions. While simple formal decision rules, whether unanimity or majority voting, affect the aggregation of member state preferences, substantive rules and complex procedures introduce a systematic bias into organizational decision processes. Fundamental organizational norms reflect organizational purpose and privilege compatible claims and proposals, while discouraging incompatible ones. Every decision must coherently fit into an ever growing set of previous decisions. Complex decision procedures involving several IO bodies confront decision-makers with organizationally produced expert perspectives and create some reflective capacity of IOs. All these procedural constraints introduce a systematic selection bias into organizational decision processes and create distinct organizational rationales. They dissociate organizational decisions gradually from member state preferences and produce an emergent ‘collective rationality’.¹⁵² While organizational culture is widely recognized for IO secretariats, it is also highly relevant for member-dominated IOs. For instance, the ‘shareholder norms’ of the Global Environment Facility contrast with the more egalitarian norms of the UN system,¹⁵³ while the World Bank and the UN have distinct perspectives on development policy,¹⁵⁴ although the memberships of these IOs are largely identical. Moreover, member-dominated IOs may ‘learn’ by collectively evaluating past action and adjusting institutional structures accordingly.¹⁵⁵ While IO secretariats are not a prerequisite for these effects, they can reinforce them. They tend to invoke organizational purpose, valid procedures, decision criteria, or precedents, and further dissociate IO decision-making from simple aggregation of member state preferences

Our conception locates the sources of IO autonomy in the institutional structures that shape and constrain member state activities and organizational decisions. Once again, it differs fundamentally from the IO-as-bureaucracy perspective and from the principal–agent model, which locate IO autonomy exclusively in the activities of intra-organizational agents. Moreover, our conception complements the narrow perspective of political authority scholars on formal decision procedures¹⁵⁶ with a set of even more important sources of organizational autonomy arising from substantive rules and complex procedures.

Conclusion

The corporate agency perspective developed in this article elucidates how IOs become actors in their own right even if the member states adopt all organizational decisions. It demonstrates that agency is not an institutional property of IOs or their secretariats, but arises from the authority relationship of an IO with its constituency. IOs gain the ability to actively influence world politics if, and to the degree that, they control governance resources and are authorized to use them for organizational purposes. They gain autonomy if, and to the degree that, they exert influence on organizational decisions. IO action capability and autonomy reflect different forms of sovereignty loss by the member states. The former requires that states sacrifice unilateral control of governance resources, while the latter relaxes the grip of member states on organizational decisions. A minimum of both action capability and autonomy is indispensable for the emergence of corporate agency. In the absence of action capability, IOs could not act in their own right (while they might still serve as coordinating forums for member states). In the absence of a minimum of autonomy, they would lack any influence on organizational decisions and could not act purposively.

Nothing in this conception is mysterious. IO agency emerges as a collective phenomenon entirely from activities of lower-level actors, especially the member states. The constituent actors pool governance resources and authorize an IO to employ these resources based on decisions of IO bodies, which operate according to jointly agreed decision-making procedures. Although IO agency arises entirely from activities of lower-level actors, it reflects an organizational quality that cannot be reduced to a particular constellation of constituent actors and their preferences.

While the corporate agency approach is fully compatible with existing actor-oriented IR institutional theory, it departs significantly from prevailing conceptions of IO agency. It coherently integrates conceptions of the rational design of IOs and the recent debate on international political authority with sociological and philosophical perspectives on organizations and corporate agency. In contrast to established IO agency theory, it does not juxtapose member states as principals and IOs as bureaucracies but demonstrates that IO agency may arise exclusively from interaction among member states. This does not trivialize the role of secretariats or other intra-organizational agents, which may increase both IO autonomy (e.g. setting the agenda of IO decision processes) and IO action capability (e.g. exercising governance rights such as dispute resolution).

The current focus of IO theory on intra-organizational agents grossly underestimates the concentration of governance power in some member-dominated IOs. If IOs are authorized to use governance resources, they become important actors in world politics, irrespective of whether they comprise extensive intra-organizational agents, having in mind that this does not ensure governance success. It matters whether states merely coordinate their own actions or centralize governance resources in an IO. The power of IOs to influence world politics depends on the nature and extent of governance resources, which they are authorized to employ, irrespective of whether decisions are made by the member states collectively or involve delegation to intra-organizational agents.

Likewise, the current focus of IO theory on intra-organizational agents underestimates the tendency of IOs to develop varying degrees of organizational autonomy. In member-dominated IOs and member-dominated IO decision processes, rules, and procedures shape organizational decisions that may depart, more or less extensively, from aggregated member state preferences. All organizational decision procedures are likely to influence IO decisions. Substantive guidelines and relatively simple procedural arrangements may fundamentally dissociate organizational decisions from a logic of preference aggregation and create distinct organizational rationales.

Our theory of corporate agency offers new perspectives for empirical research on IO agency. Its two dimensions of action capability and autonomy are open to empirical investigation. They are variable and differ across IOs, while sources of IO autonomy are likely to differ even across decision processes within IOs. The theoretical focus on corporate agency arising from member-dominated IOs and member-dominated processes in IOs complements the existing focus of IR theory on IO agency arising from the activities of agents. Thus, the theory promises to provide a more comprehensive picture of IO agency. Intra-organizational agents, like secretariats, courts, or expert groups, constitute specialized sub-systems within complex IO decision processes, much like committees composed of member-states. Many of their sources of influence, such as rule-based action, insistence on organizational purpose, or expertise, are also present in member-dominated arrangements. Awareness of the mechanisms of emerging IO agency from institutionalized interaction among member states allows accounting for the hybrid nature of most existing IOs that comprise, in various combinations, both activities of agents and member dominated areas of activity.