The Logic of Anarchy: Neorealism to Structural Realism

Barry Buzan, Charles Jones and Richard Little

New York

Columbia University Press

1993

6. Structural Realism and the Agent-Structure Debate

 

It is necessary first to examine in more detail some of the theoretical and methodological problems associated with system transformation. At first sight, the Neorealists seem unlikely to throw any light on these problems. As already noted, Neorealists such as Waltz presuppose that there has been no transformation in the structure of the international system. Indeed, the Neorealist theory of international politics is designed to explain why the anarchic structure of the international system has persisted throughout world history. Yet paradoxically, an investigation of the Neorealist position, and Waltz's arguments in particular, proves to be surprisingly revealing. Although at first it was assumed that Waltz's views on the theoretical and methodological issues associated with system transformation were unproblematic, closer investigation shows that this is not the case.

It is possible, however, to open up the idea of system transformation to theoretical investigation by drawing on the literature of the agent-structure debate. Although this debate has always been a central feature of theoretical and methodological discussions in most social sciences, the attempt to look at its implications for International Relations has been relatively recent. Waltz, for example, makes reference to both structure and agents in TIP, but he does not discuss the relationship between them, nor does he conceive of their conjunction as a problem. Nevertheless, Waltz does unwittingly provide a key that can help to resolve the agent-structure debate in International Relations, and lead the way to a better understanding of system transformation.

 

Background to the Debate

While the idea of structure is, of course, a familiar if problematic concept, ideas about agents and agency have received much less attention; moreover, the concept of agency, or, at any rate, the terminology, is much less familiar in the discipline of International Relations. Yet once attention is focused on the concept of agency, it quickly becomes apparent that structure and agent are inextricably linked.

Agency is defined in Webster's New International Dictionary as the "faculty or state of acting or exerting power." This is broadly what the term has come to mean in social theory. Although there have been attempts to restrict the use of power to human agency, these have not succeeded. As a consequence, power has been analyzed in terms of both structure and agency. The resulting conceptions of power, however, are very different. Structural explanations often turn human agents into lifeless puppets whose behavior is regulated by impersonal social forces. Explanations couched in terms of agency, by contrast, generally presuppose that it is human beings who control events in a social system. It is often suggested that an agent-based theory must embrace the intentions, the beliefs, the desires and the goals of the agents who act in the system. But Onuf (1989) has argued that a sharp distinction must be drawn between rational choice theories and hermeneutic or phenomenological theories of agents. The former type of theory presupposes that individuals are goal-oriented creatures, and that, because individuals strive to be rational, once these goals have been identified behavior becomes readily explicable and predictable. The latter type of theory, on the other hand, assumes that human beings are motivated by complex beliefs; analysts need to empathize with these beliefs before human actions can be comprehended and interpreted. Both types of theory are premised on the idea of the agent, but they are developed on the basis of very different epistemological perspectives. Onuf goes on to suggest that the study of International Relations has "no name for, or experience with" the latter type of theory (1989:56). However, the sharp distinction between these two approaches needs to be modified if the agent-structure problem is to be resolved.

The essence of the problem associated with structure and agency is quickly and easily expressed. It arises because explanations in the social sciences so frequently operate at one of two extremes. At one extreme, human beings are seen to be free agents with the power to maintain or transform the social systems in which they operate. At the other extreme, it is assumed that human beings are caught in the grip of social structures which they did not create and over which they have no control. The problem of structure and agency exists because of the failure to find any way of synthesizing these two extreme positions. Although familiar throughout the social sciences, the problem has only recently been introduced into discussions of international politics. When reference is made to the agent-structure problem in the international context, attention is almost always drawn to Waltz, who is seen to provide a perfect example of an analyst working in a purely structural mold. Critics argue that his structurally deterministic approach fails not only to leave room for human agency but also even to acknowledge the problematic nature of the relationship between structure and agent.

The accusation that Waltz has failed to accommodate the agent-structure problem does less than justice to the subtlety of his position. Indeed, there is a link between the approach adopted by Waltz and a solution to the problem of structure and agency alighted upon recently by a number of social theorists, in particular, Archer (1985, 1988), Giddens (1979), Taylor (1989), and Wendt (1987).

In one of the most extensive discussions of the agent-structure problem, Archer has suggested that it is "part and parcel of daily experience to feel both free and enchained, capable of shaping our own future and yet confronted by towering, seemingly impersonal constraints." It follows, then, that the methodological task of integrating structure and agency brings the social theorist face to face with "the most pressing social problem of the human condition" (1988:x). Yet despite the apparent universality of the problem created by the tension between structure and agency, social scientists have tended to shy away from it and have been drawn instead toward either agent-based explanations, which stress the capacity of individuals to mold the world they live in, or to structure-based explanations, which highlight the existence of unyielding social forces that shape our daily round. In the context of this bifurcation, Waltz has been neatly pigeonholed as a structuralist. Yet, with the benefit of hindsight, it can be seen that he is in the vanguard of the growing body of theorists who have reacted against both agent and structure based explanations of social behavior.

 

Problems with Structural Accounts

It is now widely accepted that any social theory must be able to generate explanations that take account of both structure and agency. One of the major problems with structural explanations, according to Giddens, is that they deny a role for the agent, thereby reducing human beings to "cultural dopes" of "stunning mediocrity"(1979:52). In structural explanations, human beings are moved around like pawns on a chessboard, by abstract, nonobservable forces that take no account of individual desires or beliefs. But as Grafstein asks, exactly how do "abstract entities constrain flesh-hued individuals?" (1988). In structural accounts, no answer is given to this question. It has been argued, for example, that Marxists have all too often operated on the basis of a structural theory of history which lacked any appreciation of the people who made this history. Marxists are increasingly coming to recognize that such an approach is inadequate and that it is necessary to ensure that any assessment of state behavior is "consistent with individual rationalities" (Przeworski 1985:383). The structural explanations of non-Marxists have also come under attack. Skocpol's (1979) highly regarded structural theory of revolution, for example, has been critically assessed by Taylor (1989). Skocpol argues that revolutions take place only when certain structural preconditions exist. It is then argued that these structural features of the system cannot be accounted for in agency terms. Taylor takes one of these preconditions, the solidarity of peasant communities, and endeavors to reveal how the solidarity is amenable to an agency based explanation. He argues that in prerevolutionary France and Russia, the state deliberately introduced measures designed to increase the solidarity of the peasantry. The purpose was either to undercut the influence of regional lords or to increase the revenue of the state. It follows that it is possible to show that Skocpol's structural prerequisites are, in fact, the outcome of rational and intentional actions.

 

Problems with Agency Accounts

But even if this argument is accepted, it remains the case, as Archer asserts so forcefully, that we are constantly aware that we are not free agents. There is, as a consequence, extreme skepticism over attempts to develop a theory of action expressed only in terms of the intentions, goals, desires, and beliefs of individual agents. Three factors negate a purely individualistic theory of action.

First, the outcome of many actions taken by one individual tends to be contingent upon action taken by another. Such situations are characterized in terms of strategic interaction. To take a simple example, if a motorist drives through a crossroad without stopping, then the outcome is determined by the intersection of this action with the behavior of the drivers on the other road. Levi has argued that one of the reasons why game theory has become so popular as a methodology in social science is because it reveals in such straightforward and remarkably clear terms "how the choice of one individual or group of individuals affects the choice of others" (1988:63). A game matrix describes how the choices of two individuals intersect, as in the case of the drivers coming to an intersection, and it illuminates how the structure of the situation they find themselves in constrains their choice of action. The matrix reveals that no rational agent operating in a structure where two decisions arrived at independently intersect can fail to take the possible actions of the other agent into account.

Second, personal experience confirms that individual action is conditioned by social institutions, some of which may even be extremely unpopular. Individuals may stand when the national anthem is being sung, for example, not out of any sense of patriotism, but simply because everyone else is standing. The example, of course, raises questions about how such institutions come into existence (Grafstein:1988). It is generally accepted that institutions are established in the first instance in an attempt to facilitate social intercourse. Institutions can help to alleviate the problems created by strategic interaction. Traffic lights can be placed at a crossroad in order to constrain the behavior of drivers, but in such a fashion as to be mutually beneficial. The traffic lights modify the existing structure. But it is also the case that social institutions can remain in place, continuing to constrain the behavior of individuals, long after they have ceased to serve their original purpose.

Third, the actions of social agents often have unintended consequences. Giddens notes that "the consequences of action chronically escape the initiators' intentions in processes of objectification" (1979:44). In a conflict situation, for example, one party may hold a parade in order to boost internal morale. An unintended consequence could be to inflame the opinions of the opposition. By the same token, a social institution established at one point in time for perfectly explicable reasons, may remain in force long after the initial reasons have disappeared. Social customs, which so often perform the unintended function of distinguishing insiders from outsiders, can frequently come about in this way.

 

Resolving the Agent-Structure Problem

The importance attached to the idea of institutional and structural constraints has encouraged the conclusion that any social theory needs to accommodate both agency and structure. In one of the best known attempts to tackle this problem, Giddens asserts that there is a dialectical relationship between structure and agency. Very far from being cultural dopes, pushed around by impersonal social structures, Giddens depicts social agents as knowledgeable and reflexive, having not only a sophisticated view of the world and how it is structured, but also the ability to monitor their actions in the light of this knowledge. As a consequence, social agents are constantly performing actions, often intentionally, but sometimes unintentionally, which ensure that social structures are reproduced.

So, for example, there is a complex set of social institutions that shape the behavior of motorists. These structural constraints are observed, not because of the sanctions imposed on deviants, but because of the reflexive ability to see what would happen in situations of strategic interaction if the structures did not exist. It follows, as Wendt notes, that "social structures are only instantiated by the practices of agents." In other words, a structure acts as a constraint only because agents choose to be constrained; if they were not constrained, there would be no structures. By the same token, social structures are depicted as being "inseparable from the reasons and self-understandings that agents bring to their actions" [Wendt 1987:359].

It is being argued, for example, that motorists stop at red lights, not because they fear that a violation of the rule would lead to prosecution, but because they appreciate the reasoning behind the decision to locate traffic lights in a particular position. So it follows that a structure constitutes a structure only because of the behavior of the agent, which in turn is intimately bound up with knowledge of the structure. As Giddens puts it, "structure is both medium and outcome of the reproduction of practices." He refers to his method as the duality of structure or structuration which he argues is an appropriate basis for social explanations because of the essential "recursiveness" of social life (1979:93). It is because of recursiveness that it is not possible to draw a rigid distinction between rational actor and hermeneutic theories.

The term "recursiveness" is not familiar in the field of International Relations. Yet it is now making its way into the literature. Onuf defines recursion as "the propensity of knowledgeable agents to refer to their own or others" past and anticipated actions in deciding how to act" (1989:62). Giddens who has discussed the idea and implications of recursion extensively, refers to the phenomenon as "reflexive self-regulation" (1979:78). But, in fact, this graphic phrase simply provides a more vivid way of expressing Schelling's more mundane reference to "interdependent decision making" which lies at the root of his conception of strategy (Schelling 1960: ch. 4). Schelling's familiar discussion of how a husband and wife, separated in a department store, have to resort to interdependent decision-making in order to make contact again can just as easily be discussed in terms of reflexive self-regulation. Recursion, therefore, may represent a new term, but it identifies a familiar phenomenon. Indeed, it can be argued further that Schelling had a handle on the "agent-structure problem" nearly thirty years before it was self-consciously identified by social theorists and then extended into the field of international relations. And the conception is certainly not unique to Schelling.

The concept takes on a new dimension when social theorists like Giddens have extended the idea of a recursive process to describe the mutual constitution of agents and structures. Giddens insists that although the activities associated with any habitual social practice will not have been introduced by any given agent, nevertheless, by engaging in the practice, the agent will "reproduce the conditions that make these activities possible" (Giddens 1984:2). By analogy, when people walk across a field, they may unintentionally create a path. Others subsequently follow and in doing so "reproduce" the path (See Little 1985). The idea of recursion gives this process a social dimension. When Americans motor in Britain, for example, they automatically start to drive on the left-hand side of the road because they anticipate that other drivers will be doing the same thing. Contemporary social theorists interested in the agent-structure problem have extended the line of analysis pioneered by Schelling, and have now linked it to the idea of recursion, to develop a theory that explains the reproduction of the structures which underpin any social system.

The solution to the problem of structure and agency presented by Giddens has not been greeted with universal approval. It has been subjected to particular criticism by social theorists who have an interest in social change and transformation. By defining structure and agency as mutually constitutive, and a product of recursiveness, Taylor argues that Giddens has made it impossible to "unravel the causal interplay between agent and structure over time and account for social change" (1989:118). Taylor wants to argue that social action can, over time, erode the existence of a social structure and replace it with another. So, for example, if one road on an intersection with traffic lights becomes more busy than the other, because of recursion or "reflexive self-regulation," there will be an increasing tendency to "jump" the lights. At that juncture, the lights become a source of danger and either the timing of the lights will be changed or they will be replaced by an alternative structure.

Archer has also disputed the approach adopted by Giddens. She acknowledges that Giddens is aware of the potential for social change, but she insists that Giddens can only depict social change in terms of a gestalt switch. Making use of the road intersection example, again, Giddens, it is argued can depict different structural solutions to the problem created in situations of strategic interaction--say, traffic lights or an underpass--but what he cannot do is analyze the process whereby one structure gives way to an other. Archer insists that an adequate solution to the structure-agency problem must make provision for an analyst being able to identify different degrees of freedom for an agent and to specify different levels of constraint imposed by a structure. It must be possible to observe the nature of the structure and the behavior of the agent changing gradually over time. As Taylor notes "to conflate structure and action (as Giddens does) is to rule out from the start the possibility of explaining change in terms of their interaction over time" (1989,149). These analysts are clearly seeking a way to understand the relationship between continuity and transformation in a system.

To achieve this objective, Archer insists that it is essential to draw an analytical distinction between structure and agent (while recognizing that it may be correct to describe them, empirically, as mutually constitutive). By doing this, it becomes possible to examine them independently, and in this way explore how the activity of social agents can, over time, bring about a transformation in social structure. Such a formulation, however, does not seem to violate the spirit of Giddens, who is unquestionably interested in how social systems develop along the dimensions of time and space. Central to his whole enterprise is the desire to identify how and why the nature of sociopolitical systems have changed so dramatically over time. So, for example, he focuses on the way that the distance between societies, expressed in terms of the conjunction of time and space, has been so very substantially compressed in the modern world. He has also increasingly acknowledged that the nature of society has been constantly influenced by its international setting. As a consequence, he has begun to develop a framework that allows him to explore the interaction between the structure of the state and the structure of the international system. This enterprise overlaps in some important ways with the approach of Section 1, where the compression of time and space in the international system, referred to there as interaction capacity, is seen to be a property of the systemic level of analysis.

 

Waltz and the Agent-Structure Problem

The task of resolving the agent-structure problem is complicated at the international level because the agent is not an individual but an institution. As a consequence, attention is centered on the formulation of a deep structure based on the configuration between political units in the anarchic international system. When one examines the complex interaction between the structure of the state and the structure of the international system, one raises the agent-structure problem in an acute form.

Neorealists circumvent the problem by treating the state as a rational actor. So although Waltz accepts that human agents are engaged not only in the task of reproducing their own political unit, but also the system within which those units operate, he appears to bracket, and then ignore, the process at the unit level. Before turning to our more complex Structural Realist approach, it is useful to see what light the recent attempts to handle the agent-structure problem throws on Waltz's apparently straightforward formulation. Superficially, Waltz appears to exemplify the work of a theorist who has resolutely ignored the problem of structure and agency; it is nevertheless possible to show that Waltz has independently alighted on a solution to the problem that coincides with the position now being adopted in other areas of the social sciences.

At first sight, it may appear that there is something perverse about this line of argument because it runs so counter to the established assessment of Waltz. TIP has been very widely discussed and it is almost invariably held, even by its critics, to be providing an unambiguously clear exposition of a structural approach to the analysis of international politics. The critics may go on to attack structuralism but they never question that Waltz is a true and faithful exponent of this methodological stance. Waltz has also been seen to play a major role in the upsurge of interest in a structural approach to the study of international relations. The growing interest has been so rapid and extensive that it has even been argued that structuralism now represents the dominant mode of analysis in international relations (Zacher 1987:173). Such a claim is debatable, but it is unquestionably the case that structuralism has become a significant feature in the discipline and that Waltz is considered to be the chief architect

It is not difficult to see why Waltz has been so closely associated with structuralism. The central theme of TIP is that there has been a persistent failure by theorists in the discipline to ascribe an independent role to the structure of the international system. It is presupposed, not always intentionally, according to Waltz, that the behavior of states (variously referred to as units, actors, and agents) can be explained only in terms of their internal properties. In other words, Waltz is primarily interested in opposing a unit-based theory of international politics. Such a theory is considered to be reductionist. But reductionism is not seen, necessarily, to be a methodological sin. Waltz accepts that the systems examined by physical scientists are amenable to a reductionist form of analysis. But he insists that the complex systems examined by social scientists do not fall into this category. In social systems, he argues, agents are constrained by the structure of the systems in which they operate and explanations need to take account of these structures. It is for this reason that Waltz has been designated as a structuralist.

The inaccuracy of this description quickly becomes apparent when Waltz's methodology is explored from the perspective of an analyst attempting to overcome the problem of structure and agency. Wendt, for example, has drawn on the idea of structuration. His line of analysis points him in the direction of the argument developed by Archer and Taylor, with the result that Wendt also asserts that both the structure of the international system and the agents operating within it must be given an independent ontological status. He adopts the epistemological posture that invisible structures have just as tangible an existence as the individual agents constrained by them. This line of argument is related to the philosophical posture identified as scientific realism. The link between structuration theory and scientific realism is now coming under close scrutiny. (Layder:1990). It will not be examined here. The whole debate about structure and agency takes on an additional complexity in the international context because states are generally identified as the principal agents. But this equation is simply a form of shorthand. Reference is actually being made to the human agents who represent the state. The state, like the international system, needs to be conceived of as a set of structures. In the international context, therefore, agents are constrained by the structure of the state as well as the structure of the international system. The state, as a consequence, is no more visible than the international system. Wendt insists that both must be accorded equal ontological status and he asserts that it is not possible to develop a coherent and effective theory of international politics without treating both the system and the state as problematic. A theory of international politics must be able to explain the emergence and development of both the state and the system. Wendt argues that this is not "mindless synthesis" because it "forces us to rethink fundamental properties of (state) agents and system structures" (1987:339).

Working from this perspective, Wendt is able to see quite clearly that Waltz cannot be classified as a structuralist. On the contrary, he argues that Waltz and other Neorealists have managed to develop a "conception of the agent-structure relationship in international relations which recognizes the causal role of both state agents and system structures" (1987:341). But he goes on to argue that the Neorealists develop an "individualist definition of the structure of the international system as reducible to the properties of states" and that, as a consequence, the approach must be considered "decidedly state-or agent-centric" (1987:341-42). Ironically, from this perspective Waltz certainly cannot be considered to be a wholehearted structuralist; and while Wendt accepts that Waltz does provide a solution to the structure-agency problem, he also makes clear that the solution is deeply flawed, not because of its emphasis on structure but because of its bias toward agency. Wendt contrasts Waltz unfavorably with the world systems approach adopted by Wallerstein, although Wallerstein is also seen to provide an inadequate solution. Wendt's assessment of Waltz runs directly counter to the one developed by Hollis and Smith (1990), who depict Waltz as an unreformed structuralist. In a subsequent debate between Wendt (1991) and Hollis and Smith (1991), neither side is willing to accept that their divergent interpretations could be reconciled by accepting that Waltz's position is compatible with the idea of structuration. To justify this interpretation, it is necessary to demonstrate that Waltz has a theory of the state as well as a theory of the international system and that, in accordance with the structurationalist methodology, the one instantiates the other. Waltz's theory of the state is examined in the next chapter.