|
|
|
|
The Logic of Anarchy: Neorealism to Structural Realism
Barry Buzan, Charles Jones and Richard Little
New York
1993
3. System, Structure, and Units
Neorealist theory is rooted in the basic ideas of system, structure and units. Any attempt to advance from it needs to start with these concepts, clarifying their definitions, and focusing on how structure and units relate to each other and to the broader idea of system that encompasses them.
System
In general terms, and also in terms of international relations, "system" refers to a group of parts or units whose interactions are significant enough to justify seeing them in some sense as a coherent set. A group of states forms an international system when "the behavior of each is a necessary factor in the calculations of the others." (Bull and Watson 1984:1) A system therefore comprises units, interactions, and structure. Interaction is crucial to the concept of system, for without it, the term system has no meaning, a point that will be more fully developed in chapter 4 and Section II. A system may be identified specifically in terms of the function of the set as a whole (as in a central heating system), or more generally by an enclosing boundary that contains the whole universe of a specified set of units (as in the international system). The functional approach, which is that of general systems theory, can be rejected as a model for the international system on the grounds that it is too highly structured (Waltz 1979:59). The international system reflects no conscious design, and has no specified purpose or function. But it is a major artifact of human evolution, and the manner of its working is an important element in defining the human condition.
The term international system has two senses, and the resultant ambiguity is a common source of confusion. Its first, and most specific, meaning refers to the system of states, and reflects the puzzling, but firmly established use of nation as a synonym for state. This usage confines the idea of system to states (singly as units, collectively as the embodiment of structure) and their interactions. One could more accurately refer to this conception as the interstate system. Its second, and most general meaning refers to the totality of human interaction on the planet, and incorporates a range of units varying from individuals, through firms, nations and a great variety of other nongovernmental organizations or entities, to states. This should more properly be called the interhuman system, or somesuch, because the term international, with its suggestion of interstate relations, contains a strong bias toward an exclusively political interpretation of the total system of humanity.
In what follows, the tension between the interstate and interhuman views of the system plays an important role. The simplicity of Waltz's theory depended in good measure on his use of highly restrictive definitions. His trailblazing cleared a basic path, and allows others to devote more energy to exploring the byways of less restrictive assumptions. Much of the logic of Structural Realism results from shifting away from a strict interstate understanding of the international system, and toward an interhuman one. To avoid proliferating terms, we will use "international system," but draw attention to points where the distinction between its two senses is important to the discussion.
System Sectors
It is a common tactic when discussing systems to seek greater specificity by qualifying the identity of systems in terms of particular sectors of activity within them (as in "the international economic system," or "the international political system"). This is the approach that Waltz adopts (Waltz 1979:79). In doing so he places himself firmly in the tradition of Classical Realism that asserts the autonomy of the political sphere. (Morgenthau 1978:5) When we refer to more specific sectors of the international system these will also be identified as the international political, military, societal, economic, or whatever, system. This approach is commonly used in academic discussions of the international system, but as a practice, has not itself been much discussed.
The partial systems identified by sectors are not subsystems in the normal sense of a subset of units located within a larger set, and containing fewer units than the whole. Instead, they are views of the whole system through some selective lens that highlights one particular aspect of the relationship and interaction among all of its constituent units. The metaphor of a lens is quite accurate (Manning 1962:2). In the physical world, one can look at an object using many different types of "lenses," ranging from the naked eye and telescopes, through infra-red sensors and radars, to X-ray machines and electron microscopes. In each case the lens is either sensitive to different types, or wavelengths, of energy (e.g., infra-red and X-ray), or else sensitive to the same type of energy in a different way (e.g., microscope and telescope). Thus even though the object observed remains the same (ignoring Heisenberg), different lenses highlight different aspects of its reality. The naked eye sees mostly exterior shape and color. The infra-red sees the pattern of heat. The X-ray sees the pattern of physical density. The electron microscope sees molecular structure. The function of sectors is the same as that of lenses: each one gives us a view of the whole that emphasizes some qualities, and de-emphasizes, or even hides completely, others.
The use of sectors/lenses has the advantage of highlighting, and therefore making easier to see, certain qualities of whatever is being observed. It is a way of unpacking the confusion of the whole. In analytical terms, it confines the scope of inquiry to more manageable proportions by reducing the number of variables in play. Thus the economist looks at human systems in terms that highlight wealth, and justify restrictive assumptions such as the motivation of behavior by the desire to maximize utility. The political Realist looks at the same systems in terms that highlight power, and justify restrictive assumptions such as the motivation of behavior by the desire to maximize power. The military strategist looks at the system in terms that highlight offensive and defensive capability, and justify restrictive assumptions such as the motivation of behavior by opportunistic calculations of coercive advantage. Each is looking at the whole, but seeing only one dimension of its reality. The obvious danger in this method is that the observer will begin to confuse the partial reality of the sector with the total reality of the whole. Sectoral blindness is an occupational hazard of all human specializations. It is a particular disease of an academia excessively blinkered by deeply institutionalized disciplinary boundaries.
In relation to systems analysis, the use of sectors raises two difficult questions. First, it introduces a definitional problem about how the different sectoral "systems" within the total system are to be distinguished from (and thereafter related to) each other. Waltz makes very clear from his title that he is attempting a sectoral theory of international politics. If theory is to be constructed within a sector then the definitional issue is inescapable. But how can one differentiate between the international political and economic systems? Ruggie's discussion of the intimate links between property relations and sovereignty illustrates the difficulty (in Keohane 1986:141-48). At root, the question here is whether a viable theory can be constructed within a sector, or whether to do so inevitably confuses the sector with the whole? In thinking about any given historical event, it is difficult to disentangle the economic, political, societal, and strategic threads that make up the whole. Because each sector is a partial view of the whole, sectors necessarily overlap and interweave. Sectors are not separated by clear boundaries like those between levels of analysis. Indeed, the whole spatial metaphor of separation, boundaries, and the like is simply inappropriate. This is because, as with lenses, the distinction between sectoral boundaries is to be found as much in the equipment of the observer as in the thing observed. By contrast, boundaries between levels of analysis are more wholly within the thing observed, serving to identify distinct elements of causality. We will consider this metaphor further in chapter 12.
Following the sectoral path to theory means assuming that sectors can be made distinct. This assumption raises a question: how do the vertical division of a system into sectors, and the horizontal division of the system into levels of analysis, interact? (See figure 3.1 below.) Do levels of analysis get defined in relation to the whole system, or only in relation to a particular sector? Generally, the definition of system in terms of units, structures, and interactions points to the validity of levels of analysis extending across all the sectors. But while the categories of levels may extend across all of the sectors, their contents may well differ. If sectors can be defined in distinct terms, it is probably by specifying the differences in their units and structures. Specificity, consistency, and clarity of usage are therefore extremely important in discussions involving system sectors.
It might be useful to construct the relationship between sectoral subdivisions and levels of analysis on the assumption that levels of analysis refer to horizontal subdivisions of the "field" of the international system, while sectors refer to vertical "subdivisions" of the same field, as sketched roughly in figure 3.1. It is important to understand here that sectors do not represent subdivisions of the whole system, but partial views of the whole.
This scheme enables us to consider levels of analysis either in terms[fjof the international system as a whole (by dissolving the sectoral distinctions), or in terms of specific sectoral subdivisions (by defining the levels in terms that are bounded by one or more sectoral subdivisions e.g., international political system, or international political economy). The question of what the complete universe of sectors might look like for the international system is beyond the scope of this discussion. The four listed certainly cover the main preoccupations of International Relations theorists. Waltz's theory is an attempt to define levels of analysis exclusively within the political sector. One needs to be aware that this confinement is a source of both the strengths and the limitations of Neorealism.
Levels of Analysis
The question of levels of analysis must arise in any attempt to understand behavior within systems because it defines the layers of the system within which explanations are sought. Each level must identify a major source of impact on behavior, and thus an explanation for events, that is distinct from other major sources. Like sectors, the idea of levels of analysis is an abstract construct. Because this is so, the definition of levels contains considerable scope for choice, which again makes clear specification and consistency of usage extremely important. The basic definition of system offers two obvious "top and bottom" candidates for levels of analysis: the structure, representing that collective element of the system which transcends the units, and the units themselves. This is the path that Waltz takes. The unit level of the system he defines as "the attributes and interactions of its parts" (Waltz 1979:18). The system level he defines "by the arrangement of the system's parts and by the principle of that arrangement" (Waltz 1979:80).
To understand why the boundary between these two levels in Neorealism has proved so troublesome to other analysts, one needs to explore deeper into its logic. In doing so, it is worth remembering Singer's conclusion to his assessment of the system and state levels of analysis: "it must be stressed that we have dealt here only with two of the more common orientations, and that many others are available and perhaps even more fruitful potentially than either of those selected here" (Singer 1961:90). In developing Structural Realism, one needs to keep an open mind not only about definitions of the system and unit levels, but also about what constitutes a level of analysis. Are the system and unit levels the only ones in play? (We will argue in chapter 4 that they are not.) Are they monolithic as sources of explanation, or can they be more finely divided into sublevels with distinct logics of their own? (The case for subdivision is made in this chapter.) Is the crude system/unit distinction the most efficient one for a theory of international relations, and how far can the separation of levels be sustained? (This issue is pursued in Section II.)
The Structure of The International Political System
If the idea of system is to represent more than just any selection of parts and their interactions, it must be accompanied by the notion of structure. For Waltz, structure means the ordering principle underlying the way in which the units relate to, and are functionally differentiated from, each other. In this context, "relate to" refers to how the parts are arranged or positioned in relation to each other, not to how they interact, which in his terms (but not ours, as will become clear in chapter 4) is a unit level phenomenon (Waltz 1979:80-81). His concern is only with the structure of the political sector of the international system. Two issues arise: (1) how is the structure of the international political system defined, and (2) once a suitable definition has been obtained, how does it relate to other definitions of structure (e.g., economic, societal, strategic) arising either from the other sectors that comprise the international system, or from a holistic view of the system?
These two questions drive in potentially contradictory directions. Seeking a definition for the structure of the international political system means accepting the Realist assumption that one can and should identify a distinctively political sector within the international system. Looking at how a political view of structure relates to notions of structure in other sectors of the international system may undermine the distinctively political emphasis of Neorealism by leading to broader definitions of structure, for example in terms of political economy. In this Section we begin by working with the Neorealist assumption about the primacy of the political sector. Section III explores the wider relationship between political and economic structures.
The problem of defining structure within the relatively narrow sectoral terms of politics is that the core defining feature of politics is itself a matter of controversy. Waltz does not attempt a formal, generalized definition of what the political sector covers, though there is no doubt that he sees the political sector of the international system as defined in terms of states. States are both the units of his international political system, and the foundation of its structure, a linkage that causes Wendt to accuse him of ontological reductionism (Wendt 1987:341). This rather crude delineation of the political sector sows the seeds of subsequent difficulties in his sparse definition of international political structure.
Politics does not have a neat, generally accepted, definition. The essence of it concerns the shaping of human behavior for the purpose of governing large groups of people. This shaping may be done both by the citizens themselves and by their leaders. If one takes government as the key to comprehending the political sector of the international system, then the theory and practice of government can be approached along three main paths, in terms of power, authority, and organization. Each of these approaches identifies a distinct emphasis in the management of collective human behavior: coercion in the case of power, legitimacy in the case of authority, and administration in the case of organization. In practice, however, none of these three can be made more than partly distinct from the others. Power and authority are conceptually distinct, but notoriously hard to disentangle in practice. Both normally depend on some element of organization. The process of government invariably involves a combination of all three. These different aspects of politics open up a variety of possible approaches to the definition of political structure. The problem is that each element taken individually has a logic and a set of applications that extends far beyond the reasonable limits of a strictly political domain. Power, for example, plays a big role in social and economic relations, as well as political ones. Only their combination for the purpose of governing produces a reasonably bounded notion of the political sector.
The Neorealist position is laid down by Waltz in his three-tier definition of political structure as: the organizing principle of the international system, plus the functional differentiation of units, plus the distribution of capabilities across units (Waltz 1979:ch. 5, esp., 100-101). This definition does not clearly enough confront the meaning of politics in relation to the sector of the international system that it addresses. The first two tiers of Waltz's definition are about government, albeit with a tightly restrictive focus on sovereignty. The organizing principle tier captures the distinction between systems with only one sovereignty, and systems with more than one (hierarchy and anarchy). The functional differentiation tier, by Waltz's argument (but not ours, as will be seen in the next part of this chapter and Section II), does not apply to anarchic political structures because all sovereign states are by definition like units. Thus in anarchic international political systems, government is strongly concentrated in the units, and only weakly, if at all, present at the system level.
The third tier, distribution of capabilities, is not about government, but focuses more narrowly on the power dimension of interstate politics, though still deriving directly from the sovereign state structure established by the first two tiers. This change is, by itself, grounds for suspicion in a supposedly integrated definition. Waltz's shift to a narrower approach in the third tier raises three issues. First, his focus on power heavily discounts the authority and organizational dimensions of international politics, an exclusion that seems to bear directly on the criticisms of Keohane and others who think that rules, regimes, and international institutions need to be brought into the definition of international political structure. Second, Waltz's formulation raises longstanding definitional problems about power and how to measure it. Third, his use of the distribution of capabilities across units as an element of structure opens the question of why he confines the principle of distribution to the single variable of capabilities aggregated into power.
The difficulties in Waltz's third tier take on added significance because they occur at the point of maximum controversy in his theory, namely the boundary between the unit and system levels. They are serious enough to require a separate consideration of the placement and nature of the boundary between the system and unit levels. Before taking up this problem, it is helpful to establish firm ground on either side of it. We will therefore consider next the first two tiers of Waltz's definition, which will collectively be referred to using Ruggie's term deep structure, (in Keohane 1986:35), after which we will look at the unit level.
Deep Structure
In dealing with deep structure it is helpful to fix a firm view of what defines the political sector of the international system. The discussion above pointed to governing as the core of what politics is about. In general, governing requires institutions that claim the right to be the makers and enforcers of laws and the executors of policy for a specified people and/or territory. These institutions must possess a sufficient combination of power, authority, and administrative capacity to put that claim into meaningful practice against rivals both within and outside their jurisdiction. The sovereign state is without doubt the hardest and clearest form of government. It is also by a long stretch the most dominant form of government in the modern international system.
But the state is not the only form that government can take. No iron law requires the functions of government to be undertaken only by sovereign entities (i.e., states). As Ruggie argues, there was government without sovereignty in the Medieval political system (in Keohane 1986:141-48). In the contemporary international system, non-sovereign bodies such as the European Community, and even more loosely the United Nations and its agencies, undertake some governmental, or at least government-like, functions. As the strictly governmental function of laws blurs into the rules and norms beloved of theorists of international society and international regimes, one's sectoral perspective begins to incorporate both the international political system and the international societal system (where the emphasis is on cultural and normative determinants of behavior). General awareness of the distinction between the international political and societal sectors is reasonably well established, most notably through Hedley Bull's emphasis on the common rules and norms of anarchical society (Bull 1977, esp. chs. 2,3; Bull and Watson 1984; Manning 1962; Mayall 1990; Wight 1977).
If government defines the makings of political structure, then one can stick with Waltz for at least the first two tiers of his definition (i.e., the organizing principle of the international system, plus the functional differentiation of units). Waltz's first two tiers identify an element of political structure that is deep in the sense of representing a basic pattern that is not only durable (on historical evidence), but also self-reproducing (in that the operation of the balance of power sustains the anarchic arrangement--and in Waltz's view also the like units). In Waltz's logic, these two tiers are so closely linked as to be nearly opposite sides of the same coin. Each tier offers two broad structural options clearly defined in terms of government. In the organizing principle tier, the options are anarchy (no central government over the units in the system) and hierarchy (central government over all the units). There is a possible ambiguity about anarchy in this formulation, since it could refer either to a system with no government at all (individual humans as the basic unit), or to a system with two or more governments controlling different parts of the system. In the international system the former condition never occurs, and is therefore of no practical interest to the theory: international anarchy always refers to a system with two or more governments.
In the functional differentiation of units tier, the options are that the units are similar or different in terms of the range of governmental functions they claim and perform. Similar effectively means sovereign, where all units define themselves as the highest authority in all matters of government over their specified territory and people, including the right to wield power independently. Different means that units claim sovereignty only over a limited range of functions, as in the Medieval system. Where claims are limited, they must logically produce difference in order for all of the political functions to be covered. A system in which all the units claimed the same limited functions would be incompletely governed, which is why similar points strongly to full sovereignty. These definitions are extremely sparse, delineating only the most basic political principles. They specifically exclude societal elements in which sovereignty is seen not just as a function of claim, but also as a function of mutual recognition between states. Because they are so basic, they leave a lot of room for variety within the scope of anarchic structure.
Taken together, the two tiers yield a 2 x 2 matrix as in figure 3.2. By Waltz's argument, however, two of these categories (numbers 1 and 4) are virtually empty because of the existence of strong, two-way interactions between organizing principle and functional differentiation of units. It is worth examining this argument in some detail, for the whole conception of deep structure (and therefore the definition of significant change) hangs upon it.
His argument runs as follows. In type 2, similar units and anarchy are opposite sides of the same coin. If all the units are subject to the survival logic of self-help, then the shoving and shaping pressures on them of life in the anarchy will push their range of governmental functions toward sovereignty. If all the units are sovereign, then the organizing principle amongst them must, by definition, be anarchic (Buzan 1991:ch. 4). Thus anarchy tends to generate like units, and like units, by pursuing sovereignty, generate anarchy.
Under anarchy, the logic of survival dictates that units are subject to the pressures of socialization and competition (Waltz 1979:76; Ikenberry and Kupchan 1990). In historical terms, one can easily see how both of these pressures work to produce homogeneity of unit type. Socialization works by the demonstration effect of the most successful units, which encourages others to copy them. In the postwar period, the former Soviet Union has been a model for some, the United States for others, and more recently, Japan has set the pace as a model of economic efficiency and prowess. The collapse of communism in 1989 not only removed one demonstration model from the system, but also illustrated the powerful socialization effect of more successful units as much of the ex-communist world struggled to adopt the forms and practices of the West. The effect of competition is rivetingly demonstrated by the period of European imperialism. The stronger European states simply took over weak units, eventually releasing them by decolonization only after they had been remade (often badly) into copies of European territorial states. Those Asian nations not colonized, such as Japan, Thailand, and China, had quickly to remake themselves in the European state mode in order to avoid becoming victims of European power.
Similarly, the Second World War allowed the victors to restructure in their own image the domestic politics of the losers. As Waltz puts it: "the close juxtaposition of states promotes their sameness through the disadvantages that arise from failure to conform to successful practices" (Waltz 1979:128). Quite so, and this logic can work as strongly for commercial relations among merchant states as for strategic relations among warrior ones. In this way, socialization and competition should push strongly away from any durable manifestation of category 4 (anarchy/different). In striving for survival and security, units will tend to converge on the model of the most successful and powerful of their number, thus moving the system toward type 2.
It must be pointed out that Waltz's logic on this point relies on the transfer from microeconomics into international politics of a very specific, and very partial characterization of competition. All the emphasis goes to the way in which competition leads to imitation--all family cars become the same style, all tomato soups adopt the same taste, and all states become centered on sovereignty, power, and security. This view discounts another side of the analogy from economic behavior, which is the search for market niches, where differentiation of function provides (temporary) refuge from the full pressure of competition. These two dynamics are not mutually exclusive, and taken together allow the dynamics of convergence to work simultaneously with those of differentiation. This view opens up the possibility of a type 4 deep structure in a way that Waltz's narrower logic does not allow, and we will make more of it in Sections II and III.
There is a matching logic in Waltz's theory that ties together hierarchic organizing principles and functionally differentiated units. In a hierarchic structure, the imperative for units to specialize is driven by the efficiency logic of the division of labor (Waltz 1979:106-7). If the organizing principle is hierarchy, as most clearly illustrated by looking at an individual state as a political system, then units within it are strongly pressured to make themselves functionally different. This is evident from any organizational chart of a national government or any bureaucracy. Indeed, there is virtually no sense in having a hierarchic structure unless the units have differentiated functions. Here the processes converge on type 3 (hierarchy/different), and away from type 1 (hierarchy/similar). The logic that pushes toward type 3 is clear if one starts from hierarchy, but is harder to illustrate if one starts from differentiated units. If units are functionally dissimilar in terms of government (i.e., claiming not full sovereignty, but only the right to govern over selected aspects of law and behavior), then their specialization must create interdependence in order to fulfill the basic requirements of governing. Because specialized units by definition do not fulfill the whole range of political functions, they must depend on each other for mutual support in relation to the political environment as a whole. Since order is a prime political value (Bull 1977:ch.4) security needs will drive differentiated units either to the collective safety of hierarchical structure, or back to self-help (and thus back to similarity, sovereignty, and anarchy). Perhaps the only strong historical case covering this point is the transformation from the Medieval to the modern system in Europe, where the choice was for anarchy, on which we will have more to say below.
Waltz's scheme identifies compelling linkages within a political system between organizing principle and differentiation of units by function. Among other things, this linkage points to an incipient theory of the state in Neorealism (on which more in Section II) that has been overlooked by some of its critics (Walker 1987:78). Yet within this logic of deep political structure Waltz takes what seems an unnecessarily extreme view in asserting that an anarchy must have like units, and that if units cease to be alike, then anarchy ends (Waltz 1979:93; in Keohane 1986:323; and interview). This rigid stance is not required by the logic of his theory, and does not follow from the essential defining condition of anarchy, which is the absence of central government. Its apparent source is Waltz's strict adherence to the "interstate" meaning of international system. If the international system is conceived purely as an interstate system, then non-sovereign, and thus non-like and non-state units simply fall outside the definition of international system. Because he puts too much emphasis on sovereign states, and too little on governing, Waltz's theory is simply unable to deal with type 4 systems (unlike units in anarchy). Waltz seems prepared to abandon a massive area of political reality in the international system in order to hang onto this definitional nicety. This seems to be a serious and unnecessary mistake, with pursuit of parsimony leading him to trade a small gain in definitional neatness for a large loss in the relevance of the theory to the international system.
Waltz's stance provokes Ruggie's justified riposte that the history of Medieval Europe provides a case where units that are not sovereign--and that therefore are differentiated by function--exist without having a hierarchical structure, and therefore in an anarchy (in Keohane 1986:141-48). At a very minimum, Ruggie's demonstration that type 4 has historical content supports Waltz's point that structures "do not determine behaviors and outcomes" and "may be successfully resisted." (in Keohane 1986:343) At a maximum, as has been hinted at above, and as will be seen in Section II, it points to a radical widening of the practical possibilities for the nature of units under anarchy.
But this point need not disturb Waltz's theoretical assertion that the logic of political structure in the international system will, over time (perhaps a long time, the theory gives no prediction about this), tend to bring organizing principles and differentiation of units into line. In interpreting his own theory too statically, Waltz seems to lose sight of his argument that structure shapes and shoves rather than determines (Waltz 1979:73-74). The power of his idea is nonetheless demonstrated by the current configuration of the international system, in which the superior power (in terms of competition) and attractiveness (in terms of socialization) of the sovereign state have virtually eliminated all other forms of government from the system. Since 1945, the system's units have become more functionally similar than ever before.
Within a Structural Realist framework, explanations for type 4 cases of functionally different units without a central government (as in Ruggie's argument, or in the more extreme views of contemporary transnational developments) can be sought in at least three ways. First, a manifestation of type 4 can be seen as a transition phase, in either direction, between anarchy and hierarchy. In the case of the Medieval system the transition could be placed historically between the long hierarchy of the Roman Empire and the long anarchy of the modern European state system. The differentiation among Ruggie's Medieval units may have reflected the collapsed hierarchy of the Roman Empire that preceded them, but in the self-help context of anarchy, the imperative governing their behavior was more "take care of yourself," than "specialize." Over the centuries, self-help security logic thus pushed the direction of change toward like units in the form of sovereign states.
Second, it can be explained as a deviation from structural norms (the anarchy-sovereignty and hierarchy-differentiation imperatives) due to the intervention of countervailing factors on the unit level. In the Medieval case these factors might include a shortage of political resources at the unit level. Small, weak units scarcely generated enough interaction in the system to drive the homogenizing logic of socialization and competition with any urgency. This question of how much, and what kind of interaction is required in order for a system to operate in Neorealist terms is vital to the theory, but as yet little discussed, and will be taken up in detail in chapter 4 and Section II. Third, it can be explained as a consequence of intervening, or overlapping, effects from other sectors within the international system. In the Medieval case, the obvious candidate is the international societal system. The existence of the Roman church as a strong, system-spanning societal entity might have helped to sustain the otherwise unstable type 4 mix. In the contemporary international system, high levels of economic interdependence and the emergence of an international society are the prime candidates for this cross-sectoral effect.
Wherever the correct historical answer lies, Ruggie is right to argue that differentiation of units remains open as a source of structural change in the international political system: type 4 systems are an area in which variation of second tier deep structure is possible. Anarchy may remain constant, but units can shift into and out of alternative patterns of differentiation of function. Both type 2 and type 4 anarchies are possible. This point does not require revision to the argument that type 4 political arrangements exist against the grain of deep structure forces. The appropriate Structural Realist hypothesis here is that type 2 structures should be more stable than type 4 ones. Type 4 should exist only when the logic of deep political structure is either overridden by forces arising in other levels or other sectors, or weakened by low levels of interaction. This reasoning significantly alters Waltz's causal nexus. The issue is more than just whether units or structures dominate outcomes within the political sector. There are also questions about the autonomy of the political sector, and the need to judge the effects of political structure against conditions in other sectors that may countervail against the narrow political logic of anarchy.
If the second tier is opened as an area of possible change, then there is a need for some definitional work to clarify the boundary between clearly similar units (no differentiation) and clearly different ones (where a principle of differentiation is in operation). This question is taken up in Section II. Waltz closed this area off, treating all sovereign states as like units. This strategy favors rather undemanding criteria for similarity, like those in such categories as fruit and mammal. Only then can all states over the modern period be treated as essentially the same type of unit. Ruggie wants to differentiate between these and the less exclusive, more overlapping, units of the Medieval system. Others might want to open up type 4 systems, and give better handles on questions of change, by arguing for tighter classifications of similarity (apples instead of fruit, humans instead of mammals). Robert Gilpin's work leans in this direction (Gilpin 1981:41-42).
It could be argued, for example, that there was a significant differentiation of units during the colonial period, with the system divided between sovereign states and varying degrees of less than sovereign vassals, ranging from colonies, through protectorates, to Dominions. If this was accepted, then decolonization, during which colonies changed into states, would mark a structural shift from a differentiated to an undifferentiated structure. Similarly, political economists might want to toy with the distinction between liberal and mercantilist states as a differentiation of function. Liberal states effectively renounce control of the economy as a function of government, turning it over to civil society, albeit remaining significant regulatory players. Mercantilist ones, especially of the socialist variety, claim the whole economy as a central function of government. Shifts in the composition of the international system between these two types might count as a second tier structural change for political economists.
The evolution of the European Community as a sui generis entity with some actor qualities perhaps also poses a challenge to the notion that the system is composed of undifferentiated units. The EC could also be seen simply as illustrating the socializing pressure of anarchy as the European states, finding themselves too small to function efficiently as major powers, seek a way to copy the power and security eminence of semi-continental states. In Waltzian terms (Waltz 1979:107), one of the fascinating elements in the European Community is how the states within it are torn between the anarchic imperative "look after yourself," and the hierarchic one "specialize." But the EC may not be simply a new state in the making. It may be a new type of semi-sovereign entity, binding together, but not replacing entirely, a group of states (Buzan et al. 1990:ch. 10).
If type 4 systems are to be made operational as part of Structural Realist theory, then their logic, dynamics, and classifications must be systematically developed. This is a major theoretical undertaking, at which we can do no more than hint here. It seems likely that the dynamics of a type 4 system would be more complicated than for type 2. Differentiated units would be more subject to interdependence, and perhaps, as in the example of colonies, to elements of hierarchy. Although balance of power might still operate strongly among subsets of like units within the system (e.g., the colonial powers) it might well be much less influential between functionally differentiated units. The relational dynamics of a type 4 system would also be much influenced by the nature of the forces preventing a reversion to type 2. A type 4 system made possible by weak interaction (thus emasculating the effects of socialization and competition) would be quite different from one in which functional specialization was supported in a high interaction environment with a strong international society among the units.
In investigating type 4 systems, it is important to be clear about the idea of function. Waltz uses it in the sense of function as a unit. He argues that all modern states function similarly as units because sovereignty imposes a similar range of governing tasks upon them. The pursuit of similar functions often leads to similar internal structures, as any comparison of states as organizations will show. This process is also encouraged by the homogenizing effect of competition and socialization under anarchy. But it is also possible to read function as including the different roles that states play in the international system. Some would like to push the idea of differentiation down this path by having it include role differences such as great and small powers, neutrals, allies, hegemons, and suchlike. This temptation should be resisted. Differentiation of function refers to function as a unit, and only in that sense does it link to the essence of deep structure, which is about government. Differentiation of roles is best dealt with at the unit level.
Given the modern history of the international system, this leaves us looking mostly at deep political structure defined in terms of the organizing principle of anarchy (types 2 and 4). Since there are no global cases of hierarchically structured political systems, types 1 and 3 are of interest mainly to those who either want to speculate about the character of, or the process of transition to, a world government, or those who want to think about the subglobal "world empires" of premodern times.
For practical purposes, anarchy thus remains a constant feature of the international system. But as revised along the lines pointed to by Ruggie, even deep structure offers interesting and accessible insights into change. Change in the first tier of structure will by definition be massive, extremely infrequent, and on the rare occasions when it occurs, extremely interesting. There has never been a global system change in the first tier, though analogous shifts from anarchy to hierarchy (the Roman and Chinese Empires) and from hierarchy to anarchy (the decolonization of South Asia, China's warring states periods) have occurred within regional subsystems. Change in the second tier is more easily accomplished than change in the first, and is historically more frequent. How much more frequent depends on what is accepted as a definition of "functional differentiation," as discussed above. Where first tier political structure operates strongly, second tier change is heavily constrained and conditioned by continuity in the first tier. Second tier change might be interesting as an illustration of first tier pressures at work, as in the twentieth-century move toward a uniform state system. But it is more likely to be interesting as an illustration of the political impact of forces from other sectors and levels overriding the narrow logic of the purely political structural theory. For those interested in first tier change, change in the second tier may also be an indicator of which way the wind is blowing. Opening up the second tier allows us to reject the argument that "structural Realist theory treats the internal characteristics of nation-states as given" (Mastanduno et al. 1989:461), and to accept, at least in part, the argument that the domestic character of states needs to be included in conceptions of international system structure. (Ruggie in Keohane 1986:147-48; Rosenberg 1990:299-303)
The Unit Level
The second arm of a pincer approach to the area between deep structure and the unit level is the unit level itself. Waltz does not say much about this. His clearest positive statement is where he defines the reductionist approach as understanding the whole "by knowing the attributes and the interactions of its parts" (Waltz 1979:18). This is reaffirmed in the negative when he says that "definitions of structure must leave aside, or abstract from, the characteristics of units, their behavior, and their interactions" (Waltz 1979:79). However, Waltz later slips into a tendency to deal with the unit level by exclusion, treating it as a catch-all for everything that falls outside his definition of structure: "structure includes only what is required to show how the units of the system are positioned or arranged. Everything else is omitted" (Waltz 1979:82, our italics). This questionable practice arises from both his preoccupation with structure, and his acceptance of a rigid two-level model of the international political system. It emerges most strongly in his response to Ruggie, and is implicit in such statements as:
Clean and simple definitions of structure save us from the pernicious practice of summoning new systems into being in response to every salient change within a system. They direct our attention to the units and to unit-level forces when the particularity of outcomes leads us to search for more idiosyncratic causes than are found in structures (in Keohane 1986:329).
Structural Realism requires a tight, positive definition of the unit level as a counterpoise to the structural one. A full system theory requires one to be as explicit about the unit level as about the system structure. Simply defining the unit level by exclusion, once a definition of structure is in place, lacks sufficient rigor to give proper definition to the unit level in its own right. It also forecloses the questions raised by Wendt (1987) about the relationship between agent and structure (which we take up in Section II).
According to Waltz's formulation, the unit level as a source of explanation for behaviors and outcomes in the international system, consists of two distinct components: the attributes of the units, and the interactions among them. Explanations in terms of attributes seek to understand the behavior of individual units by examining their domestic characteristics, components, and processes. Foreign policy analyses where outcomes are interpreted in terms of such factors as bureaucratic process, leadership personality, the organization of the government, and the organizing ideology of the state, all illustrate this method. So do theories of war which suggest that certain types of state--autocratic, fascist, capitalist, communist--are likely to be more aggressive than others because of the domestic dynamics generated by their mode of social, political, and/or economic organization. Attempts to explain behavior in terms of the preference functions of units--whether they seek power, security, welfare, or cultural values as their prime objective--also fall within the unit attribute mode of analysis.
Explanations in terms of interactions among units--often referred to as process (Keohane and Nye 1987:745) seek to understand behavior and outcomes in terms of the ways in which units respond to each other's attributes and behaviors. As such, they are heavily conditioned by deep structure. One would not expect the process manifestations of anarchy and hierarchy to be similar, nor would one expect process to be unaffected by the difference between like and unlike units. These are essentially action-reaction theories, in which the key element is a dynamic of stimulus and response. Many recurrent patterns have been found in these often very complex dynamics, and it is the elements of consistency in these patterns that inspire attempts at theory. The best known of them include war, alliance, the balance of power, arms racing and the security dilemma, and the whole range of international political economy patterns arising from protectionist and liberal policies on trade and money. Buzan's concept of regional security complexes also belongs in this category, since it is based on durable, but not permanent, patterns in the degree of amity, enmity, and indifference with which states view each other (Buzan and Rizvi 1986:chs. 1 and 9; Buzan 1991:ch. 5), as does Bull's concept of international society (Bull 1977), and the related idea of regimes. Here also lies the often confusing, but important, distinction between the structural feature of polarity (the number of major powers in the system) and the process feature of polarization (the number of opposed coalitions in the system) (Rapkin et al. 1979). These patterns are the focus of much theoretical enterprise aimed at finding ways of mitigating or controlling negative effects by constructing countervailing patterns. Regimes, for example, are proposed as a way of stabilizing trade and financial relations, and interdependence is offered as a way of muting the security dilemma. Generalized theoretical conceptions such as Choucri and North's idea of "lateral pressure" are aimed at understanding system process (Choucri and North 1975).
Because these process patterns do exhibit some consistency--for example the beggar-thy-neighbor logic of protectionism, the self-defeating logic of the arms dynamic, and the anti-hegemony logic of the balance of power--they give rise to the tendency deplored by Waltz to identify them as structures. The common sense logic here is based on an interpretation of "structure" to mean simply any pattern that either endures or recurs. Despite their persistence, Waltz is logically correct to count these patterns as unit level explanations and to reserve the term structure for the positional relations of the units. Yet there is no doubt that unit attributes and process are quite different types of explanation. Some analysts see them as different enough to justify calling them separate levels of analysis (Goldmann 1979:1-2). The choice here is between elevating process to the status of a level of analysis, or establishing a strong boundary within the unit level (between attributes and process) along the same lines as that between Waltz's first two, and third, tiers of structure. The analytical consequence of this choice is not large, and for the sake of consistency we will stick to Waltz's scheme. Waltz is at fault in this area for constructing an unbalanced systems theory by developing a highly elaborate definition of structure, while leaving the unit level as an indifferentiated mass about which it has been all too easy for confusions to multiply. The error is especially damaging, not only because it draws too hard a line across the agent-structure debate, but also because it is precisely through elaboration of process that change can best be dealt with in the context of structural theory.
To relieve the pressure on the term structure, and to differentiate process explanations from the attribute mode of analysis, some other label is needed to identify the patterns arising from the interactions of units. Given the fashion for using "structure" in a loose way, and the absence of obvious alternatives, this is no easy matter. Simply qualifying the term structure (as in "process structures") has the appeal of ease and convenience, but fatally blurs the distinction between the unit and structural levels. The word "formation" is perhaps an acceptable substitute, maintaining both the distinction from structure, and the sense of durable or recurrent patterns in behavior at the unit level. These patterns can be referred to as process formations.
It is beyond the scope of this section to delve into the labyrinths of theory at the unit level. It is important, however, to establish that the unit and structural levels are linked. Since the two levels are in some ways mutually constitutive, Structural Realism is able to serve as a theory of the international system as a whole. On this point, it is possible to agree with Wendt that agents and structures should be seen as mutually constitutive, while disagreeing with him that Neorealism simply generates structures from units (Wendt 1987:359-61, 340-44). Waltz also too easily forecloses opportunities here by arguing in favor of de-linking theories of the state from those of the international system (in Keohane 1986:340).
One of the attractions of Structural Realism's approach is precisely that it does neatly incorporate the mutually constitutive relationship between unit and system. Anarchy and autonomy are opposite sides of the same coin. As has been shown in the discussion of deep structure, structural forces play a major role in constituting sovereign states. Anarchy equals self-help (with or without states), though it does not preclude cooperation. Self-help generates a competitive and sometimes conflictual system. Competition leads toward like units by the mechanisms of copying and coercion though still leaving open strategies of differentiation as suggested above. Structural Realism certainly does not assume that states are constructed entirely by forces generated from within. Because this is so, it can and should serve as a linking framework for theories at the unit and structural levels, a theme taken up in Section II. Making this link explicit is one way to release Structural Realist theory from the high level of generalization to which Waltz has largely confined Neorealism.
Between the Deep Structure and the Unit Level
In Waltz's scheme, the only occupant of the space between deep structure and the unit level is the third tier of structure, the distribution of capabilities, which he sees as an integral part of his definition of structure. It plays an important role in Neorealism because, within it, change is frequent enough to bring the theory into some contact with the universe of policy problems. Without this level, Neorealism would be completely static on a timescale of millennia, particularly since Waltz closes off the second tier. It would retain only the (still considerable) value of its deep structure insights into what conditions behavior in the international system. Two issues need to be considered in order to clarify the logic of this area: (1) the soundness of the principle of distribution as an element of system structure, and (2) the validity of aggregating plural "capabilities" into singular "power."
The Principle of Distribution
In Neorealism, the principle of distribution is a system level element because it comes within the general definition of structure as being about "how units stand in relation to one another, the way they are arranged or positioned" (Waltz 1979:80). Even though the principle refers to unit characteristics, it hinges on their distribution, a property that fits into the concept of "relation" seen in positional terms (Waltz 1979:79-82, 97-98). This argument seems sound in itself, and provides a clear way of drawing the boundary between the structure and unit levels.
In relation to deep structure it can be argued that the distribution of capabilities derives directly from the organizing principle of anarchy. A fragmented political structure, mediated by a diverse geography, necessarily and automatically generates a pattern in the distribution of capabilities. This pattern could, in theory, be one of even distribution, but in practice it is almost certain to be uneven, and probably very uneven. Historically, the political units within the system have distributed themselves across a range of capabilities covering several orders of magnitude: the United States at one end, Vanuatu at the other. This uneven distribution will occur no matter whether the fragmentation is based mostly on territory, as in a type 2 structure with sovereign states (anarchy/similar), or more on organization, as in a type 4 structure like Ruggie's Medieval system (anarchy/different). It underlies and motivates many of the process formations at the unit level, which gives the theory some generative standing inasmuch as each tier within it strongly conditions the one(s) below. In this type of structuralism (i.e., where structure is defined by the arrangement of the units), it is impossible for structure to act as an ultimate generative source, as it does in some other (e.g., linguistic) conceptions.
In relation to the unit level, the principle of distribution clearly represents a different and higher type of generalization than either type of explanation within the unit level: "capabilities are attributes of units, the distribution of capabilities across units is not" (Waltz 1979:98). Explanation of behavior in terms of possession of a capability by a unit is quite different from explanation in terms of the distribution of capability within the system. In power terms, the fact that any single state has many capabilities is a different type of explanation for how it behaves than is the fact that the system contains only two great powers. Distribution of capabilities is also logically distinct from the action-reaction patterns that were labeled process formations above. Distribution patterns may underlie process formations, but they are not the same thing. Interaction process focuses on the action-reaction dynamics of unit behavior. Distribution patterns are about the positional logic of how units are ranked in terms of their capabilities.
Since the logic of Waltz's boundary between the system and unit levels seems sound, the main question that arises here is whether his single, but multi-tiered, definition of structure is the most useful way of presenting his ideas. The case for lumping all three tiers together is that they all conform to the logic of structure as positional relation, which is what differentiates them from the unit level. But within that logic, there are sufficient differences between the third tier and the first two to make a good case for putting more emphasis on the sub-boundary between them. Two differences stand out, one major, one minor.
The major one is that tiers one and two are connected by a powerful logic (discussed in the fifth part of this chapter) that does not extend to tier three. Tiers one and two are both concerned with "positioning" in terms of the principles of political organization. It is on that basis that the tier one logic of basic ordering principles, and the tier two logic of the functional differentiation of units, interact so closely (virtually to the point of merger in Waltz's narrow formulation). Tier three is not affected by this logic. Indeed, it is not specifically about political organization at all, but mostly concerns the relative weight of resources available to the political units within the system. The link lies in the fundamental Realist axiom that anarchic deep structure makes power the ultima ratio of international relations.
The minor difference is the relative depth of the structure identified by tiers one and two as opposed to three. Tier one is relatively deep, in the sense that major change is infrequent and has large consequences for political behavior. In Waltz's formulation tier two is closed and so doesn't count. Reopening it, as we have done, raises the question of how frequent change within it is. The answer to that question depends on the yet-to be decided matter of what qualifies as a change in differentiation of function, on which see the discussion above, and in Section II. Tier three is much shallower, in that change is more frequent and can be of low consequence. Changes in the distribution of capability may have little impact if they are simply from one high degree of diffuseness to another. A shift from a seven to a six power system might well be only of local interest. The significance of change rises as the number of powers gets smaller, and Waltz argues that only changes between one and four are significant enough to be called structural shifts (Waltz 1979:163). On this logic, the most significant shift would be one from bipolarity to unipolarity. A unipolar power structure could, but would not necessarily, trigger a deep structure shift from anarchy to hierarchy. It might equally trigger frenzied balancing attempts to stave off the final move from hegemony to world empire. There is thus a different, and less compelling logic linking tier three to the deep structure, compared with the way that tiers one and two are linked together. Like the first two tiers, however, tier three is also a useful tool for static analysis. When change is not the issue it is a fruitful source of insights into how structural continuities condition the behavior of units.
The extent of these differences between the first two tiers and the third has already meant that the terms "structure" and "structural change" cannot sensibly be used with a single meaning. In discussing political structure, whether in terms of its effects, or in terms of changes in it, one is forced to qualify which sublevel one is referring to. For this reason, it seems sensible to keep the unity of the term structure, but to distinguish within it between deep structure (tiers one and two), and distributional structure (tier 3). This distinction is within the framework of Waltz's logic, and clarifies the usage of structure. The case for identifying a sub-boundary between deep and distributional structure is reinforced when one questions the validity of aggregating "capabilities" into "power."
The Aggregation of "Capabilities" into "Power"
The second step in Waltz's logic of distributional structure is a complicated one. He aggregates his originally plural formulation of capabilities into the singular concept of power, and then restricts the application of the distributional principle exclusively to the distribution of power (Waltz 1979:97-99). His reasons for proceeding in this way are neither clear nor convincing. His formulation seems to place unhelpful and unwarranted restrictions on the application of the principle of distribution, and so on the application of the theory. He is aware that he is venturing onto boggy terrain by abstracting so closely from unit level attributes, but establishes firm ground with the principle of distribution. His tactics raise the question: why is the principle of distribution only to be applied to power? This question has two overlapping components: why must plural capabilities be collapsed into singular power; and why are attributes other than power excluded from the principle of distribution?
On the first question, Waltz's eagerness to end up with power is understandable. Because he has closed off the second tier, he needs the power dimension as a way of classifying and differentiating the units within anarchy: "the units of an anarchic system are functionally undifferentiated. The units of such an order are then distinguished primarily by their greater or lesser capability for performing similar tasks" (Waltz 1979:97). With the second tier of structure now opened, this rationale vanishes for Structural Realism. Waltz also wants to connect to the central concept in Classical Realist analysis, and makes the point that he is simply restating formally the emphasis on great powers that has always marked the study of international relations (Waltz 1979:97).
The principal justification for collapsing capabilities into power stems from the competitive self-help imperatives of anarchic deep structure:
States, because they are in a self-help system, have to use their combined capabilities in order to serve their interests. The economic, military, and other capabilities of nations cannot be sectored and separately weighed. States are not placed in the top rank because they excel in one way or another. Their rank depends on how they score on all of the following items: size of population and territory, resource endowment, economic capability, military strength, political stability and competence (Waltz 1979:131, his italics).
Is this aggregation necessary to the theory? If not, is it the best way to apply the theory? The obvious gain is that it enables Neorealism to pose as a highly parsimonious grand theory. But we will argue that aggregation often misleads more than it informs, and that under many circumstances it simply cannot be made to work at all. Grand parsimony is only occasionally available, and for the most part little is lost and much gained by treating power in terms of separate capabilities.
On the question of why the principle of distribution is confined only to power, Waltz's first justification is a technical one. He chooses power because it "is estimated by comparing the capabilities of a number of units" (Waltz 1979:98). In other words, the choice of power is made initially simply because it fits with the requirements of the principle of distribution. The key distinction here is between, on the one hand, distribution of attributes seen in direct, absolute terms (e.g., 15 democracies, 12 communist states, 30 military governments, etc.), and on the other, distribution of attributes seen in relative terms (e.g., not how many states with armed forces and GNPs of specified sizes, but how many great powers). It is the relative measurement of power that makes its distribution a structural rather than a unit level feature. This logic is sound, but it does not seal off the possibility that other attributive elements of states could also be cast in distributional terms.
Waltz recognizes this, and in his response to his critics, he argues that power is selected simply because it is a more important determinant of behavior than other characteristics of states. This reasoning harks back to Morgenthau, and the traditional Realist aspiration to derive an autonomous theory of political behavior similar in form to economic theory. Where the economists defined interest in terms of wealth, and assumed a principle of utility maximization, Classical Realists defined interests in terms of power (Morgenthau 1978:5), and many of them followed the economic parallel by assuming a motive of power maximization. While Waltz is careful to dissociate himself from the power-maximizing school of Realists, he chooses power because he takes it to be the dominant factor, not because it satisfies criteria of logical completion within categories.
In a self-help system, states are differently placed by their power. . . State behavior varies more with differences of power than with differences in ideology in internal structure of property relations, or in governmental form. In self-help systems, the pressures of competition weigh more heavily than ideological preferences or internal political pressures (in Keohane 1986:329).
This at least makes the reasoning clear, and Waltz tries to reinforce it with a plea for "clean and simple definitions of structure" (in Keohane 1986:329). The resulting formulation is not logically complete, and in Waltz's reasoning does not need to be. The issue here cannot be separated from his decision to collapse capabilities into power. The principle of distribution does not, in itself, require such aggregation. Nor does the theory require confinement to an erstwhile dominant element. Even if power was the most important variable within the distributional tier of structure, that would still be a weak reason to exclude other variables that meet the logical requirements of the theory. The Neorealist focus on aggregated power is neither the most logical nor the most effective way to deploy structural theory in an international system context.
Disaggregated Capabilities and the Application of Structural Realism
The arguments above make two points. First, the principle of distribution is solidly established as an element of structure. Second, neither the aggregation of capabilities into power, nor the restriction of the principle of distribution to power, necessarily follow from the logic of the theory. The question is whether aggregation and restriction best serve the utility of the theory, or whether a wider and more useful range of applications can be opened up. Two issues need to be considered: the case for reversing the aggregation of capabilities into power, and the case for de-restricting the application of the principle of distribution exclusively to power. In both cases the logic of sectors developed above under System Sectors and to be expanded in chapter 12, points to the utility of differentiated power. Indeed, there is nothing new in this idea in itself: discussion of disaggregated power has been a feature of the literature on interdependence for many years (Baldwin 1980; Keohane and Nye 1977).
Aggregation across a wide range of capabilities absorbs attributes into the catch-all of power that there may be good reasons for considering separately. Waltz's list of six components of state power (Waltz 1979:131) is not a mutually exclusive set. The three most complex and organizational components are also the most interesting candidates for separate treatment--military strength, economic capability, and political stability. They largely incorporate the three simpler, and more purely physical components--population, territory, and resource endowment. Waltz's explicit rejection of the separate treatment of attributes (Waltz 1979:130) seems to stem almost entirely from his concern to keep structure simple. It does not follow from the principle of distribution. Within that logic, aggregation of diverse capabilities into the gross category of power is simply one of several possible ways of proceeding. Theories are judged by their usefulness, and it is on that basis that aggregated versus disaggregated applications of the principle of distribution should be assessed. Both fall clearly within the logic of system structure as defined above.
How useful is the aggregation of capabilities into power? Although it does produce a sparse version of distributional structure, it also creates several problems. Like Morgenthau, Waltz cannot easily escape the difficulties of treating an elusive concept like power in scientific terms (Keohane 1986:10). Waltz acknowledges the difficulty of measuring the diverse and shifting components of power in any definitive way, but contrives an escape from this dilemma by appealing to a kind of mass common sense: "historically, despite the difficulties, one finds general agreement about who the great powers of a period are" (Waltz 1979:131). It is not clear whether this general agreement is thought to have existed at the time or only in historical retrospect, a point of some importance for the applicability of the theory to the present. Nevertheless, by this generalizing device Waltz seeks to preserve the aggregation of capabilities by avoiding the very serious difficulties that confront attempts to devise objective measures for state power.
Here he walks into a trap of his own making. Having derided well-informed and thoughtful individuals such as Kissinger, who have sought to disaggregate capabilities (Waltz 1979:130), he is not well placed to appeal to common sense as a means of overcoming the ambiguity of power. The fact is that under present conditions, the application of common sense produces hopeless confusion in any attempt to deal with power in aggregated terms. When the different components of state power don't line up, the aggregated category collapses. What in the late 1980s was one to make of an ostensibly bipolar system in which one of the two "superpowers" had only the third largest economy in the system, and the other had the biggest debt? How did one weigh the relative importance of the Soviet Union's large territory and endowment of resources against China's large population and ethnic coherence? How does one weigh China's nuclear weapons against Japan's wealth? At what point did the economic and political disintegration of the Soviet Union outweigh its large military power sufficiently to disqualify it as a pole? How does common sense enable one to assess polarity in the fluid conditions of the 1990s when plausible arguments can be made for unipolarity, bipolarity and multipolarity all at the same time?
Waltz's appeal to the analogy of oligopolistic markets to support the legitimacy of his aggregative approach actually supports the disaggregative case. He implies that oligopoly analysis is simply about how many major firms populate a sector of an economy, and that power analysis is therefore similarly appropriate for the international political system (Waltz 1979:131). Leaving aside the difficulty that the "sector" referred to by economists is merely a small subsector of an economic system, and therefore a much narrower construct than the political sector of the international system, the fact is that oligopoly theory has itself had to tread the path of disaggregation. Markham argues that "oligopoly is far less homogeneous than the term `few sellers' might imply," and that theorists need to break down "an unsatisfactory and meaningless aggregate into classes useful for analytical purposes" (Markham 1968:288-89). Mainstream analysis has already made progress down this path by dividing oligopolies into "concentrated" (deriving from economies of scale), "differentiated" (arising from economies of specialization), and mixed types (Sylos-Labini 1987:701). In addition, static approaches to oligopolistic competition have been challenged by more dynamic ones, which accept as systematic, rather than merely accidental, observed oscillations of competitive and collusive behavior (see Section III).
Tellingly, a review of Waltz's applications of Neorealism in chapters 6 to 8 of TIP reveals that aggregation is not a necessary condition for any of his major observations (leaving aside the question of whether or not one agrees with his applications of the theory, particularly his controversial interpretations of bipolarity and interdependence). His insights deriving from deep structure are unaffected by how distributional structure is treated. Those on the balance of power could just as well be discussed in terms of the distribution of military strength. The full meaning of military strength anyway takes in many of Waltz's components of power--and poses similar difficulties of assessment. His arguments on economic effects could just as well be discussed in terms of polarity defined by economic capability. If this were done, his bipolar case would become theoretical rather than empirical, because the contemporary international system has never displayed economic bipolarity except perhaps briefly right after the end of the Second World War. Although the use of aggregated power did enable Waltz to say something interesting about superpower relations, it did so only at the cost of missing both most of the economic reality of the contemporary international system, and most of the substance of the interdependence argument. The Neorealist approach offers no insights that could not have been gained using disaggregated power, which largely removes the rationale for a highly parsimonious theory.
Disaggregating capabilities thus seems to have few costs and many advantages. Waltz notes, but does not pursue, the idea that "distinguishing between anarchic structures of different type permits somewhat narrower and more precise definitions of expected outcomes" (Waltz 1979:70). Any move from simple toward complex has to be justified against the general rule that parsimony is preferable to elaboration in the construction of theory. Waltz is very strict on this point, but it may well be that he has pushed the pursuit of parsimony too far in some places, with results that limit the utility of Neorealism. Disaggregation does increase the diversity of what counts as structural change. But it does so firmly within the bounds of distributional structure. It does not affect the simplicity of deep structure, and it does not invite spillovers from process formations at the unit level. In addition, there are not all that many attractive candidates for defining different types of distributional structures: neither massive nor uncontrolled proliferation of system types is a problem.
One advantage of disaggregation is simply that it reduces the confusion caused by trying to use undifferentiated power, and so produces a clearer analysis. Separating military strength and economic capability, for example, gives a much more accurate characterization of distributional structure than attempting to work with an extremely unwieldy aggregate. Military and economic hypotheses can be pursued within the two resultant distributional structures, and an additional area of insight is generated by the study of situations in which these two distributional structures do and do not line up. Where they do coincide, aggregated analysis may be the most appropriate formulation. Where they don't, the interplay of the disjunctures is itself an important structural datum. Disaggregation offers scope for a wider variety of hypotheses than is possible in Neorealism. One could investigate, for example, whether increases in polarity are less likely, and decreases more likely, to trigger system-wide wars; and within that question whether military or economic polarity change accounts for more violence, or whether it is the combination of them that matters.
A second advantage of disaggregation is that it enables structural theory to be connected in a constructive way to the mass of work on process formations. A good chunk of this is taking place under the heading of interdependence (or Neoliberalism), and one key to it has been the disaggregation of power along issue lines as a means of making sense of outcomes in an interdependent international political economy. Caution is required here, because analysis is poised on the boundary between the structural and unit levels of analysis. As Keohane makes clear (Keohane 1986:182-97), the disaggregation of power, and the abandonment of the Classical Realist assumption that power (broadly conceived) is fungible across issues, is not without problems. The point is not to try to blur the boundary between distributional structure and the unit level analysis of interdependence, but to set them into a clear and complementary relationship with each other, and to place both in the context of the interaction component of system discussed in chapter 4. Waltz is right to stress that theory does not need to describe every detail of reality, and should not be judged by how accurately it does so. To push theory in the direction of describing reality is to defeat its purpose, which is to provide generalized explanations for outcomes (Waltz 1979:6-9). But maintaining theory as an abstraction does not mean that it has to be disconnected either from reality or from other levels of theory. A theory increases in usefulness to the extent that its explanations can be related both to empirical observation and to other areas of theory.
A disaggregated distributional structure can both preserve a clear boundary between itself and unit level analysis, and at the same time provide a framework within which the unit level analysis can be organized. For example, a description of distributional structure in degrees of polarity can result from analysis either in terms of economic capability or military strength. Polarity defines the distributional structure in terms of a specific capability. The degree of polarity may range from unipolar to diffuse, and structural hypotheses can be generated for the range of conditions that result. These hypotheses can serve as a foundation for unit level analysis without causing confusion or trespass between levels of analysis. Structures define the environment that conditions and constrains unit level causes.
Illustrating this approach in more detail leads to disaggregation's third advantage: it allows more finely tuned structural hypotheses. Waltz made only limited progress toward operationalizing Neorealism by generating hypotheses from it. His thoughts about the behavioral consequences of deep structure have proved quite acceptable, whereas those about distributional structure have been more controversial, not least because of his attempt to work with an undifferentiated concept of power. Trying to think in terms of aggregated power has led, inter alia, to the overambitious and inconclusive debate about polarity and war (Deutsch and Singer 1964; Rosecrance 1969; Thompson 1986; Waltz 1964) and to Waltz's interesting, but rather tangential analysis of interdependence (Waltz 1970; 1979, ch. 7).
Analysis in terms of specific capabilities opens up clearer possibilities for a considerable range of more precisely defined structural hypotheses. In relation to military strength, for example, variations in degrees of polarity generate a host of important "shoving and shaping" forces (Buzan 1987). The logics of deterrence, arms control, arms racing, and alliance all vary sharply according to degrees of polarity. Deterrence requirements and logic are transformed by the differences between unipolar, bipolar, and multipolar systems. An example of the possibilities here is provided by the work of Intrilligator and Brito, who calculated the deterrence consequences of nuclear polarity from unipolarity up through successive additions of nuclear powers to a full multipolar system (Intrilligator and Brito 1979). Arms racing is more complex in multipolar than in bipolar structures. One could begin to construct systematic theory in this rather poorly developed area by starting analysis with a specification of military polarity. One hypothesis connecting distributional structure to process formations is that, under bipolarity, a strong pressure is created for the two principal powers to engage in an arms race. Military polarity is similarly important to theorizing about arms control. For example the principle of parity in arms control is much easier to formulate in bipolar than in multipolar conditions. Polarity also affects both the possibilities and conditions for alliance making. Alliances are more likely to be rigid and long-lasting in bipolar systems than in multipolar ones. Again, starting with the relevant distributional structure seems a fruitful way to approach this subject, enabling one to construct clear and systematic links between structure and process formations. Just how does military polarity relate to degrees of polarization in the system in terms of alliance coalitions? (Rapkin et al. 1979).
In relation to economic capability, Waltz has already explored some of the logic that arises in a bipolar distributional structure. The logic of unipolarity relates in an obvious way to the study of hegemony, and so offers a connection to the work, among others, of Gilpin, Keohane, Kindleberger and Modelski (Gilpin 1981; Keohane 1984; Kindleberger 1981; Modelski 1978). Disaggregation opens up some particularly useful ground in relation to hegemony. In the military sector, and when using aggregated power, unipolarity is structurally resisted by the balance of power and hardly ever occurs. But it is a fairly common feature in the economic sector, where relational logic is much more supportive of it. Dealing in terms of aggregated power virtually excludes the possibility of discussing hegemony, because it hides economic unipolarity within a balance of military power that is peculiarly resistant to unipolar distributional structure. Again, starting analysis by specifying distributional structure is a useful way to link structural and process/unit factors in a systematic fashion. It might also link to the second tier of deep structure if the difference between liberal and mercantilist states is accepted as a functional differentiation. The difference between economic multipolarity and unipolarity raises well-known questions about system management and the sustainability of liberal economic policies. All of these "shoving and shaping" forces could be cast in the form of hypotheses deriving from distributional structure seen in terms of economic polarity.
The other "capability" liberated from aggregated power is political stability. This can more usefully be cast in terms of political cohesion, states being differentiated along these lines in terms of a spectrum running from strong (meaning closely integrated and stable state-society relations) to weak (meaning weakly integrated and unstable state-society relations) (Buzan 1988:18-27; and 1991:ch.2). In using these terms it is important to keep clear the distinction between weak and strong states, in terms of political cohesion, and weak and strong powers, in terms of military and economic capabilities. A weakish state, like Pakistan (or the Soviet Union as it was), might be quite strong as a power, and vice versa.
The variable of political cohesion can be cast in distributional terms, but does not yield structure in the form of polarity. It links interestingly to the definition of political deep structure. How lacking in sociopolitical cohesion does a state have to be before it either meets criteria of functional differentiation, or, in extremis, ceases to count as a unit of the system--for example post-1976 Lebanon? The most useful way of characterizing the distributional structure of the international political system in terms of political cohesion is to ask how homogeneous it is. Are the units all strong in terms of political cohesion? Are they all weak? Is there some mix of weak and strong? Applied in this way the principle of distribution can yield important structural hypotheses about the tendency of relations at the unit level. For example, one could posit that in a homogeneous system of strong states, intervention will tend to have low salience in international relations. In a system containing a mix of weak and strong states, like the present one, intervention by the strong into the weak will tend to be a major and unavoidable feature of international relations. In a homogeneous system of weak states all interaction would be intervention (Buzan 1988:27-31).
The process of disaggregating power liberates most of the attributes of units that are interesting in distributional terms. The remaining case for de-restriction rests on the principle of distribution itself. Logic demands that any attribute which can be cast in the relational terms of distribution be admitted to the theory. Waltz concedes the point that there are additional candidates with his reference to "other characteristics of states that could be cast in distributional terms" (in Keohane 1986:329), even though he has, in fact, buried most of them within his definition of power. In the political sector, the most obvious remaining candidate is the ideology of governments. In distributional terms, ideology can be treated like political cohesion: is a political system ideologically homogeneous or plural in terms of the number of major organizing ideologies active among its units? This question can almost be put into the terms of polarity: is the system ideologically unipolar (like eighteenth-century Europe), or bipolar (1950s), or tripolar (1970s) with the rise of Islamic politics; or, (as in the 1930s) with liberalism, communism, and fascism). Structural hypotheses arising from this variable might usefully address questions about alliance formation (polarization), international society, and system management.
The end result of disaggregating and de-restricting the distribution of capabilities is a fairly limited set of possible distributional structures based on four types of attribute: military capability, economic capability, political cohesion, and ideology. To assess the distributional structure of any given system would involve examining a set of overlapping "shoving and shaping" forces. These can be envisaged as a set of vectors (lines of force) operating within a field. Sometimes their influence would be mutually reinforcing, as when economic, military, and ideological patterns all coincide--as they did briefly after the Second World War. When that happens strong structural pressure on the behavior of units is created. Sometimes the different vectors would work at odds, perhaps canceling each other out, or at least creating weaker, more diffuse structural forces, as in the international system of the early 1990s. This could leave the field relatively devoid of influence from distributional structure, and therefore clear for the operation of unit level and deep structure factors.
Because more variables are in play, a disaggregated distributional structure will exhibit more change than an aggregated one. Any individual element might remain stable for a considerable period, as military strength has done since the end of the Second World War. But all the vectors are unlikely to remain fixed, and the influence of those that do has to be calculated against the shifting influence of those that don't. A political-economy analysis, for example, would need to deal with the increase in economic polarity in interplay with the stasis in military bipolarity. In this way, the powerful positional logic of structural analysis can be brought to bear on a wider range of events within the system. This can be done without transgressing the boundary between the unit and the structural levels. Although somewhat more complex than aggregated polarity, this form of analysis has three advantages: (1) it yields more finely tuned structural hypotheses, (2) it is applicable to a much wider range of conditions within the international system, and (3) it gives a fairly rich and replicable structural framework within which to conduct systematic comparative analysis.
Disaggregation does not lose any of the insights of the more parsimonious approach. The error of Neorealism is to try to universalize the rather special case in which the main elements of power are all distributed in the same pattern. This special case does sometimes occur, but it is not necessarily, or even probably, the normal condition of the distributional structure.
Summary and Conclusion
This move through Neorealism toward Structural Realism has left two basic elements of Waltz's structural theory firmly in place: the first tier of deep structure, and the boundary between the system and unit levels. Four new developments have been sketched out: the clarification of the sectoral boundary of the theory; the elaboration of what the unit level actually contains; the opening up, clearer demarcation, and application of the third tier, relabeled as distributional structure; and the opening up of the second tier of structure (the functional differentiation of units), which will be pursued further in Section II. The opening up of the second and third tiers of structure enormously increases the scope of Structural Realist theory, allowing it to serve as a wide-ranging and integrative general theory of international relations. The vocabulary of Structural Realism is summarized in figure 3.3.
Email this citation