Susan B. Martin

University of Pennsylvania

May 1999

Christopher H. Browne Center for International Politics

University of Pennsylvania

I would like to thank Avery Goldstein, João Resende-Santos, and Kenneth Shadlen for their helpful comments and suggestions.

Abstract:

This paper responds to recent criticisms of balance of power theory by arguing that those criticisms represent a misunderstanding of both the contributions and limitations of systemic balance of power theory. At the same time, it acknowledges that there are some problems with current work on balance of power, in particular with work that uses systemic balance of power theory as a basis for studies of state behavior. I argue that the problems with this current work are a result of the failure to make the adjustments required by the change in the level of analysis that occurs when one moves from systemic theory to the study of state behavior. In particular, there has been little analysis of what it means for a state to "balance." To address this problem, I develop a definition of balancing behavior that is consistent with systemic balance of power theory, and then use that definition to develop a general model of balancing behavior. I then show how that general model can be used to integrate and evaluate different hypotheses concerning the balancing behavior of states. I conclude that both systemic balance of power theory and studies of the balancing behavior of states can contribute to our understanding of international politics, and therefore argue that we should resist the suggestion of critics that balance of power theory should be abandoned.

Balance of power theory has been under attack recently, portrayed as a degenerating research design, as historically inaccurate, and as made irrelevant by the end of the Cold War. Of course, this negative evaluation of balance of power theory is nothing new; even those of us who work within its framework recognize and often lament the chaotic tradition labeled "balance of power." But despite the myriad different meanings given to the term balance of power, despite the number of contradictory approaches that bear its name, and despite the repeated calls to abandon the concept because it is hopelessly confused, students of international politics continue to be drawn to balance of power because it captures something essential about international politics.

This article defends balance of power theory against those who call for its abandonment while at the same time recognizing that there are problems with current work. In particular, I argue that the scope of systemic balance of power theory has been overestimated, and that attempts to use systemic balance of power theory as a basis for studies of state behavior are problematic because they fail to analyze exactly what it means for a state to balance.

The first section of this paper briefly reviews the literature on balance of power and examines the problems encountered when one moves from systemic balance of power theory to the balancing behavior of states. I argue that a more complete definition of balancing behavior, rooted in systemic balance of power theory, is both possible and necessary. After examining systemic balance of power theory in more depth and clarifying what it is about international politics that systemic balance of power theory does and does not capture, the next two sections then provide such a definition. I then present a general model of balancing behavior based upon this definition, which can be used to integrate, compare and evaluate rival hypothesis about the balancing behavior of states.

Within the tradition of balance of power theory one can identify two main strands: the first is systemic balance of power theory, which is concerned with the reoccurrence of rough balances of power within the international system, while the second focuses on balance of power theory as a guide to, or explanation of, state behavior. The former is evident in Thucydides' description of how changes in the distribution of power produced war between Athens and Sparta, in Ludwig Dehio’s discussion of the role of flanking powers in preventing the emergence of a hegemon in Europe, and in the thought of Jean Jacques Rousseau. In these discussions the focus is not on explaining the behavior of a particular state, but on explaining how changes in the distribution of power within an anarchic international system shape the constraints, opportunities, and incentives facing states. It is this strand of balance of power theory that is distilled and clarified in the systemic balance of power theory presented by Kenneth Waltz.

The second strand of the balance of power tradition focuses not on systemic constraints and international outcomes but on the behavior of states. This strand focuses on the foreign policy of states, examining whether states try to create balances of power as well as the extent to which balance of power theory provides a useful explanation of state behavior. As Inis Claude explains, this strand "treats balance of power as a policy of states or as a principle capable of inspiring the policy of states." Arnold Wolfers’ essay on balance of power is one example. In his discussion of four approaches to the balance of power, Wolfers dismisses the "automatic" or systemic notion of a balance of power as not suited to his purposes, which concern policy, not outcomes. The important question for him is "whether nations under certain circumstances do or should, as a matter of expediency, make power equilibrium rather than power superiority the target of their effort."

Edward Gulick’s work on the balance of power is another example of this strand. He sets himself two tasks in Europe’s Classic Balance of Power: First, to derive general propositions about the balance of power from the thought of the theorists and practitioners of the period he is examining, and second, to trace the effects of those propositions on the foreign policy of states. Paul Schroeder also examines how statesmen understand international politics and how that understanding influences their behavior, although he reaches quite different conclusions about balance of power theory in his work The Transformation of European Politics. Schroeder argues that after the fall of Napoleon, a change in understanding took place that transformed international politics; balance of power politics was replaced with concert and political equilibrium. Schroeder and Gulick are thus both interested in how ideas shape international politics, although they reach very different conclusions about the influence of balance of power on the actions of states.

In recent work the line between these traditions--one focusing on systemic outcomes and the other on the behavior of states--has been blurred, with analyses that use the systemic balance of power theory associated with neorealism as a basis for work that focuses on the behavior of states. It is my contention that much of the confusion surrounding balance of power theory today is a result, first, of a misunderstanding of what systemic balance of power theory can do, and second, of analysts moving from systemic balance of power theory to the study of state behavior without recognizing the implications of the change in the level of analysis.

J. David Singer argues in his 1961 article on levels of analysis that when one moves from one level of analysis to another, an "act of translation" is required: For example, when one moves from the systemic level to the state level, it is necessary to analyze what variables at the state level, what sort of state actions or policies, correspond to the relevant variable at the systemic level. The central insight of systemic international relations theory–that international outcomes are not reducible to the individual actions or intentions of states–is itself a reason why this act of translation has to take place. If an international outcome that we are interested in cannot be reduced to specific actions taken by states, then when taking propositions or ideas about the causes of outcomes at the systemic level and applying them to state behavior, we have to think carefully about what sorts of things at the state level of analysis correspond to particular international outcomes. We have to translate the variables we are interested in at one level into the corresponding variables at other levels. This act of translation has not been adequately performed by most of the analysts who use neorealism as a basis for their studies of state behavior.

The problems this lack of translation creates are two-fold. First, the propositions that can be generated from systemic theory about state behavior are stated in terms of generalities, in terms of what states "tend" to do or what the systemic incentives encourage them to do. Propositions about what states tend to do cannot be translated into propositions about what states "do;" that states "tend" to balance does not suggest that all states will balance at all times. Nor does a tendency for states to balance mean that states will balance promptly and efficiently. No determinate proposition about what states "do" can be generated from the systemic level, because state behavior is a function of causes at all three levels of analysis. This means that explanations of what states "do" that draw on the systemic level have to recognize the multiple causes of state behavior, and if they want to focus on systemic causes, they have to examine the way systemic causes interact with causes at other levels and theorize about the conditions under which those systemic causes will dominate. This is a problem not only for international relations scholars interested in the balancing behavior of states, but is one faced by all theorists who base their work on systemic theory.

The general nature of the propositions that can be generated from systemic theory can also be seen in the work of offensive and defensive realists. Both claim neorealism as the foundation for their theories, yet they make divergent assumptions about the motivation of states. Defensive realists assume that states are primarily motivated by survival, while offensive realists assume that that states are power-hungry. It is this ability of neorealism to serve as the basis for opposite claims that lead critics to charge that it is essentially unfalisifiable and a degenerating research program.

However, the problem is not with neorealism per se but with how neorealist propositions at the systemic level have been translated into propositions at the unit level. Neorealism itself portrays states as neither exclusively offensive nor as exclusively defensive. It simply assumes that states "are unitary actors who, at a minimum, seek their own preservation and, at a maximum, drive for universal domination." This assumption suffices for explanations of systemic outcomes, but to explain state behavior, more specific assumptions are necessary. The question for proponents of offensive or defensive realism is not which provides the correct interpretation of neorealism, but which school’s auxiliary assumptions about state motivations lead to the most compelling explanations of state behavior.

I suspect that this is not an either/or question, but that each assumption will help us to understand different kinds of state behavior. In other words, the question is not whether all state behavior is best explained by the assumptions made by offensive or defensive realists, but under what circumstances behavior is best explained by the auxiliary assumptions contained in each school’s application of neorealism.

A second problem created by the failure to translate concepts taken from the systemic level is evident in studies of balancing behavior, and that is the lack of discussion of what it means for states to "balance." Work on balance of power theory has taken the systemic propositions that "balances reoccur" and that "states tend to balance" and proceeded to examine whether or not states "balance," with little or no analysis of what sorts of state behavior should fall under that rubric. If we are interested in studying actions that lead to the formation of rough balances of power, we have to inlcude actions that are not normally considered "balancing." Even actions taken by states who seek hegemony must be included, for according to neorealist theory systemic effects are such that even if every state in the system aims for a preponderance of power, a rough balance of power will form.

Analysts of the balancing behavior of states tend to focus on alliances, but that is not adequate for two reasons: First, there are other ways to balance, so that by focusing on alliances we exclude some instances of balancing behavior. Not only does the focus on alliances exclude the many possibilities for internal balancing, but it also ignores other forms of external balancing. For example, Levy argues in a review of Paul Schroeder’s work that external balancing techniques include "the creation of buffer states, territorial partitions, compensations, indemnities, other forms of intervention in the affairs of weaker states, and preventive war."

Second, just as balancing incorporates a lot more than alliances, so too do alliances often include more than balancing. Alliances form for many reasons, some of which may have nothing to do with the existence of an external threat to state security. For example, Deborah Welch Larson argues that states sometimes join alliances for status and prestige, Schroeder argues that alliances can serve management and control as well as defensive purposes, and Randall Schweller develops the possibility that alliances offer opportunities for profit. Not all of these motives for alliance formation are necessarily incompatible with balancing behavior, but nether are they necessarily compatible. This means that if we equate "alliances" with "balancing," we will include some non-balancing behavior. It is impossible to sort out whether a particular alliance is an example of balancing behavior without a clear definition of "balancing."

The lack of a clear definition of balancing behavior, and perhaps even more importantly, of a clear understanding of what is not balancing behavior, inhibits the development of work in the field and leads to the criticisms that work on balance of power theory is ad hoc as well as to the idea that balance of power theorists constantly draw on the indeterminacy and generality of systemic theory to explain away any and all challenges.

I have indicated that performing Singer’s "act of translation" will help to avoid these problems. In order to demonstrate this, the rest of the article will focus on the translation of the systemic concept of balance of power as an outcome to a concept of balancing that will be useful in explanations of state behavior. The first step in this process is to return to the beginning, to systemic balance of power theory.

According to systemic balance of power theory, the formation of balances of power follows more or less directly from the anarchic nature of the international system. A systemic characteristic–anarchy–leads to a systemic outcome–the recurrent formation of rough balances of power. The theory states that any time two or more states who wish to survive exist in an anarchic system, rough balances of power form.

In this sense, systemic balance of power theory argues that anarchy is an underlying cause of the formation of balances of power, just as anarchy is understood to be an underlying or permissive cause of war. But just as a purely systemic explanation of war cannot explain how or why a particular war occurs, a purely systemic explanation of the formation of balances of power cannot explain how or when a particular balance of power forms. Systemic balance of power theory tells us that rough balances of power tend to recur and that any power advantage gained by one state over others will be temporary. The long-term maintenance of a significant disequilibrium of power would contradict the theory, although the precise meaning of the terms "long-term" and "significant" are not clear.

In this way systemic balance of power theory’s argument about the recurrence of balances of power parallels the argument that Waltz makes about the recurrence of war in Man, the State, and War. But the understanding of the international system presented in Theory of International Politics is more developed than that presented in Man, the State, and War, and that development has given rise to some confusion about the kind of claims that systemic theory makes.

In Theory of International Politics, Waltz argues that the system is composed of a structure and of units, and that two characteristics of the structure–anarchy and polarity–have two kinds of effects. The first effect of structure is as an intervening variable. Structure intervenes between the actions and intentions of states and the outcomes that result. This effect of structure means that there is a disconnect between what states intend and the outcomes that are produced. Thus Waltz argues that balances of power form even if states pursue hegemony; the effect of structure also means that two defensively-minded states may nonetheless find themselves at war because the defensive preparations of each are seen as a threat by the other. It is because structure acts as an intervening variable that the intentions of international actors cannot be inferred from outcomes; it also means that complete explanations of international outcomes require at least some systemic component.

The second effect of structure is on the constraints and opportunities that states face. The structure of the international system has this second effect because states, or rather those who act for them, are seen as sentient beings who aim to understand their environment and calculate the best way to act within it. It is this understanding of structure that leads to the concepts of emulation and selection, because in calculating how best to pursue their ends, states are expected to take note of the success and failures of other states, and to imitate those who succeed; states who fail to do this, or who do this less well than others, may be "selected out" by competitive forces.

One example of this kind of effect is the contention that anarchy encourages states to care about relative gains. A second example follows from the introduction of polarity as a characteristic of structure. Changes in polarity (more specifically, from bipolarity to multipolarity or vice-versa) modify both the constraints and opportunities available to states and the outcomes that are likely to result, without, however, changing the basic effects that follow from anarchy. Thus, in both bipolar and multipolar systems, rough balances of power are expected to form. But the opportunities and incentives that states face in bipolar and multipolar systems are different: In multipolar systems states can ally with other great powers to counter a threat, while in bipolarity each great power has to rely on internal balancing to counter the other.

It is this second kind of implication that has lead to much of the confusion about the kinds of claims that systemic theory can make. Because structure affects the behavior of states, systemic balance of power theory is often taken as a theory of balancing behavior. Waltz is often interpreted as saying that neorealist systemic theory can provide predictions or explanations of particular state actions. Elman’s belief that systemic theory can make such predictions lies behind his argument for the possibility of a neorealist, systemic theory of foreign policy, and he argues that Waltz does use his theory to make these sorts of predictions in Theory of International Politics. Vasquez, Haggard, and others agree; all take Waltz’s comments on how structure influences the constraints and opportunities faced by states and interpret them as determinate foreign policy predictions. This is perhaps most evident in Vasquez: He takes Waltz’s discussion of "balance of power theory" and interprets it as a "balancing theory," blurring the distinction between behavior and outcomes that is central to systemic theory. The "fundamental law" of international politics for Waltz is the repeated creation of rough balances of power--a law concerning outcomes. Vasquez interprets it as a law that states will always "balance."

Although structure does affect behavior, it does not determine it. Emulation and selection do not produce uniform state behavior, and systemic effects do not always, or even usually, dominate other sources of state behavior. The incentives and opportunities the system offers for balancing behavior are only one potential source of a state’s behavior. Systemic theories cannot not explain behavior per se, but they can help to account for patterns of behavior that occur despite changes in the identities and motives of the actors.

This is the fundamental reason why systemic neorealist theory is not, and cannot be, a theory of foreign policy, despite Colin Elman’s assertion to the contrary. A theory that includes only systemic variables cannot generate determinate foreign policy predictions under most circumstances. Waltz explains, "An international-political [i.e. systemic] theory can explain states’ behavior only when external pressures dominate the internal disposition of states, which seldom happens. When they do not, a theory of international politics needs help." Thus, when applying neorealism to the study of state behavior, it is necessary to move beyond the systemic level of analysis.

The question that arises when one moves from the study of occurrence of balances of power to the balancing behavior of states is, what exactly is balancing behavior? How do we know it when we see it?

As discussed above, many of the analysts who have used systemic balance of power theory as their starting point have not adequately answered this question. The lack of a clear notion of "balancing" makes it difficult to know an instance of balancing when we see it. For example, is the Concert of Europe an example of a balancing alliance, as it was a group of states who allied to prevent hegemony? Or does it represent a different kind of international politics, the exact opposite of balancing behavior, as Schroeder has argued?

The lack of a clear understanding of what constitutes "balancing" also explains why there has been little investigation of "internal balancing." The general assertion that states acquire arms in response to external threats is widely accepted, but once we examine specific decisions to acquire particular armaments, it becomes necessary to deal with other possible motives for acquiring arms–the interests of defense industries and the armed services, for example. These competing explanations become even more difficult to handle if one includes other, more indirect ways of increasing a country’s power as examples of internal balancing.

So what exactly does it mean for a state to balance? To determine this, it is necessary to translate the systemic notion of "balance of power" into the concept of "balancing."

Systemic balance of power theory tells us that anarchy leads to the formation of balances of power through the competitive international politics that it engenders. Anarchy constrains each state to rely whenever possible on itself for its survival, and to use any and every means at its disposal for that goal. As states compete for the relative power necessary for security, a rough balance of power forms.

Three adjustments have to be made in order to move from this structural understanding of "balance of power" as an outcome to an understanding of "balancing" as a state strategy. The first is that the focus of the inquiry shifts from international outcomes to the motivations behind state behavior. As noted above, it is not possible to determine if a rough balance of power forms and, if it has, argue that states were balancing, because the effect of structure as an intervening variable means that we cannot infer intentions from the outcome. The only way to identify a balancing strategy is to look at the intentions or motivations behind the action. The importance of examining motivations and intentions is reinforced by the observation that states may pursue the same policy for different reasons.

Second, it is necessary to more narrowly define the motivations that are associated with balancing behavior. Systemic balance of power theory tells us that balances result from the interaction of units who wish to survive in an anarchic environment. This suggests that the motivation underlying balancing behavior is survival. However, this is a not a useful possibility, because not all behavior motivated by survival can be considered balancing. After all, the opposite of balancing has been defined as bandwagoning, and bandwagoning is seen as a strategy that smaller states may have to pursue in order to survive. Instead of survival, I define the motivation that characterizes balancing as the desire to counter a threat.

On the basis of these adjustments made necessary by the change in the level of analysis, I argue that "balancing" can best be understood as actions taken by a state to counter an external threat. Defining balancing as actions taken to counter an external threat has three advantages. First, it puts the question of motivation at the center of the analysis. Second, it facilitates an exploration of other possible responses to threat, as well as investigations of the conditions that influence a state’s choice of response. Third, it distinguishes between external sources of threat and other sources. Each of these advantages is explained in more detail below.

The first advantage of this definition of balancing behavior is that it addresses the question of motivations directly. To determine if an alliance is an example of balancing behavior, for example, it is necessary to show that states joined the alliance in order to counter an external threat. Determining the motivations and intentions behind a policy is not always easy. One has to take into account possible attempts by policy makers to obfuscate and mislead, especially since an external security threat is often seen as a useful way to sell a policy preferred for other reasons. But the difficulty of the task does not mean that we should, or can, avoid it.

In his analysis of the fit between neorealist theory and history, Schroeder argues that by focusing on perceptions of threat, Stephen Walt’s balance of threat theory "makes it virtually impossible to distinguish between ‘balancing’ and ‘bandwagoning’ or to determine the real motives of actors, since any ‘bandwagoning’ state is likely to claim that it is actually ‘balancing’ against a threatening enemy." Schroeder emphasizes the difficulties involved in using perceptions and motivations to explain state actions, because policy makers often have an incentive to lie. But in his own analysis Schroeder cannot seem to avoid using either threat perceptions or motivations as part of his explanation for the behavior of states. For example, in the same article Schroeder argues that "almost all German states and principalities saw this move [the proposed exchange of the Austrian Netherlands for Bavaria in 1785] as a threat to the German ‘balance.’" Of course, he does not simply make this claim about threat perception and go on; he provides both evidence and a logical explanation for why that exchange was seen as a threat, and how that threat then motivated the response of the German states and principalities.

The analysis of motivations is an integral part of foreign policy analysis. Putting motivations and threat perception at the center of the definition of balancing decreases the likelihood that explanations of balancing behavior will be based on implicit assumptions about motivations and threats. By focusing our attention on motivations, it will increase our ability to evaluate when policy makers’ claims about threat are used as a smoke screen and when they were actually a motivation for policy.

A second advantage to defining balancing as actions taken to counter a threat is that it allows us to distinguish more clearly between balancing and other possible responses to threat. This in turn facilitates an examination of the circumstances under which states choose to balance rather than bandwagon, hide, or transcend. The identification of the factors that influence how a state decides to respond to threat will help move us beyond sterile definitional debates about whether or not states balance and will help to provide evidence for the usefulness of competing explanations of state behavior.

Second, it may be important to distinguish between different kinds of external threats when studying balancing behavior. There is some suggestion in the literature that British sea power posed a different kind of hegemonic threat than that posed by continental military power, and that this difference meant that states responded differently to British hegemony than they did to attempts for hegemony on the continent. The characteristics of economic threats may also differ from the characteristics of military threats in ways that are significant for how states chose to balance. Threats posed by states would also seem to differ in significant ways from the sort of collective international environment threats that have begun to attract attention. An examination of differences among these different kinds of threats will reveal if, and if so how, the characteristics of the threat faced shapes the response of states.

The definition of balancing proposed above suggests a general model of balancing behavior. In order to classify an action taken by a state as balancing according to this definition, it is first necessary to demonstrate that the action was taken in response to a perceived threat, and then to argue that the action was taken in order to counter (instead of to appease, hide from, or align with the source of) the threat. When we discuss balancing, bandwagoning, or other types of response to threats, we are actually talking about a) objective conditions that are possible sources of threat, b) the process of perception whereby those conditions are interpreted as threat, and c) a decision about how to respond to the perceived threat. Thus a model of balancing and other responses to threat must look something like this:

Insert Diagram 1

This model focuses on and separates out the different processes that systemic balance of power theory "black-boxes" and that previous discussions of balancing behavior have obscured. It focuses on states’ balancing behavior, not on the formation of a rough balance of power. It separates the question of how states respond to threat from the question of possible sources of threat from the question of when states perceive threat, and allows us to isolate and compare hypotheses about the variables that may be important at each stage. For example, a cursory examination of the balancing literature suggests competing hypotheses about possible sources of threat, about the factors that determine how a state responds to threat, and about the ways in which states respond to threat. (See Chart 1).

The lack of a precise definition of balancing behavior, and the failure to separate out these stages or processes, has led to much confusion in the balancing literature. For example, while it is clear that Waltz, Walt, and Schroeder disagree about the balancing behavior of states, it is not easy to identify exactly where they disagree. The general model of balancing presented here helps to pinpoint their disagreements. For example, Waltz and Walt disagree about potential sources of threat; Waltz focuses exclusively on aggregate power while Walt includes factors such as intentions and geographic proximity. However, they largely agree about the response of states to threats; they both focus on balancing versus bandwagoning, and they both expect that states will balance when they can, though they recognize that weak states may have to bandwagon. Schroeder’s work, however, suggests that suggest that factors other than power capabilities determine a state’s response to threat, and he argues that it is possible for states to respond to threats in many different ways.

The model of balancing presented here encourages analysts to be clear about the factors that are important at each stage. This means that in our studies of balancing behavior we will have to move beyond assertions that states do or do not balance, to arguments about the particular factors that are seen as giving rise to threat, threat perception, and the decisions of states about how to respond to threat.

As chart one makes clear, conspicuous by its absence is any work in the balance of power literature on the factors that affect threat perception. Although the process of threat perception is understudied in international relations generally, the work that has been done has not been integrated into analyses of balancing behavior. Factors affecting threat perception may include characteristics of the threatened state’s leaders and their decision process (e.g. motivated and unmotivated biases, as discussed by Jervis), characteristics of the relation between states (e.g. a high-level of tension and distrust, as discussed by Cohen), or the presence of other threats (e.g. a possible threat in one area may be overlooked because state leaders are focused on another threat). My hope is that the definition and model of balancing presented here will encourage more investigation into the factors that lead to threat perception.

While this article has generally been a defense of balance of power theory, in the sense that I have argued that many of the problems its critics point to are problems with the uses that have been made of the theory and not with the theory itself, it may strike some as a strange defense. The defense, after all, has been based on the limited scope of systemic balance of power theory--its inability to do anything more than explain recurrent patterns of outcomes and to identify incentives and constraints facing states. But these same limitations are clear in Waltz’s Theory of International Politics, and have been noted by other observers as well.

These limitations of systemic balance of power theory should not be allowed to obscure its contribution. Systemic balance of power theory accounts for the recurrent formation of rough balances of power, and systemic theory generally helps to explain the disjuncture between intentions and outcomes that routinely characterizes international politics. It identifies the ways in which the structure of the international system shapes the incentives and opportunities facing states. But systemic theory by itself cannot explain particular outcomes nor provide an explanation for the behavior of states; much difficult work remains to be done in applying that theory to the behavior of states.

A first step toward that application has been taken in this paper. It provides a definition of balancing behavior that is consistent with systemic theory yet leads to a general model of balancing behavior that allows us to draw on factors at other levels of analysis. It remains for future work to apply this model of balancing behavior, and to assess the relative explanatory power of the possible variables at each stage of the model.

Notes