Security Studies - Special Issue: Debating Glaser's "Rational Theory of International Politics" (2011)

Security Studies, 20:416–423, 2011 Copyright © Taylor & Francis Group, LLC ISSN: 0963-6412 print / 1556-1852 online DOI: 10.1080/09636412.2011.599189

Dilemmas About Security Dilemmas ROBERT JERVIS

Ever since Thucydides students of international politics have focused on the security dilemma as a cause of conflict.1 Defined as the unfortunate fact that policies designed to increase the state’s security often have the effect of decreasing the other’s security (even though this is not desired or intended), at first glance this appears to be more a problem than a dilemma. The latter aspect appears once we realize that while arming (or making alliances) can end up heightening tensions, making the state less secure, and even producing an unnecessary war, the failure to take such precautions can endanger the state if it is faced by adversaries who will take advantage of its weakness. So the question of how many and what kind of arms a state should procure not only presents trade-offs with consumption and investment, but confronts the state with difficult judgments about how to maximize its security. Over the past fifteen years, Charles Glaser has been wrestling with the security dilemma, what it tells us about international politics, and how leaders should react to it, and Rational Theory of International Politics pulls together and extends this body of work.2 As is his wont, Glaser proceeds with great care and thoroughness, generously acknowledging what others have said. He makes major contributions by laying bare basic arguments that had become obscured and pushing forward current lines of argument, but this kind of enterprise necessarily excludes positions outside its purview and runs the risk of accepting too quickly some propositions that need further probing.

Robert Jervis is Adlai E. Stevenson Professor of International Politics at Columbia University. President of the American Political Science Association in 2000–01, his most recent book is Why Intelligence Fails: Lessons from the Iranian Revolution and the Iraq War (Cornell University Press, 2010). 1 For other recent treatments, see Ken Booth and Nicholas Wheeler, The Security Dilemma: Fear, Cooperation, and Trust in World Politics (New York: Palgrave Macmillan, 2008); and Shipping Tang, A Theory of Security Strategy for Our Time: Defensive Realism (New York: Palgrave Macmillan, 2010). 2 Charles L. Glaser, Rational Theory of International Politics: The Logic of Competition and Cooperation (Princeton: Princeton University Press, 2010), subsequent references cited in text.

416

Dilemmas about Security Dilemmas

417

Downloaded by [Florida International University] at 17:19 17 September 2011

GLASER’S ARGUMENT Although he does not propose a formal model, Glaser’s logic is very much in the spirit of rational choice in that his objective is normative in the sense of saying how a rational leader who seeks security should navigate the difficult waters created by the security dilemma. Given the challenging nature of the international environment, the incentives of leaders to get it right, and the fact that leaders are usually quite sophisticated, there is reason to expect that the theory will also give a good account of what they do. Even when it does not, it will provide a baseline from which to detect deviations that call for explanations, most obviously in terms of domestic politics or decision-making pathologies. Of course Glaser realizes that such a normative argument cannot be tested directly by looking at national policies and international outcomes, so he makes use of thought-experiments and counterfactuals. By necessity, these are speculative, but they spell out for the reader the implications of his argument and, by and large being plausible, provide support for his theory. His starting and fundamental point is that an understanding of the security dilemma leads us to see that “the international system does not consistently favor competitive policies” (7). While this may seem obvious to defensive realists, in fact it is both contested and important: it is not uncommon for critics of realism to claim that this school of thought expects and preaches belligerence, and from the other end of the continuum John Mearsheimer does indeed argue that under most conditions even states that seek security have no choice but to try to reduce others’ power.3 But Glaser is clearly right that competition is costly in consuming resources and dangerous because it can increase conflict and make the state less secure. Glaser argues that there often are fairly reliable ways for a state to judge whether others are a menace and, if it so desires, for the state in turn to reveal that it is benign. I will return to problems with this argument, but want to note here that the security dilemma is also driven by the fact that even if the other is benign today, it can turn malign in the future, a nasty possibility that leads some to say that states should and must make worst-case assumptions, which inevitably leads to severe competition. Glaser correctly notes that such blanket pessimism is unwarranted. The fact that a more cooperative policy has costs and risks does not mean that these are greater than those entailed by more competitive ones. Furthermore, Glaser’s theory points to the importance of timely warning, a concept underpinning many arms control agreements. The grim facts of uncertainly about current motives and the possibility that they can change means that states need to track indicators of the threat posed by the adversary. But this will do the

3

John Mearsheimer, The Tragedy of Great Power Politics (New York: Norton, 2001).

Downloaded by [Florida International University] at 17:19 17 September 2011

418

R. Jervis

trick only if the state has time to put in place protective measures, most obviously by increasing its arms. Even in the face of significant mistrust, then, cooperation is possible if arrangements can be made so that each side has reason to believe that it can respond appropriately after detecting that the other has become troublesome. More specifically, Glaser argues that the general outlines of a state’s security policy should be guided by three kinds of variables: motives, material factors, and information. Motives flow from the state’s values, interests, and goals, and the key distinction is between states that primarily seek security and those that are “greedy,” a formulation to which I will return. Material factors focus on “hard power,” which means mainly the military power that a state has or can readily develop and whether offensive capabilities are stronger than and different from defensive ones. Informational variables concern what the state knows and can know about the other’s motives and power, and involves questions of how states can accurately signal each other, especially when they want to establish a cooperative relationship. These factors of course are not unmoved movers, and Glaser explicitly puts aside exploring what determines them. Some might argue that this is what we need to focus on, but I agree that even if we take the state of these variables as a given, there is plenty of work to be done. The discussion is so clear, full, and rich that all those concerned with security policy will want to study it at length for themselves. So rather than provide more summary, I will turn to questions and lines of criticism. One set focuses on the school of thought that the book belongs to and another more narrowly involves the key variables.

PROBLEMS WITH THE REALIST VIEW OF SECURITY Most obviously, objections to the general framework of realism apply to this book. It would not be fair to expect Glaser to engage them fully; if he did, he would not have time to develop his own theory. But it is at least worth mentioning that some goals and motives are omitted here, as they are from most realist accounts. The desire to see one’s ethnic group thrive (or dominate) is associated with civil wars, but can be a cause of international conflict as well. Indeed, it can be argued that the cause to which Hitler was dedicated was not Germany, but the Aryan race. Here the unit of concern shifts, but much of Glaser’s framework might be retained. I think this would be more difficult when states and individuals are driven by nonmaterial goals such as honor, glory, or altruism. These break out of most of our standard frameworks and yet play powerful roles.4 Concern 4 Donald Kagan, On the Origins of War (New York: Doubleday, 1995); and Richard Ned Lebow, A Cultural Theory of International Relations (New York: Cambridge University Press, 2009).

Downloaded by [Florida International University] at 17:19 17 September 2011

Dilemmas about Security Dilemmas

419

for respect and status can rationally lead to behavior that is quite different from that which is generated by Glaser’s theory, and to the extent that many countries, including the leading powers, now have a high degree of security, these motives may become even more important.5 A second limitation is also widely shared. Like most abstract analyses of security policies, Glaser simplifies the world by imagining that it consists of only two actors. For obvious reasons, we often did so during the Cold War. Furthermore, many of our analytical tools cannot readily handle multiple actors. While discussions of multipolarity and, relatedly, of alliance bargaining necessarily have to pivot on the role of third actors,6 and studies of specific phenomena like proliferation and some dynamics (such as the domino effect) crucially involve many players, a theory like Glaser’s would probably become impossibly complicated if it widened its aperture. At some point, however, we are going to have to take on the difficult theoretical and empirical problems that emerge when we do so. In the contemporary world, we cannot fruitfully analyze the security choices the United States faces in its policies toward Iran and North Korea without considering many of the other states involved. Indeed, the basic American motivation is the threat to its friends in the regions, not to itself directly, and pressure in the form of economic sanctions can be effective only if many other states join in them. Like much realism, Glaser’s study downplays the importance of domestic politics and ideologies. To the extent that these are basic causes of the state’s motives and power, Glaser can put them aside because he explicitly excludes looking at the causes of these variables. But if they are basic drivers of security problems, the cost of the exclusion is high.7 Glaser also scants the influence of ideologies and domestic sources of foreign policy in his treatment of the information one state has about another. For example, when he looks at the end of the Cold War he concentrates on shifts in American perceptions of Soviet power and motives (207–12), but he minimizes the extent to which these were strongly influenced by the belief that the Soviet domestic system was changing. At bottom, many Americans believed, 5 Robert Jervis, American Foreign Policy in a New Era (New York: Routledge, 2005), chap. 1; and Deborah Larson and Alexei Shevchenko, “Status Seekers: Chinese and Russian Responses to U.S. Primacy,” International Security 34, no. 4 (Spring 2010): 63–95. Randall Schweller’s more fundamental criticism in his contribution to this symposium is consistent with this line of questioning. In his contribution, James Fearon makes the related point that once we realize that the fundamental unit is the individual rather than the state, we see that the former can have goals other than the aggrandizement and even the security of the latter. Also see Barry Buzan, People, States and Fear: An Agenda for International Security Studies in the Post-Cold War Era (New York: Harvester Wheatsheaf, 1991). 6 For one of the best examples see Glenn Snyder, Alliance Politics (Ithaca: Cornell University Press, 1997). 7 For a strong argument about the role of conflicting ideologies as a fundamental cause of international conflict see Mark Haas, Ideological Origins of Great Power Politics (Ithaca: Cornell University Press, 2005). Neoclassical realism stresses the need to keep domestic factors in the analysis: see, for example, Steven Lobell, Norrin Ripsman, and Jeffrey Taliaferro, ed., Neoclassical Realism, the State, and Foreign Policy (New York: Cambridge University Press, 2009).

420

R. Jervis

the Soviet Union had been expansionistic because it was a power-hungry dictatorship permeated by an ideology proclaiming the duty and necessity of spreading revolutions throughout the world. When Mikhail Gorbachev became less dictatorial and abandoned the class struggle as the foundation for Soviet foreign policy, the American perception of threat rationally decreased enormously.

Downloaded by [Florida International University] at 17:19 17 September 2011

PROBLEMS OF INFORMATION AND MOTIVES Let me now turn from questions about the framework of defensive realism within which Glaser operates to those internal to this approach. Informational problems are central to the security dilemma, and so one would think that there would be an extensive literature on threat perception and the related subject of how states can show that they are not threats. In fact, there is not. The problem is not that Glaser’s analysis is not state of the art, but that the state of the art itself is weak; while his discussion of how he thinks information should be used has much to be said for it, there are severe inherent limits here. The concept of costly signaling has major difficulties,8 and the degree of ambiguity in states’ behavior is likely to remain high. When a state forgoes offensive capability, is it showing that it is a security seeker or that it is just biding its time, perhaps building up defenses to a point at which it can more safely prepare for expansion, as many members of the Ronald Reagan administration (but eventually not Reagan himself) believed Gorbachev was doing? Today what could Iran do to show that its (presumed) desire for nuclear weapons is motivated only by the quest for security, and what could the United States do to show that it is willing to respect Iran’s vital interests? The North Koreans do not appear to have paid much attention to the hard-won security guarantees that the United States provided in the hope that they would lead to a deal over Korea’s nuclear program.9 Should North Korea have been reassured? Should it be reassured or frightened by the fact that the United States recently turned over more military authority to South Korea? In parallel, the debates about Soviet motives have outlasted the opening of many documents after the Cold War. Difficult problems plague theory building (and leaders) here, and the normative and the empirical dimensions are intertwined. The choice of what signals one state should send to make a desired impression on the other depends on how the other will perceive them and, conversely, the inferences the state should draw from the other’s behavior depends on its theories 8 I have discussed some of them in “Signaling and Perception: Drawing Inferences and Projecting Images,” in Political Psychology, ed. Kristin Monroe (Mahwah, NJ: Earlbaum, 2002), 293–312. 9 Victor Cha, “What Do They Really Want? Obama’s North Korean Conundrum,” Washington Quarterly 34, no. 4 (October 2009): 119–38.

Downloaded by [Florida International University] at 17:19 17 September 2011

Dilemmas about Security Dilemmas

421

(or hunches) about international relations in general and the other state in particular. Those who hold second-image theories will believe that the nature of the other’s domestic regime tells a great deal about how it will behave, for example. In large measure, this explains the divergent views of the danger posed by Saddam Hussein’s Iraq held by George W. Bush and his colleagues on the one hand and most academic realists on the other, and I do not think that a rational theory can prove that Bush is wrong and we are right. When Soviet First Secretary Nikita Khrushchev made deep and unilateral cuts to his army, American Secretary of State John Foster Dulles was not impressed, claiming that this would only cut waste and allow the Soviet economy to grow faster, thereby presenting a greater threat in the future. This response was driven by his irredeemably suspicious view of Soviet motives,10 but he may even have been right, and certainly we cannot say that the only rational response would have been to conclude that Khrushchev was not a threat. Most discussions of security policies in general and the security dilemma in particular talk in terms of states’ intentions. Glaser instead talks of motives. Although not without its difficulties, the shift in terminology is welcome. In parallel, he abandons the common categorization of states as status quo on the one hand or expansionist or aggressive on the other, substituting the terms “security seekers” and “greedy states.” Glaser’s essential point is that the more common terms are not only value-laden, but refer to behavior, which is troublesome in disguising the fact that the same behavior can be caused by different motives and the same motive can lead to different behaviors depending on the circumstances. Most obviously and most commonly, a state that wants to maintain or increase its security can believe that this requires it to expand at the expense of others. The state is a security seeker, but this motive leads it to be aggressive. In one sense, the security dilemma does not apply here since the state foresees and intends that its policies will make others less secure. But in a deeper sense, this does exemplify the security dilemma because the state seeks security and expansion is not valued intrinsically, but only as the necessary path to security. That is why Mearsheimer, who sees dynamics of this kind as dominating, calls his book “the tragedy of great power politics.”11 It is a tragedy because states often want security, but they cannot reach this goal through the cooperative mechanisms that defensive realists like Glaser believe often should be employed. Just as Robert Powell clarified the relative versus absolute gains problem by distinguishing between preferences over strategies and preferences over

10 Ole Holsti, Making American Foreign Policy (New York: Routledge, 2006), 28–29; also see his “Cognitive Dynamics and Images of the Enemy: Dulles and Russia,” in Enemies in Politics, David Finlay, Ole Holsti, and Richard Fagen (Chicago: Rand McNally, 1967), 25–96. 11 John Mearsheimer, The Tragedy of Great Power Politics (New York: W. W. Norton, 2001).

Downloaded by [Florida International University] at 17:19 17 September 2011

422

R. Jervis

outcomes,12 Glaser shows that we must distinguish between goals (for example, expansion) that are sought because they have intrinsic value and those that are instrumentally valued as a means toward higher goals (for example, security). The analytical difficulty, of course, is that many goals partake of both. For example, during (and after) the Cold War the United States sought to replace radical regimes to bolster its own security, to increase the influence it could wield for other purposes, and because it believed that such regimes were intrinsically evil. Intentions and motives often are combined or blurred here. What states have to care most about in setting their security policies is how the adversary is likely to react to alternative policies that they could follow. Although leaders rarely put it quite this bluntly, what they want to know is whether the other side will cooperate if their state does. Separating motive from intent can be quite useful here because it makes us see that drawing inferences about hostility may be more complicated than we would think at first. Analysis is likely to be subject to dispute by rational observers, however. Thus during the Cold War, hawks were quick to argue that high Soviet defense spending pointed to that country’s intention to create and exploit American vulnerabilities. But it is also possible that high levels of Soviet spending and arms indicated great fear of the United States. Today, most of those who believe that Iran is moving as rapidly as possible to acquire nuclear weapons assume that the purpose is to dominate the region, expel American influence, and menace if not destroy Israel. But the intent to go nuclear could be motivated by the desire to be secure. The difficulty—with both the theory and international politics itself—is that if the adversary believes that its security requires expansion, then from the state’s perspective it does not matter that from a motivational standpoint the adversary is a security seeker rather than a greedy state. For example, one analysis of the Cold War holds that while both the United States and USSR were security seekers, each believed that there could be no stability in an ideologically heterogeneous world and that the other side would inevitably try to undermine it, thus making cooperation impossible. In cases like this, although mutual security might be desired, it cannot be reached, or at least cannot be reached by normal means. One of the virtues of Glaser’s analysis is that it leads us—and diplomats—to think about possible responses to the difficult situation of benign motives producing hostile intent. If in some cases states will simply have to treat the adversary as an enemy, in others it may be possible to convince the adversary that there are alternative paths to its security. Even when the normal methods of seeking cooperation under the security dilemma such as dividing up a large transaction into a series of small 12 Powell, “Absolute and Relative Gains in International Relations theory,” American Political Science Review 85, vol. 4 (December 1991): 1,303–20; and “Anarchy in International Relations Theory: The Neorealist-Neoliberal Debate,” International Organization 48, no. 2 (Spring 1994): 313–44.

Downloaded by [Florida International University] at 17:19 17 September 2011

Dilemmas about Security Dilemmas

423

ones, bolstering the advantages of defense, or providing transparency and other safeguards against cheating would not be sufficient, it may be possible to change the beliefs and worldviews that have convinced the adversary that competition is the best if not the only route to security. Indeed, in considerable measure this is what Secretary of State George Shultz tried to do in his conversations with Gorbachev and his foreign minister, Eduard Shevardnadze in the last years of the Cold War, and the end of the conflict owed more than a little to these efforts. Glaser clearly shows that while cooperation is not always the best policy for a security seeker, neither does the structure of international politics invariably rule it out. Even if I am right that there are problems with it, his theory is a big step toward understanding why and when sensible leaders can cooperate in this uncertain world.

Security Studies, 20:424–430, 2011 Copyright © Taylor & Francis Group, LLC ISSN: 0963-6412 print / 1556-1852 online DOI: 10.1080/09636412.2011.599190

Realists as Idealists JOHN MEARSHEIMER

Charles Glaser established himself as one of the leading defensive realists in the mid-1990s when he published an important article entitled “Realists as Optimists.”1 Rational Theory of International Politics is a sophisticated extension of that article.2 There is much to admire in Glaser’s new book, as he employs his first-rate analytical mind to make arguments that are comprehensive as well as logically sound. Nevertheless, his theory has shortcomings that limit its usefulness as a guide for understanding international politics. There are two ways to assess any theory of international relations. One way is to analyze its internal logic: its core concepts, assumptions, and logical consistency. Alternatively, one can take the theory as it is and test it against the historical record: how well does it explain the past and present behavior of states, and how useful might it be for explaining future actions? I have disagreements with some of the main elements in Glaser’s theory. For example, I do not think that the offense-defense balance, which is at the heart of his theory, is a useful concept. Nor do I agree with Glaser’s contention that it is possible for states to signal their intentions to each other in meaningful ways. But I would like to put aside my criticisms of the theory’s underpinnings and instead evaluate it on its own terms. In what follows, I take the theory as given and ask how useful it is as a guide for understanding how states can maximize their security in the real world.

THE GREAT TRANSFORMATION PROBLEM Taken on its own terms, Glaser’s theory has two problems that limit its usefulness. First, because his theory is normative—not explanatory—there John J. Mearsheimer is the R. Wendell Harrison Distinguished Service Professor of Political Science and the codirector of the Program on International Security Policy at the University of Chicago. 1 Charles L. Glaser, “Realists as Optimists: Cooperation as Self-Help,” International Security 19, no. 3 (Winter 1994–95): 50–90. 2 Charles L. Glaser, Rational Theory of International Politics: The Logic of Competition and Cooperation (Princeton: Princeton University Press, 2010), subsequent references cited in text.

424

Downloaded by [Florida International University] at 17:20 17 September 2011

Realists as Idealists

425

has to be a fundamental transformation in how policy makers think about international politics for his theory to operate. But Glaser does not tell us how this radical shift might occur, and given that so much past behavior by the great powers contradicts his theory, there is good reason to think that little will change in the future. Glaser makes it clear from the outset that his theory “analyzes the strategies that a state should choose—which is essentially the same as assuming that the state is a rational actor” (2). He later writes, “This book develops a rational theory—a theory of what states should do to achieve their goals, given the constraints they face; in this sense, it is a prescriptive, normative theory” (23). Significantly, he does not claim that his theory—which has a rational actor assumption at its core—has much explanatory power, because it does not. Instead, it provides a guide for how rational states should act toward each other under different sets of circumstances. What is going on here? When Glaser and other defensive realists look at the historical record and consider how the great powers have behaved over time they quickly recognize that those states have often behaved in ways that run counter to what their theories predict. In other words, their realist theories do a poor job of accounting for the past and present actions of the major powers in the international system. To put the point somewhat differently, great powers often behave in ways that the defensive realists consider reckless rather than rational. In practice, states frequently do not operate as strategic calculators, which obviously contradicts what one would expect from a realist theory built around a rational actor assumption.3 The root of the problem is that defensive realists believe that the structural constraints of the international system should discourage great powers from adopting aggressive strategies and especially from starting wars with each other. For them, the structure of the system should push states to behave defensively and to maintain rather than upset the balance of power. Of course, this logic explains why realists can be optimists. In practice, however, this is not what the world looks like; great powers often behave aggressively and are prone to engage each other in intense security competitions that sometimes lead to devastating wars. For the defensive realists this is reckless behavior that contradicts their deeply held belief that the structure of the international system is largely benign. All of this foolish and dangerous behavior explains why Kenneth Waltz, the preeminent defensive realist, insists that his theory of international politics does not employ a rational actor assumption. It also explains why he

3 I interpret the historical record differently from the defensive realists. Where they see states acting recklessly, I often see them acting as strategic calculators. Of course, states acting rationally sometimes miscalculate and end up worse off, not better off. John J. Mearsheimer, The Tragedy of Great Power Politics (New York: Norton, 2001). The analysis in this article assumes that Glaser’s description of the past is correct.

Downloaded by [Florida International University] at 17:20 17 September 2011

426

J. Mearsheimer

maintains that his theory does not explain state behavior, but instead explains international outcomes. He argues that a separate theory of foreign policy is needed to explain state behavior, although he does not provide one.4 Consider the work of two other prominent defensive realists, Jack Snyder and Stephen Van Evera. They also see much evidence of reckless behavior on the part of the great powers, but unlike Waltz, Snyder and Van Evera try to explain it.5 Each of them has invented what is in essence a compound theory comprised of: (1) a realist component that can account for those cases where a great power recognizes the dangers of pursuing aggressive policies and consequently shows the appropriate restraint in its behavior, and (2) a unit-level component that accounts for the reckless behavior so often displayed by the great powers in the past. Their theories, in other words, aim to account for both strategic and nonstrategic behavior. Snyder and Van Evera fit neatly in the Waltzian tradition because the realist component of their theories is essentially a theory of international politics while the unit-level component of their theories is in effect a theory of foreign policy. Unlike Snyder and Van Evera, Glaser does not offer a compound theory with a realist and a unit-level component, because he is not attempting to explain how the great powers have behaved in past and recent times. Like Waltz, he simply offers a systemic theory or what Waltz calls a theory of international politics. There is, however, an important difference between them: as noted, Waltz maintains that his theory can explain international outcomes, even if it cannot account for state behavior. Thus, Waltz is claiming that his theory has a great deal of explanatory power, although its domain is limited. Glaser, on the other hand, does not make this distinction between state behavior and international outcomes and instead concedes that his theory has little explanatory power. Thus, he is left with little choice but to tell us how states should behave, not how they actually have acted toward each other over time. His normative theory, in other words, simply provides a rational baseline for assessing how states should think about the strategic choices facing them. This discussion of the normative nature of Glaser’s theory raises the allimportant question: will he be able to convince future policy makers to think about the world in the manner he prescribes? The problem he faces is that history is littered with examples of states that have not acted according to the dictates of his version of defensive realism. Indeed, the fact that countries have frequently acted in reckless ways probably means that nonstrategic

4 Kenneth N. Waltz, Theory of International Politics (Reading, MA: Addison-Wesley, 1979), 60–73,118, 121–23, 128; and John J. Mearsheimer, “Reckless States and Realism,” International Relations 23, no. 2 (June 2009): 241–56. 5 Jack L. Snyder, Myths of Empire (Ithaca: Cornell University Press); and Stephen Van Evera, Causes of War (Ithaca: Cornell University Press, 1999).

Downloaded by [Florida International University] at 17:20 17 September 2011

Realists as Idealists

427

behavior is deeply wired into the international system. Thus, for Glaser’s theory to be useful, there will have to be a great transformation in state behavior. In effect, he has to sell his ideas to foreign policy elites across the world and those ideas have to have continuing influence over time. Otherwise, his theory will have little impact outside of academia, which is not what the author of any normative theory wants. Glaser, however, has no explanation for how to transform the behavior of states and get them to act as strategic calculators. The reason he has no story about how his theory might catch on in policy-making circles is that he has no theory of state behavior. If he had such a theory, as Snyder and Van Evera do, he could explain when states are likely to act according to the dictates of defensive realism and when they are likely to act recklessly. Snyder, for example, argues that cartelized political systems in which powerful interest groups logroll with each other are especially prone to pursuing foolish foreign policies, while democracies and unitary oligarchies are likely to act more intelligently and not pursue aggressive and dangerous policies. Thus, if democratization was inexorably spreading across the globe, as Francis Fukuyama famously claimed when the Cold War ended, one could expect more and more states to think and act like upright defensive realists, which would lead to a more peaceful world.6 But that is Snyder’s account, which grows out of the unit-level component of his theory. Glaser, however, has no such story, and thus no explanation for how to cause the great transformation that his theory demands.

THE FIRST-MOVER PROBLEM There is another problem with Glaser’s theory that concerns the first countries that might adopt his theory. He believes that reckless behavior has been commonplace in international politics, which is not to say that he thinks that every action by every great power has been strategically unwise. But there are many examples of misguided state behavior in his story, including the glaring cases of Imperial Germany, Imperial Japan, and Nazi Germany, which all started wars that ended in their destruction. This means that the first states that decide to behave according to the dictates of Glaser’s theory will be doing so in a world that is populated with other countries that are given to behaving recklessly. This is a fundamentally different situation from a world in which almost all of the other states are committed to acting like defensive realists. Glaser maintains that rational states will usually emphasize cooperation over competition. When they do compete, they should do so in a restrained

6

Frances Fukuyama, “The End of History?” National Interest (Summer 1989).

Downloaded by [Florida International University] at 17:20 17 September 2011

428

J. Mearsheimer

fashion. There is little room for war and no room for the pursuit of hegemony in his theory. There is not much question that if all or nearly all of the states in the system embraced his theory, the world would be much more peaceful. But the problem is: how do you get to that point? It is unlikely that all of the great powers will simultaneously undergo an epiphany and start acting like defensive realists. It is much more likely that one or maybe two states will find Glaser’s logic compelling and think seriously about adopting it. But these first-movers will be living in a world where some number of powers will continue to act recklessly. What does this mean for the first-movers? How does it affect their behavior? It is worth noting that Glaser maintains that his theory “prescribes the strategy, or at least narrows the range of strategies, that a state should pursue when facing a rational adversary” (270). The problem, however, is that first-movers cannot assume they will be facing rational adversaries. On the contrary, as first-movers they face a world where reckless states are still a very real possibility. There is good reason to think that the first-movers would often have to behave offensively, not defensively, to ensure their security, thus perpetuating interstate competition and conflict. Specifically, in a world in which there is significant potential for the great powers to act recklessly, those countries that are rational have strong incentives not just to preserve the balance of power, but also to take concrete measures to increase their share of world power for purposes of self-protection. Indeed, states in such a dangerous situation often have good reasons to pursue risky—which is not to say foolish—strategies to gain additional increments of power. In fact, aggression may sometimes be the smart strategy for states simply worried about their survival in such a world. A first-mover operating in a system where there might be powerful but misguided adversaries runs the risk that one or more of those dangerous adversaries might attack it, possibly with the aim of destroying it. Even if no rival great power seems to fit that profile at the moment, a state can never be certain that will always be the case, especially since Glaser maintains that reckless states show up with some frequency in the existing international system. A savvy first-mover will therefore be constantly thinking about how best to prepare itself for the possible appearance of a dangerous opponent. As a good defensive realist, Glaser would probably advise a threatened state to build a balancing coalition in the event that an aggressive adversary appears on the scene. However, a first-mover faces two major obstacles to crafting an effective balancing policy. It is more difficult to make deterrence work when dealing with powerful and reckless states rather than strategic calculators. By definition, misguided states will sometimes pursue policies that violate strategic logic. That means they might initiate a war in circumstances where a rational state would sit tight and not start a fight. This is because domestic political considerations are likely to push them to pursue

Downloaded by [Florida International University] at 17:20 17 September 2011

Realists as Idealists

429

strategies that are unnecessarily risky. They are also likely to have more than the usual amount of trouble that all countries face when they assess the balance of power as well as the systemic constraints and incentives facing them. After all, they are not rational states. Furthermore, a threatened first-mover cannot be certain that its potential balancing partners will be there when they are needed, if only because balancing coalitions usually face significant collective action problems. In addition, there is always the possibility that those allies might behave foolishly themselves, in which case the balancing coalition would not come together in time to deter a potential aggressor. One might concede that balancing is not always efficient and that war sometimes results, but argue that a balancing coalition would eventually come together and the threat would be contained or eliminated in the end. That is possible, maybe even likely, but hardly guaranteed. Moreover, the first-mover would still have to fight a war to check the aggressor, and that undertaking would involve huge costs, which a prudent state would surely want to avoid. Given the difficulties of making containment work effectively in a world where there might be some number of reckless great powers, the best way for a first-mover to protect itself is to be especially powerful. Striving to be the preponderant power in the system—which is not to say attempting to become a hegemon—would appear to be a sensible policy. In addition, that first-mover should be willing to pursue risky strategies to gain more power or retain the power advantage it has over other states; it should have opportunities to acquire more power, because putting together a balancing coalition that might deter the first-mover will be especially difficult in a world where most of the great powers are prone to act in foolish ways. In particular, preventive war should be a serious option for a rational state facing a rising power that might one day foolishly aspire to be a hegemon. The logic here is straightforward. The more powerful a first-mover is relative to the other states in the system, the less likely it is that a reckless state would attack it. There is no guarantee that a state prone to foolish behavior would not start a losing war, but it is less likely if that potential aggressor is badly outgunned. Plus, if deterrence fails and there is a war, the first-mover would be well positioned to win it quickly and decisively. Finally, a rational state that is the preponderant power in the system is likely to be able to contain a misguided aggressor by itself and not need to rely on a balancing coalition to do the job. This takes the problem of inefficient balancing off the table, as the first-mover no longer has to worry about unreliable allies. First-movers face another troubling dilemma. If they act offensively to protect themselves from their reckless adversaries, there is a good chance that those reckless states will feel more threatened and respond even more aggressively. A rational state bent on defending itself may do its best not to provoke the reckless states in the system, but it will find that goal hard

430

J. Mearsheimer

to achieve. The main reason is the familiar security dilemma, which says that measures a state takes to increase its own security usually decrease the security of other states. In spite of this tendency, or perhaps because of it, potential first-movers will sometimes go on the offensive to try to gain power at the expense of their reckless foes and thus increase their prospects for survival.

Downloaded by [Florida International University] at 17:20 17 September 2011

THE LIMITS OF IDEALISM The bottom line is clear. It may be rational to act defensively in a world where most or all of the great powers have accepted Glaser’s logic, but in a world of first-movers surrounded by reckless states, it sometimes makes sense to act offensively. Such behavior, needless to say, is not going to lead to the benign world that is at the heart of defensive realism. This tragic reality will make it hard for Glaser to convince world leaders to follow his theory’s prescriptions. His only hope would be to convince almost all of them to adopt his theory at essentially the same time. Otherwise, he is stuck with the first-mover problem. But Glaser has no explanation for how he can convince any policy makers to act like defensive realists, much less get the majority of them to do so simultaneously. Until he comes up with a clever marketing strategy that can address that fundamental problem, his theory is likely to be of interest only to his fellow academics. There is nothing wrong with that, of course, although one suspects that Glaser was hoping for something more.

Security Studies, 20:431–440, 2011 Copyright © Taylor & Francis Group, LLC ISSN: 0963-6412 print / 1556-1852 online DOI: 10.1080/09636412.2011.599192

Two States, Two Types, Two Actions JAMES D. FEARON

I agree with Barry Posen’s and William Wohlforth’s book jacket comments describing Rational Theory of International Politics (RTIP ) as a “must read” for students of international relations theory.1 In this brief comment I will first explain why and then raise two concerns I have with the approach Glaser takes. RTIP provides the most careful and extensive informal development of a simplified model of international politics that I think of as the “Jervis model,” a set of arguments and assumptions suggested by Robert Jervis in his 1976 book Perception and Misperception in International Politics and his 1978 World Politics article “Cooperation under the Security Dilemma.” The model could also be called the “Kydd model,” after Andrew Kydd’s formal statement and development of it in his 1997 World Politics article and 2005 book. Or it could be called the “Glaser model,” after Glaser’s analysis in RTIP and in precursor articles going back to 1992. Or you could name it after all three, as they all deserve substantial credit.2 In this model for thinking about interstate cooperation and conflict, the analyst theorizes under the simplifying assumptions of two states, two types of each state, and two actions.

Two States Jervis typically theorized about “the state” and “the adversary.” This assumption betrays the Cold War origins of the model, but it remains a useful

James D. Fearon is Geballe Professor of Political Science in the School of Humanities and Sciences, Department of Political Science, Stanford University. 1

Charles L. Glaser, Rational Theory of International Politics (Princeton, NJ: Princeton University Press, 2010), subsequent references cited in text. 2 Robert Jervis, Perception and Misperception in International Politics (Princeton, NJ: Princeton University Press, 1976); Robert Jervis, “Cooperation under the Security Dilemma,” World Politics 30, no. 1 (1978): 167–214; Andrew H. Kydd, “Game Theory and Spiral Model,” World Politics 49, no. 3 (1997): 371–400; Andrew H. Kydd, Trust and Mistrust in International Relations (Princeton, NJ: Princeton University Press, 2005); Charles L. Glaser, “Political Consequences of Military Strategy: Expanding and Refining the Spiral and Deterrence Models,” World Politics 44, no. 4 (July, 1992): 497–538.

431

Downloaded by [Florida International University] at 17:22 17 September 2011

432

J. D. Fearon

starting point for many settings besides “bipolar” relations between superpowers. The idea is to focus on a dyadic interaction, provisionally treating actions by third parties as parametric rather than part of the strategic problem being considered. Thus, the model might be applied to US-Iran relations or US-North Korean relations in considering how each side’s actions affect responses by the other. Actions by other relevant parties can be considered but need to be built in rather than treated as fully endogenous in the analysis. This is a liability, especially the deeper one goes into specific policy questions about a specific case and for certain classes of problems that are inescapably triadic like moral hazard in defender-ally-attacker relationships. It can, however, still be very useful as a first cut.3

Two Types Continuing an IR theory tradition that goes back at least to Hans Morgenthau, Jervis developed many arguments based on a dichotomy between “status quo states” that are interested only in their own security and “aggressive” or “expansionist” states that value taking more territory, or changing the international status quo on some issue, for their own sake. To emphasize that this is a distinction about preferences over outcomes rather than preferences over foreign policies, Glaser helpfully uses “greedy” rather than “aggressive” or “expansionist.” As he stresses, a status quo state could be expansionist about territory purely out of desire to protect what it already has.

Two Actions The states in the model are depicted as choosing between two actions, one of which can be described as cooperative and the other competitive. In RTIP , Glaser, like Jervis and Kydd, mainly considers a state’s choice between building arms and not building arms, but he notes that the actions could include other foreign policies such as whether to extend political or military control over buffer states or colonies or whether to deploy a particular military technology.4

3 The “two state” assumption may also have been rendered somewhat less objectionable by nuclear weapons (and perhaps some other developments), as these have made old-school, nineteenth-century style alliance politics a dead letter among the major powers. In the nuclear era, the major powers do not have the same need for allies to deter invasion by military coalitions of other major powers. Instead, for the United States, at least, principal security problems now frequently involve dyadic or “n versus 1” relations with minor or middle powers that have or that might acquire nuclear weapons. 4 Glaser often refers to the choice not to build as “arms control,” which is not really right since this takes two. Also, even if both sides do not build, we should distinguish between mutually arranged coordination (arms control) and restraint that is the result of dominant strategies (that is, when not building is a best reply regardless of the other’s action). More generally, RTIP has little to say about arms

Downloaded by [Florida International University] at 17:22 17 September 2011

Two States, Two Types, Two Actions

433

Closing the model so that it can produce implications, whether formally or not, requires two additional sorts of assumptions. First, we need to assume, as Jervis et al. do, that the states have beliefs about the likelihood that the other state is a greedy or status quo type. For later use, let the probability α 2 be state one’s belief that state two is a greedy type and α 1 be two’s belief that one is greedy.5 Second, we need to make assumptions about how the different possible arming choices translate into political or military outcomes. The idea in Jervis, Kydd, and Glaser is always that building when the other has not provides a military advantage in the event of conflict or coercive bargaining and that greedy types have a preference-based reason to exploit an advantage while pure security seekers do not. In one type of formalization, after the states have made and observed their building choices, they might choose whether to attack the other.6 Alternatively, one can treat conflict in a more reduced-form way, by assuming that the military balance translates directly into changed distributions of territory (or issue resolutions) via an unmodelled bargaining process.7 Informal versions of the model tend to be vague on this issue, so that arguments about the security dilemma are widely assumed to have something to say about when and why wars occur, but for no good reason. RTIP provides some discussion of how building decisions might be related to war risk (112–23), but this is an area where I think more work is needed to connect the arguments about competition via arms racing (or other competitive actions) with the question of war. Reduced form or not, this is where the offense-defense balance and offense-defense distinguishability enter the model. If, when state one builds and state two does not, state one’s odds of winning a war increase a great deal, then offense is favored in the balance. If the states can choose types of military assets that differ in their aptitude for defense versus attack, then the choice of asset may be able to communicate information about a state’s type. Jervis famously argued that defense dominance and distinguishability favor cooperation despite anarchy.8 In RTIP , Glaser works systematically through

control negotiations or how to think about arms control in theoretical terms, for example, as coordination on one of multiple equilibria in the 2-2-2 model under some parameter conditions. 5 In game models like Kydd’s in “Game Theory and the Spiral Model” and Trust and Mistrust in International Relations, we need to make further assumptions about what the states believe about the other’s beliefs about their own type. As Kydd shows, such assumptions are also required to fully explicate some of the key arguments in Jervis’s Perception and Misperception about “the spiral model.” 6 Robert Powell, “Guns, Butter, and Anarchy,” American Political Science Review 87, no. 1 (1993): 115–32; Kydd, “Game Theory and the Spiral Model”; James D. Fearon, “Arming and Arms Races” (paper presented at the Annual Meetings of the American Political Science Association, Washington, D.C., 2–5 September 2010). 7 Kydd, Trust and Mistrust in International Relations; Stergios Skaperdas, “Conflict, Cooperation, and Power in the Absence of Property Rights,” American Economic Review 82, no. 4 (1992): 720–39; Jack Hirshleifer, “Anarchy and Its Breakdown,” Journal of Political Economy 103, no. 1 (1995): 26–52. 8 Jervis, “Cooperation under the Security Dilemma.”

434

J. D. Fearon

the implications of these aspects of military technology in the context of the 2-2-2 model.

Downloaded by [Florida International University] at 17:22 17 September 2011

VALUE ADDED Both in Kydd’s formalization and in Glaser’s informal analysis, the basic model yields a set of intuitive implications that nonetheless go against important articles of faith espoused by many self-described realists, and offensive realists in particular. The basic question is: when does it make sense for a security seeker to choose competitive policies? Not surprisingly—but as Glaser points out, at odds with common claims by some realists—each state’s belief about the other’s type is important. In the simplest versions of the formal model, a security seeker arms if the cost of arming is less than the probability that the other state is greedy (α j ) times the cost of possible losses suffered if that other state turns out to be greedy. So if there is a positive cost to arming, then there is always a positive threshold probability α ∗j such that a status quo type prefers not to take the competitive action if it believes the other state is less likely to be greedy than α ∗j . This is true no matter how bad it would be if the other state turned out to be a greedy type. So, when both put low enough weight on the other being greedy, status quo types will choose not to arm while greedy types would. This means that if in fact both states are security seekers, their actions reveal this, producing a “downward spiral” of beliefs. In this instance, cooperative policies reveal that the states are not aggressive types, reducing further subsequent needs for arms. (In a richer model, states would presumably want to keep some arms as insurance against a change of types.) By contrast, if the initial beliefs α 1 and α 2 are above the relevant thresholds, then security seekers will arm—acting the same way as greedy states—so that beliefs should not change after the states observe each other arming. In this simplest case, then, there is never an “upward spiral” of beliefs that the other is greedy when in truth it is a status quo type. Kydd showed that if the states are additionally uncertain about each other’s beliefs about one another—for instance, state one is unsure if state two puts high or low odds on state one being a status quo type—then upward spirals are possible between status quo types.9 In RTIP , Glaser notes this mechanism and suggests some others that he conjectures would imply the possibility of mutual arming that leads (correctly, on average) to increased beliefs that the other is greedy. The issue is important because if upward spirals between security seekers are possible, then there can be a cost to arming that is independent of the material cost of procuring and deploying weapons. Namely, arming will increase the other’s fear that a state is greedy and so increase the 9

Kydd, “Game Theory and the Spiral Model.”

Downloaded by [Florida International University] at 17:22 17 September 2011

Two States, Two Types, Two Actions

435

risk of preemptive or other protective actions that could harm the state. This danger then strengthens the conclusion from the simple model that for low enough initial beliefs that the other is greedy, status quo types will rationally not want to arm, both to save direct costs and perhaps also to send a signal of their type to reduce the risk of conflict if the other is also a security seeker. As Glaser stresses, a common claim made by some self-described realists is that even if all states are security seekers, international anarchy forces them to behave in a highly competitive manner.10 The argument behind this claim is unclear, but it usually seems to come from an assertion that security seekers should rationally adopt worst-case assumptions about possible adversaries’ types. But if you think the other state is almost surely a status quo type, then, as the 2-2-2 model clarifies, to act as if it was surely greedy would be bizarrely irrational. The 2-2-2 model suggests, reasonably, that the extent of costly competition in international politics will depend on states’ beliefs about each other’s underlying preferences for controlling more territory and on their actual preferences, both of which are claimed to be unimportant by some realists. A second set of results from the basic model concerns comparative statics on the offense-defense balance and offense-defense distinguishability. Defensive advantage lowers the costs of not arming when the other arms (and is greedy) so that the threshold belief α ∗j below which state i does not build is higher. That is, defense dominance makes cooperation safer and the best choice for a larger range of beliefs about the probability that the other is greedy. As Glaser argues, then, beliefs about type and the offense-defense balance interact in the sense that improving on one relaxes the constraint posed by the other. These arguments can be efficiently derived from simple formalizations of the 2-2-2 model, along with additional implications concerning, for instance, the cost of arming and of fighting wars.11 In RTIP , Glaser carefully works through the model’s implications informally. In terms of theoretical development, I see the main value added as coming from his more extensive analysis of the role of military technology than I have seen in formal versions (or informal treatments). To date, the formal models represent only the most basic aspect of the offense-defense balance—the idea that offensive advantage means that small changes in the military balance have a large impact on the probability of winning a fight. The implications of offense-defense distinguishability have not been explored as far as I know. Glaser also makes interesting proposals about how variation in relative power should affect arming strategies, a subject little discussed in the formal literature. He argues that power disparity reduces the strong state’s incentive to compete while 10

See, for example, John Mearsheimer, The Tragedy of Great Power Politics (New York: Norton,

2001). 11

Kydd, “Game Theory and the Spiral Model”; Kydd, Trust and Mistrust in International Relations.

Downloaded by [Florida International University] at 17:22 17 September 2011

436

J. D. Fearon

increasing the weaker state’s, and he further discusses a set of interactions between relative power and the belief and offense-defense variables.12 More generally, a new feature of Glaser’s analysis is his attention to how his four main exogenous variables—beliefs about type, actual types, military technology, and relative power—interact with each other to determine optimal strategies in a large number of situations. Glaser makes his arguments informally but carefully, and he pays close attention to elaborating his assumptions in terms familiar to self-described realists. For these reasons his demonstration that a rationalist realism cannot plausibly hold that anarchy must imply high levels of interstate competition may have more impact on this part of the field than prior formal and informal arguments to similar effect. It would be progress if, as a result of this book, a realist analysis of foreign policy choices for pairs of states came to mean an analysis that takes into account the four central independent variables and their interactions along the lines that Glaser develops.

TWO OBJECTIONS: SECURITY IS A MEANS, AND ALL STATES ARE SOMEWHAT GREEDY I agree with Glaser that the 2-2-2 model usefully develops the central argument that the degree of interstate competition should depend on variation in states’ preferences for acquiring territory, beliefs about this, and relative power and the offense-defense balance. Still, the model has features that I think are problematic limitations that should be overcome in future work. The problems I will discuss here concern, first, the assumption that states are typically motivated by a desire for security and second, the assumption of two types of state preferences, greedy and security seeker. The assumption that, with few exceptions, states are interested in security may be useful for some purposes, but in terms of theoretical foundations, it is the wrong place to start. The right place to start, for a rationalist theory of this sort, is with the assumption that the people who make up or run the state want to maximize consumption, whether of material goods or symbolic things like nationalist satisfaction from having certain people inhabit a territory. Security seeking should be derived as a possible means to such ends, not posited as a nearly universal end in itself. In building his theory on the assumption that “most states are motivated by at least security” (37), Glaser follows the extremely influential argument of Kenneth Waltz, who wrote that “survival is a prerequisite to any goals that 12 Contrary to Glaser’s argument, in formal versions of the model power disparity by itself usually reduces incentives to compete for both states. Since the stronger state has less to fear if the weaker state is actually greedy, it is more inclined to cooperate if it is a security seeker, which in turn can make the weaker state less nervous.

Downloaded by [Florida International University] at 17:22 17 September 2011

Two States, Two Types, Two Actions

437

states may have other than the goal of promoting their own disappearance as political entities,” and “only if survival is assured can states safely seek such other goals as tranquility, profit, and power.”13 I think this argument has unintentionally misled a generation of realist and some constructivist scholars as well. It makes sense to assume that most people have a basic desire to survive and to keep what they have, but the analogy fails for states because states are themselves the means people construct to advance other ends. Suppose, for example, that the primary goal of the people or leaders in two neighboring countries is to maximize their incomes and that a merger into a single state happens to be the best way for them to achieve this. Then it would not be true that, as Glaser says, “Security is required to enable a state to pursue any other motives and goals over the long term, including domestic prosperity” (37). In short, it is false that the state’s survival and continued control of its territory are prerequisites for any other ends that the people who constitute the state might plausibly have. Now it may be true that seeking to survive as an independent state often happens to be the best feasible means to maximizing material or, in the nationalist age that we live in, symbolic consumption. For example, even though federation might increase total economic output by allowing lower defense expenditures, the leaders of two states might have difficulty constructing power-sharing institutions that would credibly commit them against attempts to grab the whole “pie” rather than sharing the larger revenue stream. (This problem is arguably at the heart of many civil wars.) But then security-seeking state preferences are an outcome to be explained as a consequence of particular facts about domestic politics and institutions, along with anarchy and the basic motive of consumption.14 Making consumption the starting point also forces us to consider more directly the domestic tradeoff between “guns and butter,” which is mentioned occasionally by Glaser but always set to the side. In the very basic 2-2-2 model sketched above, if arming is costless, then security seekers arm for any beliefs about the adversary whatsoever. This suggests that arming costs should be a more important variable in Glaser’s theory.15 In addition to having more defensible first principles because it starts with the ends being sought rather than a means to them (security), the assumption that states want to maximize consumption (material or otherwise) has the added value of showing that assumptions about political institutions

13

Kenneth N. Waltz, Theory of International Politics (Reading, MA: Addison-Wesley, 1979), 91–92,

126. 14 James D. Fearon, “Domestic Politics, Foreign Policy, and Theories of International Relations,” Annual Review of Political Science 1 (1998): 283–313, 294. 15 In “Guns, Butter, and Anarchy,” Powell shows how to think theoretically about the arming problem when one starts from the assumption that states are interested in maximizing consumption, subject to the constraints that they can attack each other but cannot form a political union.

Downloaded by [Florida International University] at 17:22 17 September 2011

438

J. D. Fearon

and regime type need to be built into the foundations of a coherent structural IR theory.16 If you start with consumption, you have to ask, “Whose consumption?” This depends on domestic political institutions and in particular on whether the regime is democratic or not. For example, democracies may tend to have less value for territorial conquest than do narrowly held autocracies if the gains must be more widely shared, and they may be more able to parcel out aspects of sovereignty because they have already developed successful institutions for power sharing at home.17 The larger point is that if we start in the right place—assuming consumption, broadly conceived, rather than security as the core motive—we are likely to develop both a richer and conceptually simpler IR theory in which regime type is an integral component rather than an add-on. The second problematic feature of the 2-2-2 model worth discussing here is the two-type assumption. Are any states really pure security seekers, in the sense that they would not want to change anything about the territorial status quo or the status quo with respect to other issues, if they could do so at no cost and with no consequences? Can leaders or analysts assign their own state’s type to the dichotomous greedy/security seeker categories? (This is necessary if the theory is to be usable for producing policy recommendations, as Glaser would like.) And do Glaser’s theoretical arguments go through if the reality is that all states are at least somewhat greedy, and uncertainty is over a more continuous distribution of levels of greed? The answer to the first question—are there any pure security seekers?—is almost surely, “no.” In a footnote, Glaser says that “an actual state may not have to entirely lack greedy motives to be productively characterized as a security seeker” (36), and he recognizes and considers the possibility of variation in degrees of greed and in the value a security seeker puts on keeping what it has. But I have doubts about the full relevance of much of the analysis if in the real world all “security seekers” have at least some revisionist preferences. In this world, even relative security seekers sometimes have reasons to engage in coercive bargaining with each other. RTIP is written mainly from the perspective of “what should a pure security seeker do?” including the evaluation of the rationality of arms racing decisions by states such as Wilhelmine Germany and interwar Japan. But what if

16 By “structural” I mean analogous to the microeconomic approach to oligopolistic markets, as Waltz wanted to develop his theory in Theory of International Politics. Waltz argued that this meant there should be no variation assumed of the actors except their relative power. I think that is too restrictive an interpretation of, say, the Cournot model of oligopoly, where firms can vary in their production costs among other things. See Fearon, “Domestic Politics.” 17 For models developing such arguments, see Matthew O. Jackson and Massimo Morelli, “Political Bias and War,” American Economic Review 97, no. 4 (2007): 1353–73; James D. Fearon, “A Simple Political Economy of Relations among Democracies and Autocracies” (paper presented at the Annual Meetings of the American Political Science Association, Boston, 27–31 August 2008).

Downloaded by [Florida International University] at 17:22 17 September 2011

Two States, Two Types, Two Actions

439

there are no pure security seekers and what if even small amounts of greed can be consequential in some strategic interactions? To illustrate, consider the question of whether the United States is greedy with respect to Iraq in 2003 or to Iran or North Korea today. In none of these cases did or does the United States aspire to formally annex territory, so I suspect Glaser would say that, in the core sense used through most of the book, the United States is a security seeker in these relationships. But US administrations over the years have intensely disliked Iranian and North Korean regimes, as well as Saddam Hussein’s. This was not solely because of fear or security reasons: in all cases, even a credible guarantee of no expansion would not have made US administrations completely happy with the dictatorial incumbent. So in that sense, in terms of having revisionist preferences about regime change, the United States was and is slightly “greedy.”18 Arguably, even these mildly greedy preferences may contribute to producing a great deal of conflict if the larger strategic context is right. These regimes know we dislike them as such, and because we are very powerful, they fear us. Given its revisionist preferences over regime types, the United States cannot credibly commit not to use its power to undermine and act against these regimes given the opportunity, which is an important reason why they can have an incentive to try to build nuclear weapons and also to resist monitoring.19 This in turn creates major security fears and worries on our part.20 If Glaser’s theory were applicable and the United States was a pure security seeker with respect to these regimes, our leaders would be happy to say, “We publicly promise not to attack, to push you around, or to try to destabilize your dictatorship with our greater power if you credibly promise not to develop nuclear weapons” (say, by allowing extensive international monitoring). Following Glaser’s arguments, if we were a pure security seeker with respect to these regimes, we should be willing to try to send signals to this effect. But for several of these examples and perhaps all of them at different times, it is hard to imagine US leaders doing this, I think in part because we are inevitably somewhat greedy on the regime change dimension. In sum, I think that most of the action in international security politics is between states that have at least somewhat revisionist preferences, so

18 One can argue further that democracies are intrinsically threatening to non-democracies, because their mere existence provides an example that might inspire people in non-democracies to rebel. 19 This is not the sole reason; some greed or revisionist preferences on their part can be another, as well as pure status-seeking and domestic political reasons. 20 Note that this is not a “security dilemma” as IR scholars usually use the term, since an important premise was that the United States is not a pure security seeker with respect to, say, Iran or Saddam Hussein’s Iraq. Instead it is a commitment problem akin to the basic Prisoners’ Dilemma-like problem of arming: the United States cannot commit not to exploit a relative power advantage if the other state is disarmed (does not have nukes).

Downloaded by [Florida International University] at 17:22 17 September 2011

440

J. D. Fearon

that much of what we see is coercive bargaining in an environment with primitive commitment technologies. For Glaser and the 2-2-2 model, the central dilemma for a security seeker is that arming provides protection if the other is a greedy type but may cause unnecessary conflict and expense if the other is a security seeker. I think that probably more often the central dilemma is how to get what you want on changing the status quo without paying large costs in armament or war, while states’ beliefs about “types” in Glaser’s main sense are largely accurate or play a secondary role. There are some strong parallels between the two dilemmas, but they are not the same. Both of the objections outlined above—that state preferences over territory should be derived from more primitive assumptions about the political economy of states, and that “the right model” of Glaser’s sort should have some level of conflict of interest between even what he calls “security seekers” and the possibility of bargaining—are more in the manner of next steps for theory development than they are unnoticed problems that would vitiate Glaser’s approach. The 2-2-2 model is a natural and useful theoretical simplification, well worth the careful analysis that it gets in RTIP . It should replace a lot of less careful claims about what follows from variously posed “realist assumptions,” and Glaser’s treatment of it should help to shape a new round of theoretical work on the strategic dynamics of contentious dyads.

Security Studies, 20:441–450, 2011 Copyright © Taylor & Francis Group, LLC ISSN: 0963-6412 print / 1556-1852 online DOI: 10.1080/09636412.2011.600144

Rationalist Theories of International Politics and the Problem of the Future DALE C. COPELAND

Charles Glaser’s Rationalist Theory of International Politics nicely extends his work on contingent realism begun in the mid-1990s.1 The rationalist theory he puts forward focuses on three main causal variables: material factors such as the offense-defense balance and power; informational variables, primarily leader uncertainty about the motives of key adversaries; and a state’s motives per se, namely, whether it is a security-seeking actor or has “greedy” non-security reasons for acting. Glaser’s core contribution comes with his balanced reworking of one of the foundational concepts of international relations: the security dilemma. In a chapter entitled simply “The Theory,” Glaser effectively shows that depending on its intensity, the security dilemma can cut both ways, either encouraging states to compete or convincing them that moderate behavior is best. When offense has the advantage and states have information suggesting that adversaries have greedy motives, the security dilemma is intense, and states will be pushed into competitive spirals of arms racing, territorial expansion, and war. But when defense has the advantage and adversaries appear to be security seeking, the security dilemma is moderated, and states will avoid hard-line actions that might cause others to doubt their motives and intentions (see esp. 57–90). The main target of Glaser’s book is the spare version of offensive realism put forward most notably by John Mearsheimer. Mearsheimer argues that anarchy forces rational security-maximizing states to assume worst case—that other states are out to get them either now or in the future—and thus to compete fiercely for position and territorial control in the system.2 In one of the book’s most important contributions, Glaser puts to rest any

Dale C. Copeland is an associate professor in the Department of Politics at the University of Virginia. For helpful comments on earlier drafts, I wish to thank Kyle Lascurettes and Brandon Yoder. 1 Charles L. Glaser, Rational Theory of International Politics: The Logic of Competition and Cooperation (Princeton: Princeton University Press, 2010), subsequent references cited in text. 2 John J. Mearsheimer, The Tragedy of Great Power Politics (New York: Norton, 2001).

441

Downloaded by [Florida International University] at 17:22 17 September 2011

442

D. C. Copeland

notion that a properly specified rationalist theory can be founded upon a generalized worst-case assumption. Because security-driven actors know that overly hard-line behavior can cause unwanted spirals of mistrust and counter-balancing they would be irrational to always assume worst case. This would lead them into constant conflicts that would sap their long-term ability to play the power game. In situations of highly intense security dilemmas, worst-case thinking might be appropriate. But for most other environments, rationalist theory must allow the beliefs about the other’s type to vary so that states can calibrate the severity of their policies relative to both the material and informational dimensions of their situations. This short critique will not reveal any hidden logical flaws within Glaser’s argument. The book is internally well argued and indeed offers the most powerful fusion of defensive realism and neoclassical realism yet developed.3 It also captures some of the main insights of neoliberal institutionalism regarding the importance of shared beliefs and informational transparency, even as it is able to bound its optimism. When the security dilemma is intense, states can find themselves stuck in repeated patterns of competitive behavior that institutional frameworks cannot easily resolve. The security dilemma, in short, hangs over states in a way that forces them to be prudent. Cooperation can still obtain, but only under certain conditions. Glaser’s careful analysis of how the offense-defense balance, power, and information about motives interact to shape behavior is an exemplar for an IR field that too often stresses monocausal arguments or considers multiple variables only in isolation. My main concerns center not on what the book says, but on what it does not say. Glaser has provided a plausible rationalist theory, but he does not set it against alternative rationalist theories that might challenge either the veracity of his deductive logic or his interpretations of history. There are many such approaches one could discuss, including rational deterrence theory, theories of crisis escalation, and the bargaining model of war. Given space considerations, I will focus on two groups of arguments that I find particularly powerful, arguments that also start from a rationalist foundation but which potentially undermine the generalizability and empirical value of Glaser’s theory. These are: dynamic versions of realism that stress power trends and the malleability of character type; and theories that examine the effects of economic dependence and trade expectations on power politics. Both approaches emphasize the problem of future changes in the realities that actors face and how the anticipation of such changes can drive current behavior. As such, they are opposed to what Glaser refers to as his inherently “static theory” or “static picture” of international politics (110–11), a deductive logic based on snapshots of current variables rather than on actor expectations of the future. I should note upfront that my goal is not to argue 3 For a parallel effort, see Andrew Kydd, Trust and Mistrust in International Politics (Princeton: Princeton University Press, 2005).

Downloaded by [Florida International University] at 17:22 17 September 2011

Rationalist Theories of International Politics

443

that Glaser is wrong. It is only to show that once we bring in dynamic rationalist alternatives, the extent to which he is right is very much up in the air. Any debate between Glaser and these and other such rationalist theories, given that the latter emphasize different independent variables and different causal logics, can only be resolved by the kind of in-depth empirical tests that Glaser’s book did not have the space to perform. Let us begin with the interesting title of the book. By leaving out the definite or indefinite article, Glaser leaves us wondering whether he has presented “a” rationalist theory of international politics or “the” rationalist theory of international politics. Throughout the book, he reminds us that he is offering “a” rationalist baseline for evaluating the optimality or suboptimality of state actions. A careful reading of the book, however, indicates that Glaser is not interested in setting his argument against other competing rationalist theories. His goal is much more ambitious. Without explicitly stating that his is the only rationalist baseline one should employ, Glaser structures the book to convince us that his theory replaces, or is the necessary foundation for, all other rationalist logics. The book does this in two ways. First, any rationalist arguments that might otherwise seem to compete with his, such as dynamic realist models, the bargaining theory of war, and neoliberal institutionalism, are treated simply as “extensions” of his theory (108–26). The question of whether these alternative perspectives actually challenge the core implications of his theory either deductively or empirically is sidestepped by absorbing them within his apparently more comprehensive framework. Second, whenever actors in the real world fail to act as his theory would expect them to act, their behavior is automatically labeled as “suboptimal,” that is, as a product of domestic-level pathologies that interfered with truly rational decision making (156–57, 172–80, 229–68). The possibility that these actors, instead of being irrational, might be operating according to a competing rationalist logic is never considered. These two conceptual moves serve to shield the book from any direct criticism. If alternative rationalist theories are mere extensions of Glaser’s theory, then by definition they cannot fundamentally question his argument. They may elucidate his argument’s underplayed aspects, but they can never undermine our confidence in it. And if empirical evidence at odds with Glaser’s framework is necessarily evidence of irrationality at the unit level, then the theory can never be disconfirmed. It is only occasionally trumped by domestic and psychological pathologies that no rationalist theory could have ever fully anticipated. In this sense, it appears that Glaser has offered us an inherently non-falsifiable theory. Any historical behavior that does not fit his version of what a rationalist theory would expect is by definition suboptimal and thus outside of his theory’s framework.4 4 That this is Glaser’s modus operandi is shown by his two empirical chapters, 8 and 9. These chapters offer no event or policy across a century and a half of history that is seen as disconfirming

Downloaded by [Florida International University] at 17:22 17 September 2011

444

D. C. Copeland

Once we bring in alternative rationalist theories and set them against Glaser’s argument, we can deal with anomalies to his causal logic without necessarily having to fall back on pathologies at the unit level. Consider Glaser’s theory in comparison to a properly specified dynamic version of systemic realism, one that takes into account both the trends in a state’s military power and its uncertainty about the future character, incentives, and existential situations of other actors. Such an alternative rationalist theory would allow variability in actor type (it would not assume worst case). But it would also stress that variability in future realities, rather than any variability in the present, is most likely to drive state behavior now. In such a dynamic theory, whether the other is currently security seeking or greedy would be of little real consequence if the state was losing relative power and had good reason to fear the other’s future type and future behavior.5 In fact, even when a rational security seeker knew for sure that the other was currently also a rational security seeker, it might have reason to compete for systemic position. It would be acting to avoid a future situation where the other was militarily dominant and had “changed its stripes,” becoming either a greedy actor or one that pursued security in a less-than-rational way.6 Glaser’s book has a radically different theoretical set-up from such a dynamic theory, despite the common baseline assumption of rationalism. As the first three chapters make clear, Glaser focuses only on snapshots of the present moment—whether a rational actor believes it is facing a greedy state or a security-driven state, an offense-dominant or defense-dominant situation, or a power advantage or disadvantage. He then derives all of his key predictions from various mixtures of these present-moment informational and material variables.7 In chapter 4, Glaser argues that his static argument of his logic. Rather, all events and policies at odds with his expectations are labeled suboptimal and irrational, without any exploration of the documents to see if actors were indeed violating rational-actor assumptions due to poor information searches, cognitive closure, bolstering, the distortions of domestic politics, and so forth (see esp. 229–67). 5 For references to the vast preventive war literature and for my own take on such a dynamic realist approach, including the conditions under which preventive war or preventive containment actions are likely, see Dale C. Copeland, The Origins of Major War (Ithaca: Cornell University Press, 2000), chaps. 1–2. 6 Glaser cannot assert that the security dilemma would be “eliminated” if both actors knew with certainty that the other was also a security-seeker (81, 86–87). Once we bring in the fact that future character can change, the security dilemma races back in, forcing states to be concerned about their future power positions. Thus the irony of Glaser’s defensive realism and all theories such as constructivism and liberalism that underscore the importance of variation in state types: it is precisely because the character of actors matters that the problem of changing future intentions hangs over all actors, even when they know each other to be good security-seekers. For more on this, see Dale C. Copeland, “The Constructivist Challenge to Structural Realism,” International Security 25, no. 2 (Fall 2000): 187–212. For the related commitment problem, see Robert Powell, “War as a Commitment Problem,” International Organization 60, no. 1 (Winter 2006): 169–203. 7 The variable “state motives” is brought in only briefly at the end of the main theory chapter, chapter 3 (see 87–90), to cover states that may be more aggressively competitive than would security-seeking actors under similar circumstances. This neoclassical addition to Glaser’s defensive realist core is then

Downloaded by [Florida International University] at 17:22 17 September 2011

Rationalist Theories of International Politics

445

can be extended by bringing in “dynamic considerations.” The fact that “all of the theory’s key variables could change over time” means that a rational state must consider “not only the international situation it currently faces, but also the situation it might face in the future” (108). No debate there. Unfortunately, nothing in the book’s theoretical set-up prepares us for this sudden introduction of dynamic variables. None of his variables is conceptualized in dynamic terms, so it is not clear exactly how his logic could be “extended” to include a state’s expectations of future power and state type. In fact, his short discussion of dynamic issues (108–12) is used not to provide a dynamic version of his logic but to show how his logic restrains or moderates some of the deleterious effects caused by changing power, technology, and regime type.8 This is revealing. It suggests Glaser understands that dynamic realist arguments are not really extensions of his theory, but rather theories that directly compete with his own predictions. Given this, he is obliged to show that their insights should not cause us rethink one of his core points against offensive realism, namely, that the nature of the system and fear of the future do not tend to push states in a competitive direction, all things being equal.9 The lack of a dynamic underpinning to Glaser’s theoretical logic plays havoc with his effort to understand real-world cases. Consider two of his strongest historical examples. He argues that the United States should have sought a ban on MIRV’d ICBMs in the early 1970s. But instead of exploring the difficult trade-offs that the Nixon-Kissinger team confronted at the time, he simply labels America’s MIRV buildup a suboptimal product of domestic used in the empirical chapters to mop up variance that is left unexplained by changes in the intensity of the security dilemma. As such, this extra variable represents a degenerative aspect of Glaser’s research program: it is used only as an ad hoc variable to explain events outside the theory’s core deductive logic. On this, see Jeffrey W. Legro and Andrew Moravscik, “Is Anybody Still a Realist?” International Security 24, no. 2 (Fall 1999): 5–55. 8 Glaser’s three paragraphs on preventive war argue that if the current offense-defense balance is defense dominant, the rising state is presently security-seeking, or the declining state is a security seeker, the incentives for preventive war are reduced. To the criticism that state motives can change, Glaser counters that cooperation can still obtain because of the costs of preventive war are often high and because rising actors may indeed remain peaceful in type (109–112). These are reasonable qualifications. But they are only counter-arguments that suggest his variables might occasionally keep states from launching preventive actions. They are not arguments explaining the push to preventive war or preventive containment strategies, behaviors that would clearly undermine his claim that cooperation is typically the most rational strategic option for all players. Interestingly, he is more candid when discussing the implications of a changing offense-defense balance. His paragraph on the subject acknowledges that a balance moving from defense-dominance to offense-dominance would leave states “less confident in their security than [my] static theory suggests.” Nonetheless, cooperation can be saved by the existence of nuclear weapons that “moderate these dynamic concerns” by enhancing deterrence (110–11). Aside from the fact that nuclear weapons are an offensive technology that achieve deterrence by their sheer ability to punish, Glaser’s discussion here again shows that he is only countering the counter-arguments using his own static variables rather than demonstrating how his theory could be made truly dynamic in and of itself. 9 After all, it is the malleability of the future, not any worst-case assumption, that gives offensive realism its greatest analytical bite.

Downloaded by [Florida International University] at 17:22 17 September 2011

446

D. C. Copeland

pathologies, one that reduced strategic stability (260–65). Yet in the face of a growing Soviet ICBM force and Vietnam-driven legislative resistance to increased defense spending, Kissinger reluctantly fell back on MIRVs to maintain a measure of parity during a difficult transition period in American foreign policy. Fears of decline, combined with concerns about Soviet cheating, were determinative. Even with our best cases, therefore, a research strategy that compares competing rationalist explanations will help us temper any tendency to assume that a behavior was irrational simply because it did not fit our particular rationalist theory and because it had a deleterious effect on system stability.10 Balanced empirical testing requires that we avoid relying on 20-20 hindsight and that we instead examine the information available to actors at the time. In-depth use of documents may indeed reveal that leaders were not following the guidelines of the rational actor model, but the documents may also show that leaders were simply operating according to another rationalist logic, one that calls our own theory into question. Glaser also seeks to show that the arms racing prior to World War I and the German initiation of that war were the products of suboptimal decision making. The 1910–14 period is critical to his overall thesis since it has always been seen as the preeminent case of the offense-defense balance mattering in world politics. It is odd, therefore, that Glaser ends up agreeing that the Germans initiated war largely for preventive reasons (the fear of the rise of Russia) and despite the fact that they alone understood that technology favored the defense. He even acknowledges that the huge German military buildup from 1912 to 1914 was designed largely to create a power superiority that would compensate for the natural defensive advantages of the French, British, and Russian armies on the battlefield (244–46). Yet by the end of his discussion he falls back on the idea that the Germans “could have adopted a more defensive doctrine”—that is, a strategy more in line with the prescriptions of Glaser’s normative theory—and thus moderated the incentives for arms racing and war. The fact that they did not suggests that Germany’s “entire military strategy . . . was suboptimal” (247). The problem with Glaser’s conclusion is not just that he supplies no evidence to support it. It is more that it skirts the key theoretical issue, namely, whether an alternative rationalist theory with a more dynamic baseline might do a better job than Glaser’s theory in explaining this difficult case. In a paragraph that suggests he is aware of the problem, Glaser notes that “whether Germany’s army buildup was suboptimal depends on a still larger debate over whether Germany required an offensive doctrine: if offense was required, the arms buildup was arguably Germany’s best option; in contrast, if 10 Glaser does not discuss the arms racing of the first fifteen years of the Cold War, and for good reason. Here it is quite obvious that ongoing fears of the future—delivery systems that the other might be deploying and technological innovations that it might only reveal later—were driving the intense efforts of two largely security-seeking actors to increase their strategic military power.

Downloaded by [Florida International University] at 17:22 17 September 2011

Rationalist Theories of International Politics

447

defense was sufficient, the arms buildup was suboptimal.” He does not state his position on this debate but simply hints that those arguing for the “cult of the offensive” and other irrationalist views have the upper hand (247). Aside from the fact that his previous discussion shows that German leaders did not believe in a cult of the offensive, Glaser has sidestepped the larger theoretical question. It is precisely the possibility that leaders might launch a preventive war for rational reasons, given their fears of the future power and intentions of others, that most challenges his own theory’s predictions. And it is thus against such theories that his theory must be tested. For if the alternative rationalist arguments can explain the case without dipping down to the unit level, they not only offer a superior explanation for such crucial cases, but they avoid the degenerative move of bringing in factors not part of the original theory (in this case, unit-level pathologies) to cover empirical variance left unexplained by the theory.11 It must be stressed that I am not arguing that a dynamic realist theory will always offer a better explanation for provocative state behavior. Moreover, sometimes unit-level variables do intervene. But unless we test Glaser’s rationalist theory against alternative rationalist theories, dynamic or otherwise, we will never know. And if we fail to perform such tests, we will be constantly inclined as theorists to presume that behavior we dislike in international relations must be the product of “crazy” leaders or states suffering from severe internal pathologies (a tendency that many defensive realist scholars suffer from). Another group of rationalist models, among many, that can give Glaser’s theory a run for its money includes those theories that consider the importance of commercial access and expectations of the future trade environment for the security of rational states.12 Apart from two paragraphs about US-Chinese relations at the very end of the book (280-81), Glaser does not discuss the impact of international economic factors on the behavior of great powers. This is unfortunate given that in the nuclear age, competition over resources and markets offers one of the few ways rational great powers might be drawn into militarized struggles that could escalate to war. Throughout history, states that have been economically dependent have feared that, in the future, they might be cut off from access to the countries that provide the raw materials and export markets critical to long-term economic growth.13 Without a strong economic foundation, a decline in relative military power is inevitable. Great powers have thus had a big incentive to protect their 11

See n. 7. For summaries and references, see Patrick J. McDonald, The Invisible Hand of Peace: Capitalism, the War Machine, and International Relations Theory (Cambridge: Cambridge University Press, 2009); Dale C. Copeland, “Trade Expectations and the Outbreak of Peace,” Security Studies 9, nos. 1–2 (Autumn– Winter 1999–2000): 15–58; Dale C. Copeland, Economic Interdependence and War (forthcoming). 13 See John J. Mearsheimer, “Disorder Restored,” in Rethinking America’s Security, ed. Graham Allison and Gregory F. Treventon (New York: Norton, 1992), 223. 12

Downloaded by [Florida International University] at 17:22 17 September 2011

448

D. C. Copeland

trade routes through the projection of military power into regional areas of dependency (the Middle East since 1940, for example). Yet the very act of projecting power around the system, even if only to protect commerce from unexpected threats, can spark a different form of security-dilemma spiraling, one driven by fears that other states’ projections of power reflect economically aggressive rather than merely defensive intentions. Glaser’s brief discussion of China’s growing resource dependence shows he is aware of these issues. He notes that an effort by China to protect its sea lanes of communications (SLOCs) through the buildup of its air and naval capabilities could create a spiraling of tensions with the United States. But he treats the question of economic dependence and fear of future cut-off as simply one of a handful of specific issues “that lie . . . below the sweeping view of grand international relations theory” presented in the book. He does not explore whether the rationalist theories that examine this question, by stressing the great-power drive to reduce vulnerability when conditions are increasingly uncertain, might directly challenge the validity of his theory’s “optimistic analysis of the international environment” (280–81). Consider Glaser’s take on Germany’s naval buildup after 1897. Not surprisingly, he sees it as yet another example of Germany’s suboptimal decision-making process rather than as a rational response to a threatening international situation. But instead of asking whether alternative rationalist arguments might potentially explain why German elites of all stripes felt the need for a naval buildup, Glaser argues that the buildup was suboptimal since it ended up wasting money and alienating Britain (238–43). Context matters, however. Germany was hardly alone in wanting a blue-water naval presence around the world. Indeed, every great power except Austria had begun a major naval buildup prior to Germany’s entry into the race: France, Italy, and Japan built up in the 1880s and the United States and Russia in the early 1890s.14 It is not enough for Glaser to focus on Germany—whose importance is clear only in retrospect, given its primary role in the initiation of World War I—and to assert that because a policy led to hostile relations, it must have been suboptimal. Instead, he must explain why almost every great power in the system was doing pretty well the same thing and why other dyads (for example, US-Britain 1895, Britain-France 1898, Russia-Japan 1904) came close to war or actually fell into it. The need to protect one’s economic security and the fears that others were seeking to create restrictive economic realms in Latin America, Africa, and Asia goes some distance to providing a generalizable answer. The fact that rapid industrialization was taking place within each great power—with Austria again the exception that proves the rule—only reinforces the point. Each great power needed larger 14 By 1890, Germany ranked behind the United States and even Italy, the weakest of the European great powers, in total warship tonnage. See Paul Kennedy, The Rise and Fall of the Great Powers (New York: Random House, 1987), 203.

Downloaded by [Florida International University] at 17:22 17 September 2011

Rationalist Theories of International Politics

449

export markets and an expanding variety of resource inputs to sustain its economic growth and to play the power game effectively. The German naval buildup was rooted in this impulse, even if, because of Germany’s advanced economic base and geographic location, its new policy was ultimately more frightening to Britain than the concomitant arms racing of America, Japan, France, and Russia.15 The need to protect far-flung economic interests from future threats suggests an additional problem for Glaser’s argument. To deter regional actors and other global powers from trying to control the resources and markets of economically important countries, a great power will have to be able to project a particular type of military power into a region: namely, offensive power. That is, even if a great power only wants to defend its current economic interests abroad, if it is going to “extend” its deterrence to a small third party, it must be able credibly to threaten to punish an adversary for any revisionist behavior. And for this, only offensive powerprojection capability—the power to cause damage to the other side’s forces or population—will do the trick. Sitting back behind a defensive shield that protects a great power’s homeland will not suffice. It may deter attacks on the homeland itself, but it can do little for an ally hundreds or often thousands of miles away. All this has a huge implication for Glaser’s theory. It means that great powers can rarely afford to emphasize a global posture of defense, even if defensive technologies at a snapshot in time are seemingly “dominant.” The political necessity of safeguarding economic interests far from home will override any desire to build only defensive strength. Notwithstanding Glaser’s optimism that security-driven states in a defense-dominant environment can escape the security dilemma (81-87), great powers in a globalized system will be forced to develop offensive technologies, if only to deter revisionist states or to respond to their encroachments abroad. And yet as Glaser himself emphasizes, the building up of the mere potential for causing harm can lead adversaries to question a state’s motives, undermining the peace even when all states are good security seekers.16 This does not mean that the need to protect economic assets abroad will necessarily lead to great power crises and war. Yet Glaser’s book, by ignoring the economic dimensions of great power politics, has left us with little sense of the conditions under which intense security-dilemma spirals over raw materials and markets can be avoided. In sum, Glaser’s rationalist argument cannot stand alone. It must be tested against alternative rationalist theories that incorporate causal variables 15 On the economic origins of German Weltpolitik or “world policy,” see Woodruff Smith, The Ideological Origins of Nazi Imperialism (Oxford: Oxford University Press, 1986). 16 The offense-defense balance in this regard is endogenous to a state’s global grand strategy rather than an exogenous variable shaping that strategy, as Glaser’s theoretical set-up must presuppose.

Downloaded by [Florida International University] at 17:22 17 September 2011

450

D. C. Copeland

that are not part of Glaser’s framework but that could turn out to be highly salient in historical practice: power trends and fears of decline, the other’s future type as opposed to just its present type, present and future economic and trade variables rather than simply a state’s military power, the induced incentive to build offensive power to defend far-flung allies and economic interests, and so forth. Without a doubt Glaser’s book provides the most comprehensive defensive-realist discussion of the security dilemma and its implications in the literature. And by effectively challenging the validity of the worst-case assumption, he forces one strand of offensive realism to rethink its fundamental starting point. Yet the book does not constitute a challenge to other rationalist theories of international politics, particularly those that bring in the importance of actors’ dynamic expectations of the future. To prove the value of his theory, therefore, in future work Glaser must offer head-to-head empirical tests of the main rationalist theories, including his own, across a variety of historical settings. Otherwise he will be left with a purely normative theory that offers merely one view of how rational states should act, a theory that too easily labels most states in history “irrational” simply because they do not follow his policy prescriptions.

Security Studies, 20:451–459, 2011 Copyright © Taylor & Francis Group, LLC ISSN: 0963-6412 print / 1556-1852 online DOI: 10.1080/09636412.2011.599193

Mission Impossible: Measuring the Offense-Defense Balance with Military Net Assessment KEIR A. LIEBER

Charles Glaser’s Rational Theory of International Politics argues that state security competition is not an inevitable consequence of international anarchy.1 To be sure, sometimes the attempt by one state to increase its security has the unintended and unavoidable effect of decreasing the security of others—triggering a spiral of arms racing and diplomatic hostility that results in war. At other times, however, this security dilemma can be avoided. According to Glaser, under a surprisingly wide range of conditions, states should be able to feel secure, build military forces to signal peaceful intentions, and thereby avoid unnecessary competition and conflict. In short, much of the tragedy of international politics could be ameliorated if state leaders were more attuned to conditions conducive to cooperation. What are these conditions? For Glaser, offense-defense variables are key. The nature of the offense-defense balance (ODB), and whether offense and defense are distinguishable, helps determine whether states can avoid negative security spirals. When the ODB favors defense, and the forces that support offense are distinguishable from the forces that support defense, securityseeking states can build defensive forces at less risk and without threatening the adversary’s ability to defend itself. This reassuring defense posture signals benign motives, encourages cooperative arms control policies, and has the potential to improve political relations over time. In sum, offense-defense variables are a crucial determinant of a state’s security environment and vital to the causal logic and policy prescriptions of Glaser’s theory.

Keir Lieber is associate professor in the Edmund A. Walsh School of Foreign Service and the Department of Government at Georgetown University. For helpful comments, the author thanks David Edelstein, Robert Lieber, Daryl Press, and Sebastian Rosato. 1 Charles L. Glaser, Rational Theory of International Politics: The Logic of Competition and Cooperation (Princeton: Princeton University Press, 2010), subsequent references cited in text.

451

Downloaded by [Florida International University] at 17:23 17 September 2011

452

K. A. Lieber

Glaser’s analysis is clear, rigorous, and comprehensive. The book is a major contribution to international relations theory. However, the theory’s reliance on offense-defense variables—specifically, its faith in the ability of states to assess the offense-defense balance and shape military policy accordingly—undermines its explanatory and normative power. Scholars have previously challenged the validity of offense-defense explanations.2 Glaser responds to many of these critiques in his book; the full debate does not need to be recounted here.3 Instead, the discussion below focuses on Glaser’s crucial claim—which has yet to be fully debated—that when it comes to measuring the ODB “the analytic tasks required are the same as those required to perform military net assessments” (140, emphasis added). As he stipulates, “If useful net assessment is feasible, useful estimates of the offense-defense balance are also feasible” (140). “Useful” in this context means that states are capable of measuring offense-defense variables, performing the analysis envisioned in the theory, and adopting the appropriate prescribed strategy when conditions permit. The argument that states are capable of assessing offense-defense variables as long as they are capable of performing military net assessment is unconvincing for two reasons. First, the task is impractical. The standard tools of military net assessment are not nearly as fine-grained as necessary for states to identify the conditions conducive for cooperative security policies as outlined in Glaser’s theory. Second, the task is logically flawed. Even if practically possible, states could not use the results of that analysis to shape arming and force posture decisions to yield the security benefits envisioned in the theory. In short, there is a fundamental mismatch between Glaser’s theory and its real-world applicability. The sections below briefly summarize the role of offense-defense variables in Glaser’s theory, discuss the limitations of net assessment compared to the demands of estimating the ODB, and explore why the carefully calibrated force structures envisioned by the theory would make little sense in the real world.

Offense-Defense Variables in Glaser’s Theory Material variables—those that determine a state’s military capability—play a major role in determining whether a state can choose a cooperative security 2 Critics have argued that one cannot distinguish between offensive and defensive weapons, the ODB cannot be measured, political and military planners do not think in offense-defense terms, and states have not behaved according to offense-defense predictions. See Keir A. Lieber, War and the Engineers: The Primacy of Politics over Technology (Ithaca: Cornell University Press, 2005); Stephen Biddle, Military Power: Explaining Victory and Defeat in Modern Battle (Princeton: Princeton University Press, 2004); Richard K. Betts, “Is Strategy an Illusion?” International Security 25, no. 2 (Fall 2000): 5–50; and Sean M. Lynn-Jones, “Offense-Defense Theory and Its Critics,” Security Studies 4, no. 4 (Summer 1995): 660–91. 3 Since Glaser describes his theory as “a theory of what states should do, not necessarily what they do do,” critiques about empirical validity can seemingly be set aside (143).

Downloaded by [Florida International University] at 17:23 17 September 2011

Mission Impossible

453

strategy. For Glaser, the key material variables are power, the ODB, and offense-defense distinguishability. He argues that the standard realist emphasis on power alone is problematic because offense-defense variables are necessary to understand how power is translated into a state’s ability to perform relevant military missions, such as defending or conquering territory. The relative ability to defend or attack—not simply the balance of power—largely determines the severity of the security dilemma. The offense-defense balance signifies the relative ease of conquest. It is defined as “the ratio of the cost of the offensive forces the attacker requires to take territory to the cost of forces the defender has deployed” (43). The ODB depends on a variety of factors, most importantly the nature of technology and geography, and is assessed assuming the attacker chooses the “best available offense to the defender’s best available defense” (45). This variable raises the prospect—if defense advantage is great enough—that even states much less powerful than their adversaries might satisfy their defense requirements. Offense-defense distinguishability indicates whether states can build defensive forces as a means of signaling benign motives and encouraging arms control. “When offense and defense are completely distinguishable,” Glaser writes, “the forces that support offensive missions do not support defensive missions, and vice versa; when offense and defense are entirely indistinguishable, the forces that support offensive missions can be used as effectively in defensive missions, and vice versa” (45). The distinguishability variable is crucial for Glaser’s theory because it helps determine whether a state can deploy forces that are useful for self-defense, but not for attacking an adversary, and whether a state can signal benign motives by foregoing forces with offensive potential. According to Glaser, a security-seeking state considering whether to pursue an arms buildup should ask whether doing so would be destabilizing and counterproductive. That is, it should ask whether building more arms would decrease its adversary’s defensive capability, whether its buildup is likely to signal malign motives incorrectly, and whether it would be better to pursue security gains more cooperatively. Offense-defense variables point the way to answering these questions: all other things being equal, defensive advantage and the ability to distinguish between offensive and defensive forces should allow states to protect themselves, signal their benign motives, and avoid negative military and political spirals. Under these conditions, rational security-seeking states should opt for arms control, defensive force postures, and unilateral restraint. Defense dominance renders doing so less risky, and distinguishability makes benign communication more feasible. In sum, when offense-defense conditions permit, security seekers should aim to maintain the military capabilities required to deter and defeat adversaries, but also seek to preserve other states’ security by avoiding policies that undermine opposing military capabilities. According to Glaser’s theory,

454

K. A. Lieber

the degree to which a state will have difficulty in balancing these potentially conflicting strategic objectives is largely a function of the severity of the security dilemma, which, in turn, is determined largely by the ability of states to measure and understand offense-defense variables.

Downloaded by [Florida International University] at 17:23 17 September 2011

The Limits of Net Assessment Versus the Demands of Estimating the Balance A key claim in Glaser’s argument is that statesmen can ascertain the ODB. The analytic tasks required to comprehend the ODB, Glaser asserts, are “the same as those required for military net assessments—analyses of the ability of a state’s forces to perform military missions against the forces of an opponent” (139–41). If states can perform net assessment, they can estimate the ODB: Both require analyzing the ability of a defender’s forces to succeed against an attacker’s forces. Standard net assessments analyze the ability of the forces deployed by the attacker to defeat the forces deployed by the defender. The same tools can be adapted to determine the ODB. Instead of focusing on deployed forces, to measure the balance we allow the attacker’s forces to vary. The balance is defined where the offense is large enough to defeat the defense; the ratio of the cost of these offensive forces to the cost of the defensive forces is the offense-defense balance (140).

Glaser acknowledges that accurately estimating the ODB is complex, requiring a number of demanding assumptions and judgments—but he is optimistic that none of the required tasks are unmanageable (138). Equating net assessment with measuring the ODB belies a fundamental misunderstanding. Net assessment simply cannot yield the kind of precise knowledge necessary for states to make the strategy and force structure decisions called for in Glaser’s theory. War is a highly uncertain business, so the tools of net assessment are blunt; they are not intended to produce fine-grained results. The best that net assessment can typically yield, even with sophisticated combat modeling, is a simple range of possible outcomes: most commonly, that the attacker will likely win, the defender will likely win, or the outcome of the war is uncertain because the two forces are relatively evenly matched. Of course, that determination is often suitably accurate for states to use as a general guide for force procurement and deployment decisions. But the methods of net assessment cannot somehow be fine-tuned to produce the much more precise assessments that states would require in order to opt for the strategic policies Glaser recommends. Net assessment can be undertaken in many different ways, but it essentially involves the analysis of relative military capabilities. The process involves analyzing the interaction of opposing forces in realistic battle scenarios in order to determine strategic vulnerabilities and opportunities against

Downloaded by [Florida International University] at 17:23 17 September 2011

Mission Impossible

455

a given adversary in a given theater of operations. Net assessment typically begins with basic questions: Who is the adversary? What are the conflicting political objectives? Where would a military conflict take place? What forces would be involved and what is their expected quality? Given the actors, objectives, location, and forces involved, net assessment seeks to determine the military strategies that each state can be expected to employ to achieve its objectives, the appropriate models or analytical tools to assess the relative military effectiveness of those strategies, the results produced by those models, and the sensitivity of those results to changes in basic parameters (especially alternative forces, locations, strategies, and the models themselves). The typical payoff for undertaking this exercise is some indication of whether a given military conflict is likely to be a closely fought competitive battle, or whether one side is far superior to the other. In other words, either one actor will appear to have a clear military advantage, or the situation will appear competitive enough that the ultimate outcome of battle will depend on any number of factors—including relative skill or will, luck, the friction of war, and other unanticipated strategic wildcards—that cannot be fully predicted beforehand. The analytical community has developed increasingly sophisticated military models to predict battle outcomes, but the basic limitations on net assessment remain. Among several excellent recent net assessments—for example, of the US-Iranian conflict in the Strait of Hormuz and US military operations at Tora Bora in Afghanistan—none purport to do more than suggest likely outcomes and all are careful to qualify the findings. Peter Krause argues that the US battle at Tora Bora in 2001 would have been more likely to have succeeded with more troops, but concludes that operations would have been difficult and complicated, success was by no means guaranteed, and much about the outcome would have depended on logistics, intelligence, al Qaeda’s strategic choices, the performance of various US and allied forces, and the highly unpredictable factors of weather and terrain where the fighting would occur. Similarly, Eugene Gholz’s analysis of Iran’s ability to cripple the supply of oil to the global market and Caitlan Talmadge’s evaluation of the potential interaction of Iranian and US military forces in the Strait are models of net assessment; each aims to improve public debate and strategic decision making by suggesting how different assumptions about military capabilities might generate different predictions about likely outcomes. Neither predicts just how a US-Iranian conflict would turn out, recommends specific US force requirements, or purports to overcome a host of military and nonmilitary factors that would preclude making such concrete conclusions.4

4 Peter John Paul Krause, “The Last Good Chance: A Reassessment of U.S. Operations at Tora Bora,” Security Studies 17, no. 4 (October 2008): 644–84; Eugene Gholz and the LBJ School Hormuz Working Group, “Threats to Oil Flows through the Strait of Hormuz,” (unpublished manuscript, March 2010); and

Downloaded by [Florida International University] at 17:23 17 September 2011

456

K. A. Lieber

Glaser’s theory contends that we can take net assessment—with its already demanding information requirements, as well as modest objectives— and ask it to do much more. According to Glaser, states should adapt net assessment techniques to measure the ODB in order to “maintain the military capabilities required to deter and defeat adversaries, but also to preserve other states’ security by signaling benign motives and avoiding policies that undermine opposing military capabilities” (53). The goal is to meet a state’s military requirements for successful defense while simultaneously minimizing the threat those capabilities pose to an adversary. This Goldilocks strategy (as I label it) aims to build just the right amount and kind of military force: not too much, not too little, not too menacing, not too reassuring. Leaving aside the question of whether this balancing act is theoretically possible, in practice it would require information that far exceeds the capability of existing military modeling tools. Consider, for example, the debate over the NATO-Warsaw Pact conventional military balance in Europe in the 1980s. Analysts disagreed on basic assumptions about forces, strategies, and doctrines, as well as the relative merits of various models used to predict battle outcomes, but most views fell into one of two schools: those who argued that NATO was hopelessly overmatched against a conventional Warsaw Pact attack and those who believed that NATO might be able to defend successfully. The most serious debate occurred within the latter group, with some analysts believing that NATO could probably thwart a Pact attack and others contending that NATO would probably lose. But both the qualified optimists and the qualified pessimists understood that the conventional balance was highly competitive—the outcome of the battle would largely depend on whether NATO received adequate strategic warning of an attack and mobilized its forces accordingly.5 None of the analyses attempted to define the precise point at which NATO forces would be just strong enough to stop a Warsaw Pact attack, let alone determine the cost ratio of building those hypothetical forces. Other examples abound. Most analysts agree that the Israeli Defense Forces (IDF) could defeat a Syrian invasion today. But figuring out exactly how much the IDF might cut its air and ground forces to the point that it could just barely defend Israel without unnecessarily threatening Syria would be analytically impossible. It would also be strategically foolish. In 1973, the Israelis believed the IDF to be unassailable. But a coalition of Arab states led by Syria and Egypt succeeded in launching a surprise attack, and the vastly superior IDF came perilously close to being swept from the Golan Heights, which would have given the Syrians a short, open road to Tel Aviv. Had the Caitlin Talmadge, “Closing Time: Assessing the Iranian Threat to the Strait of Hormuz,” International Security 33, no. 1 (Summer 2008): 82–117. 5 See John J. Mearsheimer, Barry R. Posen, and Eliot A. Cohen, “Reassessing Net Assessment,” International Security 14, no. 4 (Spring 1989): 128–79.

Mission Impossible

457

aimed lower and tried to build a Goldilocks force, it would have likely suffered a decisive defeat. The problem is manifest even in more limited conflict scenarios that have been closely studied for years. For example, what is the Goldilocks force for the US mission to maintain air supremacy against China in the event of a regional conflict? American planners have identified the air battle as a vital component of the broader strategy to defend allies in the region, and have thus analyzed it carefully. Yet the tools of net assessment cannot tell us what amount and kind of US capabilities would meet minimum military requirements without unnecessarily threatening Chinese security. Even with sophisticated net assessment, military planners today face great uncertainty about US capabilities against Chinese air and air defense capabilities during a regional conflict. Some analysts believe US air power could easily gain and maintain air supremacy. Production of the F-22 Raptor fighter ended in 2009 in large part because US planners were satisfied with existing capabilities for any air superiority mission.6 Others argue that US air supremacy is not only doubtful in a future conflict scenario against China, but also uncertain if a conflict were to occur today. In the assessment of these analysts, Chinese surface-to-surface missile attacks against air bases in Japan and Guam combined with surface-to-air missile attacks would dramatically degrade US air capabilities and defeat the air superiority mission. This view calls for a huge bolstering, not cancellation, of F-22 production simply to meet minimum requirements to conduct the defensive mission.7 Serious analysts using sophisticated tools of net assessment have advanced both positions, with no clear answer. The more ambitious type of net assessment required by Glaser’s theory—to determine a force that would just barely win—would likely generate even greater uncertainty. Net assessment is often used as a valuable tool to inform political judgment and strategic choice, but one should not get carried away in thinking that these analytical techniques can somehow indicate just the right level of risk for a state to assume in the pursuit of peace and security.

Downloaded by [Florida International University] at 17:23 17 September 2011

IDF

One Goldilocks Force, Many Missions There is a second major reason why Glaser’s optimism about a state’s ability to make useful assessments of offense-defense variables is unconvincing. Even if one could overcome the substantial practical obstacles to measuring 6

See, for example, Robert M. Gates, “DOD News Briefing With Secretary Gates from the Pentagon,” 6 April 2009, http://www.defense.gov/Transcripts/Transcript.aspx?TranscriptID=4396; and Michael Donley and Norton Schwartz, “Moving Beyond the F-22,” Washington Post, 13 April 2009. 7 See, for example, Gary Schmitt and Thomas Donnelly, “Shore Up America’s Air Superiority,” Wall Street Journal, 17 January 2011; Michael Auslin, “The Case for Reviving the F-22 Fighter,” Wall Street Journal, 24 February 2011; and Mark Bowden, “The Last Ace,” The Atlantic, March 2009.

Downloaded by [Florida International University] at 17:23 17 September 2011

458

K. A. Lieber

the ODB and using that understanding to guide force production and deployment decisions, the causal logic of the theory is inapplicable in a world where states have more than one adversary, pursue security interests beyond their own territory, or face multiple military missions. Glaser suggests that the narrow, dyadic version of his theory can be extended: “A natural next move would be to address more fully situations that include more than two major powers” (271). But doing so would undermine the theory’s core normative prescriptions. Simply put, the right force structure for a single dyad—that is, that which reduces the adversary’s vulnerability and signals benign motives without compromising a state’s own security—will likely be the wrong force structure for another dyad. Furthermore, because forces can be concentrated, several separate Goldilocks forces inevitably undermine the purported security benefits of each individually tailored force. The theory developed in Rational Theory of International Politics seeks to explore the security dynamics between two states. It assumes a state can determine the ODB for a single conflict scenario against one adversary in a lone theater of operations. The theory calls for states—under permissive conditions—to build and deploy a force that balances the capability to deter and defeat that adversary without undermining the adversary’s own security. In the real world, many states—and all great powers—plan for a variety of conflicts against multiple adversaries in several theaters of operation. Any force that addresses a range of threats in a range of scenarios will have great capacity for offensive operations in any single theater. For example, even assuming that the ODB favored the defense, and offensive and defensive forces were distinguishable, how could the United States configure its armed forces to be able to defend South Korea from invasion, protect Japan, keep East Asia’s sea lanes open, defend Kuwait and other Gulf Allies, keep the Strait of Hormuz open, and fulfill US commitments to NATO without having a force that could be concentrated for an overwhelming invasion of North Korea? What imaginable force would be barely sufficient to prevail in each of these missions without unnecessarily threatening potential adversaries? Or imagine that offense-defense conditions allowed China to build just sufficient air and naval power to keep the US Pacific Fleet at bay—but nothing more. In trying to convey benign motives and build just enough capability to stalemate US forces in the Pacific, China would inevitably generate enormous security fears in Japan, Indonesia, Australia, and other regional states. In short, a Goldilocks force against one adversary is likely to be excessive against another. Even if a state were able to adapt net assessment to usefully estimate a favorable ODB and tailor a separate military force for each mission, the fact of the matter is that these forces could likely be concentrated for battle in any single campaign. Even in the Cold War, when defense against a Warsaw Pact invasion in Western Europe was the main strategic challenge, US forces earmarked for other missions—including naval and air assets

Downloaded by [Florida International University] at 17:23 17 September 2011

Mission Impossible

459

around the world—could have contributed substantially to the central mission if war occurred. Similarly, even if the United States could configure a Goldilocks force in East Asia aimed at allaying Chinese insecurity, many other US capabilities (particularly naval forces) deployed to other regions for other purposes could be concentrated in the region fairly quickly. To be sure, any state faces potential costs to concentrating its forces—for example, in accepting temporary vulnerability against threats elsewhere—and adversaries are likely to understand and seek to capitalize on those costs (perhaps by allying with those other threatening states in order to undermine a concentration strategy).8 Yet understanding such regional and global strategic dynamics highlights the problem with analyzing the offense-defense balance—and a Goldilocks force—between a pair of states in isolation. In short, the sum of separate mission forces—or even the combination of a fraction of these forces—is likely to exceed the level necessary to signal benign motives, minimize the adversary’s insecurity, and create conditions favorable for a cooperative security spiral. In the real world of international relations, states facing multiple security missions would see little sense in trying to simultaneously estimate the minimum force necessary to meet security requirements across the range of scenarios, the maximum capability they could deploy without unduly threatening their adversaries in these scenarios, and the right kind of force that would split the difference. Only a state without multiple adversaries, alliance commitments, and distributed interests would seem to have a chance of following such policy prescriptions. Unfortunately, this rules out virtually all major-power states today.

8 Note that a coordinated strategy among states to balance against the potential concentrator would require those countries to adopt greater offensive forces and strategies (to be able to threaten the potential concentrator) than would be necessary for an adequate measure of security in a purely dyadic context. Even if many other hurdles were overcome, defining, developing, and deploying a Goldilocks force in this context appears impossible.

Security Studies, 20:460–468, 2011 Copyright © Taylor & Francis Group, LLC ISSN: 0963-6412 print / 1556-1852 online DOI: 10.1080/09636412.2011.599196

Rational Theory for a Bygone Era RANDALL L. SCHWELLER

In its breathtaking comprehensiveness, analytic power and clarity, creative fusion of theoretical materials, and sheer density of logic, Rational Theory of International Politics qualifies as a must-read for any serious student of international politics.1 Although I have some minor quibbles with parts of the analysis, the book strikes me as the last word on the security dilemma, offense-defense balance theory, the spiral model, and, more generally, rational choice theories grounded in Waltzian structural realism. At least I hope it is the last word. The time is long overdue for realists of all stripes to lay down their swords on these issues and move on. There is no time to waste, either, for the train left the station well over a decade ago. Anyone who has witnessed Charles Glaser’s mind in action quickly learns that only a fool would choose to debate him on his own terms. Accordingly, the thrust of my critique concerns the book’s relevance to today’s problems. It seems to me that Rational Theory is more about adjudicating and synthesizing claims made by contending schools of thought than about present concerns and issues confronting states and other actors on the world stage; it is more about contemporary international relations theory than about contemporary international relations. The book presents a theory of international politics rooted in classical eighteenth-century balance-of-power logic, a world largely defined by territorial concerns that determine who has power, prestige, and security; a world in which states live in constant fear of territorial attacks by greedy and insecure neighbors. Its portrayal of international politics would be very familiar to Napoleon Bonaparte, who, if he suddenly awoke from his long sleep and read the book, would conclude that little had changed in his absence. But the world has changed and in dramatic ways. Therein lies the problem. Classical balance-of-power theory—no matter how much it is

A previous version of this paper was presented at the annual meeting of the American Political Science Association (APSA), Washington, D.C., 2010. Randall L. Schweller is Professor of political science at Ohio State University. 1 Charles L. Glaser, Rational Theory of International Politics: The Logic of Competition and Cooperation (Princeton: Princeton University Press, 2010), subsequent references cited in text.

460

Rational Theory for a Bygone Era

461

modified by only slightly more modern “Cold War-era” theories about the security dilemma and offense-defense balance—is an anachronistic lens, one that filters out most of what is most important in current world affairs. It is not only the wrong end of the telescope but the wrong instrument through which to view the world.

Downloaded by [Florida International University] at 17:23 17 September 2011

TERRITORY, TERRITORY, TERRITORY Glaser’s theory combines three material variables—power, the offensedefense balance, and offense-defense distinguishability—with two “information” variables that revolve around state motives. How are these independent variables defined? The short answer is by means of territory and territorial concerns. Territory is front and center in all of Rational Theory’s independent variables and core concepts. Power is defined by Glaser as “the ratio of the state’s resources that can be converted into military assets to the adversary’s resources”; it determines how a state’s “potential military compares to the military that an adversary could build.” Capabilities are limited to “military capabilities,” which refer to the “state’s ability to perform military missions” (41–42). Thus, the concept of power boils down to a state’s ability to defeat or deter rivals on the battlefield, its ability to take territory or defend territory. Does this narrow view of state power and capabilities make sense in today’s world? How much of the warp and woof of twenty-first century world politics can this limited view of power be expected to explain? Similarly, the offense-defense balance is defined as “the relative effectiveness of forces deployed to defend (hold) territory compared to forces deployed to attack (take) territory” (4). Measuring the offense-defense balance “requires adapting net assessment tools that are designed to evaluate the outcome of military battles” (138). State motives—central to the theory’s information variables—are defined in terms of greed and security. A state can be either a greedy non-security territorial expander, or a greedy but also security-seeking territorial expander, or a pure security seeker that may seek territorial expansion but only as a consequence of the dangerous environment it finds itself in. Security is defined as the “state’s prospects for preserving control of its territory, avoiding war to protect its territory, and suffering low costs in fighting if war occurs” (35). This definition of security does not include the state’s ability to protect its prosperity and way of life more generally or to prevent coercion and competition that would transform the state or undermine its prosperity and values. Next come the information variables. One would imagine that they would have something to do with the information or digital age—a revolution so profound that it rivals the industrial revolution. Consider this fact: every day the National Security Agency alone intercepts and stores nearly two

Downloaded by [Florida International University] at 17:23 17 September 2011

462

R. L. Schweller

billion separate e-mails, phone calls, and other communications. In Rational Theory, however, information has nothing to do with the information age. Instead, it refers to beliefs about others’ intentions. Is the adversary motivated by greed or security? Information also refers to what the state thinks the adversary thinks about its motives (but this dimension drops out of the analysis). The book’s goal is, in Glaser’s words, to construct “a theory of foreign policy. Most generally, states choose between cooperative and competitive strategies; more specifically, yet still broadly defined, states can choose between building up arms, acquiring allies, negotiating arms control agreements, making concessions, and launching wars” (26). These are the state’s objectives? This is the entire menu of state cooperation and competitive strategies in the twenty-first century? Glaser goes on to say that the state can achieve its objectives by acquiring military capabilities to deter attack, to coerce concessions, to defend territory, and to take territory or by influencing its adversary’s objectives, bringing them in line with the state’s own objectives (34). The book’s discussion of cooperation and competition focuses on arming decisions. Cooperation refers to coordinated policies designed to avoid arms races and improve political relations; competition refers to unilateral military buildups that can generate arms races and the formation of alliances. Foreign policy therefore revolves around three fundamental issues: territory, alliances, and arms buildups. Does this sound like the concerns of powerful states these days? Not to me. Glaser claims that his rationalist, strategic-choice theory makes the following assumptions: states exist under anarchy, are unitary rational actors, and are black-boxed by the theory (no looking inside at their domestic politics). True, but he leaves out the big and important assumptions that drive his analysis, all of which are rooted in classical balance-of-power logic. First, states expect to settle their differences by fighting. War is not just possible but anticipated by the Great Powers. Second, territory is the key component of power, security, and wealth. Territorial concerns dominate foreign policy goals and strategies of cooperation and competition with others. Third, territorial expansion—even greedy, non-security expansion—is a legitimate goal of states. There is no norm against taking over countries or grabbing someone else’s territory. Do any of these assumptions hold today? I think not.

THE TRANSFORMATION OF WORLD POLITICS The digital age is fundamentally changing international politics. It is transforming the nature of power, the ability to exercise power, the way power is exercised, and the goals of power itself. To be sure, control over resources and access to routes (for example, sea lanes vital for the transport of oil and

Downloaded by [Florida International University] at 17:23 17 September 2011

Rational Theory for a Bygone Era

463

projecting military power) will continue to be strategic objectives, especially among the Great Powers.2 And disputes over territory and regional influence will continue to fuel rivalry among the emerging powers themselves and with the United States. This is most evident between China and India, two countries that share a disputed border extending roughly two-thousand miles and that remain keenly sensitive to changes in their relative military capabilities. Finally and most generally, people will continue to define their identities largely by means of territory: nationalism will continue to hold greater sway over international politics than cosmopolitan universalism. But geography and policies rooted in geopolitics have become less relevant to the formation of strategy and politics than they were when raw materials and land were the major prerequisites for state power. We no longer live in a world governed by the logic of the mercantilist age, when military conquest to control territory and achieve autarky (or a monopoly on goods) was the surest route to riches and power. Today, the traditional link between territory and wealth has been largely broken. The current era of high technology, instant communication, and nuclear weapons has significantly raised the benefits of peace and the costs of war. What matters most today is not a state’s ability to exert direct control over resources but its capacity to purchase them in a free global market. Accordingly, the foundation of modern state power has shifted away from traditional military power toward an emphasis on economic production and a sustained capacity to generate ideas and commercial innovations that create wealth. To be perfectly clear on this point, innovation and economic growth remain key building blocks of military power; I am not suggesting otherwise. Rather, I am saying that military power is no longer an essential building block of economic growth and wealth creation; and this has deeply changed the nature of international politics and how the game is played. Given the new realities of the nuclear revolution and the rise of globalization and “knowledge economies,” territory is no longer the coin of the realm in terms of power assets. Taking more of it no longer makes states safer or more powerful. Indeed, land grabs, naked or clothed, will certainly make the state less safe and secure in the end. When was the last time a security-seeking state actually gained security through territorial conquest? I ask the question because Glaser features this scenario as a “very important” and counterintuitive insight of the book.3 2

See Jakub J. Grygiel, Great Powers and Geopolitical Change (Baltimore: The Johns Hopkins University Press, 2006), chap. 7. 3 That said, Glaser’s point echoes the familiar logic at the core of the security dilemma. As Robert Jervis observed decades ago, when the security dilemma is at its most vicious, “the only route to security lies through expansion. Status-quo powers must then act like aggressors; the fact that they would gladly agree to forego the opportunity for expansion in return for guarantees for their security has no implications for their behavior.” Robert Jervis, “Cooperation Under the Security Dilemma,” World Politics 30, no. 2 (January 1978), 187.

464

R. L. Schweller

Downloaded by [Florida International University] at 17:23 17 September 2011

Although security is a benign motive, a state seeking security could find territorial expansion to be desirable and therefore be willing to engage in an arms race and start a war to change the status quo. Expansion could increase a state’s security in a variety of ways, including increasing its resources and resource autonomy; decreasing its adversary’s resources and in extreme cases eliminating its adversary as a sovereign state; and improving its ability to employ its resources effectively by, for example, providing a buffer zone against invasion, strategic depth, or more defensible borders. (36)

Glaser’s logic is unassailable. And yet, I can think of few recent cases of a state gaining security by attacking its neighbors or eliminating its adversary as a sovereign state. Leaving security aside, Glaser focuses a good deal of his attention, as well he should, on states motivated by greedy expansion (some of which also have security concerns).4 Glaser’s distinction between greedy and securityseeking states is a necessary refinement of neorealism’s more simple assumption that all states are security maximizers—an assumption that, by leaving out power maximizers motivated by pure greed, creates a “status-quo” theoretical bias that consigns a considerable number of great powers in modern history as exceptions to be accounted for.5 Indeed, from a neorealist perspective (as well as a defensive realist one), most great power behavior in the pre-1945 era is not only inexplicable but pathological.6 Given the amount of space and analytic firepower Glaser devotes to the motives and behaviors of greedy states, however, one might reasonably conclude that interstate wars of aggression were still commonplace in international politics; that conquest and territorial expansion were still rational policies for accumulating state power. Yet, here again, I cannot think of a single recent case (or even one since 1945) when a state gained power and prestige by conquering its neighbors. If territorial expansion is still a genuine

4 Opportunistic expansion has always been a core principle of realism because, in the words of John Mearsheimer, “Anarchy and uncertainty about other states’ intentions create an irreducible level of fear among states that leads to power-maximizing behavior . . . ; the structure of the international system, not the particular characteristics of individual states, causes them to think and act offensively and to seek hegemony.” John J. Mearsheimer, The Tragedy of Great Power Politics (New York: W. W. Norton, 2001), 43, 53. Disagreeing with this view, Glaser and other non-offensive realists limit the generalizability of Mearsheimer’s claim by various scope conditions, including the offense-defense balance, the cumulatively of resources through conquest, and whether the state is motivated by greed, security, or some combination of the two. 5 See Randall L. Schweller, “Neorealism’s Status-Quo Bias: What Security Dilemma?” Security Studies 5, no. 3 (Spring 1996): 90–121. 6 For a critique of defensive realism along these lines, see Fareed Zakaria, “Realism and Domestic Politics: A Review Essay,” International Security 17, no. 1 (Summer 1992), esp. 190–98. In his defense of structural realism (especially of the Gilpin variety), Zakaria makes a significant refinement to neorealism: states are not assumed to be security maximizers but rather influence maximizers; that is, the stronger a state gets, the more influence it wants. See ibid., 194, n. 43.

Downloaded by [Florida International University] at 17:23 17 September 2011

Rational Theory for a Bygone Era

465

goal among powerful actors today, why has interstate war, especially major war, become such a remarkably rare phenomenon since the end of the Cold War?7 Why have there been so few cases—and virtually no successful ones—of greedy expansion since 1945? Indeed, if a greedy state is one defined by its “nonsecurity motives for expansion, which can include the desire to increase its wealth, territory, or prestige, and to spread its political ideology or religion, when these are not required to preserve the state’s security” (36), then one would be hard-pressed to identify such a state—much less a “purely greedy state” (one entirely without security motives)—in the world today. Peter Liberman may be correct that sufficiently ruthless occupiers can extract gain from territory, especially in industrialized countries, even in the modern world.8 But if conquest still pays, leaders of contemporary states do not seem to know or believe it.9 The absence of greedy states is especially puzzling in what was formerly known as the Third Word, which consists of regions defined by significant power inequalities among neighboring states. According to balance-of-power realism, capability imbalances are dangerous precisely because they encourage opportunistic expansion. Yet, as Jeffrey Herbst observes, very few Third World states have fought interstate wars or even confronted major external threats. Even in Africa, the continent seemingly destined for war given the colonially-imposed boundaries and weak political authorities, there has not been one involuntary boundary change since the dawn of the independence era in the late 1950s, and very few countries face even the prospect of a conflict with their neighbors. Most of the conflicts in Africa that have occurred were not, as in Europe, wars of conquest that threatened the existence of other states, but conflicts over lesser issues that were resolved without threatening the existence of another state.10

Likewise, K. J. Holsti comments, “The search for continental hegemony is rare in the Third World, but was a common feature of European diplomacy under the Habsburg, Louis XIV, Napoleon, Wilhelmine Germany, Hitler, and Soviet Union and, arguably, the United States.11 7 See John Mueller, The Remnants of War (Ithaca: Cornell University Press, 2004); John Mueller, “Why Isn’t There More Violence?” Security Studies 13, no. 3 (Spring 2004): 191–203. 8 Peter Liberman, Does Conquest Pay? The Exploitation of Occupied Industrial Societies (Princeton: Princeton University Press, 1996). 9 For a “domestic politics” explanation for the puzzle of under-aggression and under-expansion since 1945, see Randall L. Schweller, “Neoclassical Realism and State Mobilization: Expansionist Ideology in the Age of Mass Politics,” in Neoclassical Realism, the State, and Foreign Policy, eds. Steven E. Lobell, Norrin M. Ripsman, and Jeffrey W. Taliaferro (Cambridge: Cambridge University Press, 2009), chap. 8. 10 Jeffrey Herbst, “War and the State in Africa,” International Security 14, no. 4 (Spring 1990), 123. 11 K. J. Holsti, “International Relations Theory and Domestic War in the Third World: The Limits of Relevance,” in International Relations Theory and the Third World, ed. Stephanie G. Neuman (New York: St. Martin’s Press, 1998), 106.

Downloaded by [Florida International University] at 17:23 17 September 2011

466

R. L. Schweller

To be sure, headlines that dominate the news today—from Iran to DPRK to China-Japan-Vietnam to US-China—clearly suggest that plenty of states still care about territory. These issues, however, are a far cry from past bids to take over the world or a large portion of it. Indeed, the United States’ most likely peer competitor—China—more resembles Wal-Mart than the former Soviet Union. Modern leaders seem to understand that, in the twenty-first century, states move up the ladder of international power and prestige by means of knowledge economies that generate dynamic growth; that the present and future competition among states will be decided by technological innovation, connectedness within global networks, the ability to steer complex technological innovation processes, and who best creates environments that facilitate flexible and timely innovation choices. These are the kinds of issues Americans focus on—or should focus on—when they consider how to prevent their country’s declining global position from accelerating. At present, Americans maintain fears that have little to do with security defined narrowly in terms of whether other states will attack the United States or its allies.12 They understand that America’s position in the world will be largely determined by the health of its economy—an economy saddled by enormous public debt set to double in the coming decade from $5.8 trillion in 2008 to $14.3 trillion in 2019. They have seen America’s share of world product fall 32 percent since 2000;13 unemployment remains at nearly ten percent; foreclosures have forced millions of Americans out of their homes; and real incomes have fallen faster and further than at any time since the Great Depression. Americans worry about having less influence in the world, about putting the country’s fate in the hands of others. They worry about China and other emerging countries taking market share from US companies, about dangerously high current account deficits, about the effects of globalization and outsourcing on the average American’s standard of living, about the welfare and education of their children and grandchildren, about terrorism, cyberspace crime and attacks, the spread of infectious diseases and the potential emergence of a new pandemic, nuclear proliferation, climate change, energy, healthcare, savings, pensions, and biosecurity. Add to this list of potential troubles the decline of the US dollar, which arguably poses the most serious threat to the future of American power. If the US dollar relinquishes 12

They do fear terrorist attacks, however, and their fears are largely unwarranted and overblown. For the most recent and thorough discussion of this phenomenon, see John Mueller and Mark Stewart, Terrorism, Security, and Money: Balancing the Risks, Benefits, and Costs of Homeland Security (New York and London: Oxford University Press, 2011). 13 Robert A. Pape, “Empire Falls,” The National Interest 99 (January/February 2009): 21–34. For contrary views, see Stephen G. Brooks and William C. Wohlforth, World Out of Balance: International Relations and the Challenge of American Primacy (Princeton: Princeton University Press, 2008); and, to a lesser extent, Fareed Zakaria, The Post-American World (New York: W. W. Norton, 2008).

Rational Theory for a Bygone Era

467

Downloaded by [Florida International University] at 17:23 17 September 2011

its status as the world’s sole reserve currency to the Renminbi, gold, SDR, Bancor, or something else, not only will its value plunge but US credit markets will collapse, commodity prices and interest rates will soar, and the Federal Reserve’s balance sheet will explode beyond all reason.14 These are the issues and security threats that will determine the fate of America and the world in the twenty-first century, whether the United States remains far and away the most dominant country in the world, falls back to the pack, or, worse still, simply collapses.15 Only madmen still believe that the path to security and greatness lies in imperialism and territorial conquest.

WHERE DO WE GO FROM HERE? Am I proclaiming the triumph of Davos-style liberal cosmopolitanism and the death of realism?16 Certainly not. International politics is still fundamentally a competition for power and influence over others and the environment. As in the past, states want to maximize their wealth, influence, security, prestige, political autonomy, and freedom of choice. But the world has changed in important ways. To better fit the realities of the twenty-first century, therefore, realism must be uprooted from its foundation in the industrial age, geopolitics, and concerns about military power and territory to the exclusion of everything else that matters in the world. Realism must become more a theory of consumption and influence maximizing—shaping others’ preferences to get the outcomes you desire—than one about military capabilities and security defined narrowly as safety from territorial attack. To this end, realists must reconceptualize their core tenets in a way that retains the perspective’s focus on competition, power, prestige, and influence-maximizing but reconsiders the nature and bases of power assets, how power is and can be exercised, the goals and fears of modern states, and the routes to security and prosperity in the new millennium. Otherwise, realism will tell us less about what makes the clock tick in contemporary international politics than it otherwise could. What realism needs 14 For various views on the future status of the US dollar, see Barry Eichengreen, Exorbitant Privilege: The Decline of the Dollar and the Future of the International Monetary System (New York: Oxford University Press, 2011); Barry Eichengreen, “The Dollar Dilemma: The World’s Top Currency Faces Competition,” Foreign Affairs 88, no. 5 (September/October 2009): 53–68; C. Fred Bergsten, “The Dollar and Deficits: How Washington Can Prevent the Next Crisis,” Foreign Affairs 88, no. 6 (November/December 2009): 20-38; Roger C. Altman and Richard N. Haas, “American Profligacy and American Power: The Consequences of Fiscal Irresponsibility,” Foreign Affairs 89, no. 6 (November/December 2010): 25–34. 15 See Niall Ferguson, “Complexity and Collapse: Empires on the Edge of Chaos,” Foreign Affairs 89, no. 2 (March/April 2010): 18–32. 16 Referring to the annual meeting of Westernized elites in Switzerland, Samuel Huntington coined the term “Davos culture,” comprising an elite group of highly educated people who operate in the rarefied domains of international finance, media, and diplomacy and share common beliefs about individualism, market economics, and democracy. See Samuel P. Huntington, The Clash of Civilizations and the Remaking of World Order (New York: Touchstone, 1996), chap. 3.

468

R. L. Schweller

Downloaded by [Florida International University] at 17:23 17 September 2011

is an overhaul; what it gets from Glaser’s book are incremental modifications of defensive and structural realism. I, for one, have little interest in further debates among offensive realists, defensive realists, classical realists, structural realists, and neoclassical realists. Give me, instead, a realist theory for the twenty-first century.

Security Studies, 20:469–489, 2011 Copyright © Taylor & Francis Group, LLC ISSN: 0963-6412 print / 1556-1852 online DOI: 10.1080/09636412.2011.599197

Defending RTIP, Without Offending Unnecessarily CHARLES L. GLASER

Maybe the greatest honor that a scholar can receive is for leading scholars to take his work seriously. Thus, I greatly appreciate the time and care that the contributors to this Security Studies symposium have invested in providing critiques of my recent book, Rational Theory of International Politics (RTIP).1 Of course, critiques would be uninteresting if they did not offer incisive criticism. Thus, while I appreciate the commentators’ praise for RTIP, I also look forward to this opportunity to respond to a diverse range of issues their essays raise. Because there is little overlap in the essays’ key themes, my response deals with them sequentially, instead of working to integrate across them.

GOALS FOR RTIP To set the stage for my responses, I would like to highlight four of the goals that I had for RTIP. First, I strove to construct a comprehensive statement in which individual arguments add up to more than the sum of their parts and can be appreciated as part of a larger whole, which reflects an important lens through which to view international politics. RTIP develops a stronger foundation—by motivating its choice and characterization of key assumptions and variables—and a sharper, more fully systematic presentation than is possible in an article or even series of articles. The case for including three Charles Glaser is Professor of Political Science and International Affairs at the George Washington University and Director of the Institute for Security and Conflict Studies at the Elliott School of International Affairs. In addition to Rational Theory of International Politics, he is the author of Analyzing Strategic Nuclear Policy (Princeton, 1990). He would like to thank Eric Grynaviski, Jim Goldgeier, Harris Mylonas, and Tristan Volpe for their helpful comments on an earlier draft of this essay. 1 Charles L. Glaser, Rational Theory of International Politics: The Logic of Competition and Cooperation (Princeton: Princeton University Press, 2010), subsequent references cited in text. Some of these essays build on a panel on RTIP that was held at the APSA 2010 Annual Meeting. Web videos of this panel can be viewed at http://media.elliott.gwu.edu/video/89.

469

Downloaded by [Florida International University] at 17:24 17 September 2011

470

C. L. Glaser

types of variables—motives, material, and informational—the analysis of the security dilemma and signaling, and my conclusions about the conditional desirability of international cooperation and competition are all stronger as part of this integrated package. Second, I wanted to make clear the theory’s boundaries and its complementary relationship to other types of theories. I emphasize that the rational theory developed in RTIP can be envisioned as sitting in the middle of three layers of theory—between those that explain the inputs to the rational theory and those that explain suboptimal behavior, that is, deviations from it. The layers do not compete with each other. They stand on their own, are complementary, and can be combined to create a more complete explanatory theory. Third, I tried to emphasize that the core theory quite naturally could be extended to encompass a wider range of theoretical questions and to incorporate greater complexity when more nuanced assessments were required. RTIP sketches a variety of what I term “within theory” extensions. These extensions flow directly from the theory’s key variables, allowing for variation within types of states (that is, within security seekers and greedy states), uncertainty about material variables, and changes over time in both material and information variables. Introducing this greater complexity leads toward a richer theory that preserves the defining characteristics of the core theory. This extended theory does not compete with the core theory; it is better envisioned as increasing the core theory’s reach, sometimes at the cost of parsimony. Fourth, I worked to demonstrate the strength of the rational theory. This task is much less straight forward than usual, because RTIP recognizes that states do not consistently adopt rational/optimal policies. As I explain below, this does not reduce the value of the rational theory. But it does mean that we cannot evaluate it by simply exploring how well it explains states’ behavior. RTIP employs three approaches instead. I devote a chapter to evaluating the theory “from within”: exploring the adequacy of its assumptions, the completeness of its variables, the feasibility of its decision-making requirements, and the deductive strength of its logic. A second chapter adopts a more familiar approach, demonstrating that the theory provides good explanations for a number of important historical cases. A third chapter employs counterfactual analysis to demonstrate that states that adopted suboptimal arming policies—especially arming when they should have engaged in unilateral restraint or bilateral cooperation—were hurt by their errors. In combination, these approaches provide substantial confidence in RTIP’s core theory.

NARROW BOUNDARIES AND DIFFICULTIES OF SIGNALING Robert Jervis divides his critiques into those that plague all realist theories and those that apply only to defensive realism and the more general

Downloaded by [Florida International University] at 17:24 17 September 2011

Defending RTIP, Without Offending Unnecessarily

471

framework developed in RTIP. He identifies a number of ways in which realism’s basic framework is too narrow, including its focus on security-seeking motives—while excluding honor, glory, status, and altruism—and its limited scope for addressing domestic politics and individual ideology. These are important issues for RTIP because if we are going to use its rational theory as a baseline against which states’ actual behavior is judged, then we need to understand whether its assumptions are adequate and whether the relevant variables are included. RTIP addresses some of these issues in chapter 7, which evaluates the theory from within. I explore the theory’s key assumptions and argue that, while the anarchy, rationality, and unitary actor assumptions are adequate, one assumption—black-boxing the adversary—is not. An adversary’s domestic system could influence its ability to extract and convert resources, which could influence its effective power; changes in the adversary’s regime type and domestic institutions could provide the state with information about its adversary’s motives; and the state’s policy could influence the balance of domestic politics within the adversary’s state, which could in turn shift both the adversary’s motives and adversary’s information about the state’s motives. Opening the black box in any or all of these ways would provide a fuller rational theory and, I conjecture, that under some conditions a state’s preferred strategy could be different from the one identified by the RTIP baseline. Consistent with Jervis’s critique, I believe this is likely to be the most important direction for expanding RTIP beyond its current boundaries. This chapter also explores whether important variables are left out of RTIP, focusing on variables that are not a natural extension of the theory. Among the possibilities are motives other than greed and security—which could include the altruism that flows from collective identities and humanitarian interests—and international norms. Although all of these could influence a rational state’s strategy, I argue that, unlike the black-box assumption, none is especially important for the central questions that RTIP is designed to address. Jervis argues that honor, glory, and status can lead to quite different state behavior; their implications relative to security and greed deserve to be explored as well. Reflecting on this chapter, I should note that I consider it to be something of an experiment for evaluating a theory—I could not find examples of this type of analysis—and a challenging one, especially to handle relatively briefly. Although undoubtedly biased by having invested enormous effort puzzling through these issues, I do believe the undertaking has been fruitful, while recognizing that it leaves room for the analysis to be pushed further. Turning specifically to RTIP, Jervis addresses the difficulties inherent in costly signaling, a subject on which he speaks with special expertise. Jervis identifies two separate but related problems: the ambiguity in a state’s signal is often quite large, leaving it open to multiple and often contradictory

Downloaded by [Florida International University] at 17:24 17 September 2011

472

C. L. Glaser

interpretations; and states’ and especially their leaders’ assessments are likely to be biased. Jervis raises a large subject that I cannot deal with adequately here, so instead I offer a few quick points. According to Jervis, standard costly signaling arguments tend to overlook the degree to which actions are ambiguous. His examples of the deeply skeptical reactions that some US leaders had to quite substantial unilateral restraint by Soviet leaders Nikita Khrushchev and Mikhail Gorbachev provide a powerful illustration of this problem. At first consideration these actions appear to be quite unambiguous—what could be clearer than unilateral restraint? However, the problem may be less severe than it seems initially. First, we need to look more closely at the strength of the alternative/skeptical arguments. Examining the specific argument, beyond its general logic, may expose weakness. For example, cutting defense spending does not always contribute much to economic growth, so we need to ask, for example, whether Khrushchev’s force reductions could have fueled sufficient growth to significantly reduce future US security. If not, then a rational actor should reject this interpretation, which greatly reduces the ambiguity of the signal. Second, we need to ask whether all costly signals are open to such divergent interpretations. It seems likely that certain actions taken in a specific international situation are clearer than these examples of unilateral restraint. For example, what if a state that enjoys a large power advantage agrees to a ratio of deployed forces that is much smaller than its power advantage? This was the policy the United States adopted toward Japan in the interwar years. What if two countries agree to limit offense and these limits will not generate large savings, thereby vitiating the argument that savings will generate economic growth. Limiting multiple independently targetable reentry vehicles (MIRVs) in the 1970s would fit this description. Third, if an action is ambiguous, a state that focuses on one of two or more sound interpretations is not acting rationally. If there are multiple sound interpretations, then a rational decision maker should weigh the probability of the alternative explanations, or at least recognize that there are competing explanations and modify the information the action provides accordingly. In some situations this properly balanced assessment will restore some useful information to the signal. Biased assessments of the adversary’s actions raise a quite different set of issues. One reaction is simply to dismiss this concern because bias is a form of suboptimality and therefore lies outside of RTIP’s boundaries; some of my defense of the rationality assumption follows this basic tact. However, two additional points are relevant here. First, although Jervis’s work on misperceptions demonstrates the serious limits of individuals’ capabilities, states can reasonably be expected to do significantly better than individuals. Especially for noncrisis decisions, states can rely upon well-established analytic units that are staffed by experts with diverse views, or who are at least well versed in the competing views; can create units across the government that

Downloaded by [Florida International University] at 17:24 17 September 2011

Defending RTIP, Without Offending Unnecessarily

473

have different institutional missions and perspectives; and can integrate these perspectives into a balanced state-level assessment. And even if states have often failed to build such sophisticated domestic structures for evaluating their adversaries’ actions, the question from the perspective of the rational theory is whether doing so is within their means. Second, although a nonrational/biased adversary does not fit within RTIP’s boundaries, the theory can be extended to provide useful policy guidance, if a state knows that its adversary’s interpretation of its signals will be biased. This would be a large project, but a couple of examples help illustrate how its arguments might evolve. If facing an adversary/leader that was known to be biased toward exaggerating the extent of the state’s malign motives, the state should recognize that communicating its benign motives will take larger and/or repeated costly signals. In addition, the state should recognize that the adversary’s failure to reciprocate restraint is less likely to be an indication of its own greedy motives, reflecting instead its flawed interpretation of the state’s signals. Of course, none of this is costless. A revised strategy that requires sending larger signals could be too risky, making a competitive strategy the state’s best option, and introducing recognition of the adversary’s bias into the state’s decisions introduces greater complexity into an already challenging analytic task, which could make it unmanageable.

WHAT IF NOT ALL STATES ARE RATIONAL? John Mearsheimer focuses on what might appear to be an inconsistency in RTIP—my theory addresses rational states, but I also acknowledge that actual state behavior has not always met this standard.2 This latter point reflects recognition of research over the past few decades that has found that states often adopt flawed policies and that has developed a range of theories to explain this suboptimal international behavior. As a result, Mearsheimer offers two broad, related critiques: RTIP is of little use because it does not explain state behavior; and RTIP’s prescriptions will be dangerous for rational states that in fact live in a world plagued by suboptimal behavior. Both critiques raise interesting issues, but Mearsheimer greatly exaggerates the problems and limitations they create for RTIP. Mearsheimer’s first critique focuses on explanation, arguing that RTIP will only be useful if there is a “great transformation in state behavior” 2 At the beginning of his essay Mearsheimer notes a number of specific disagreements with RTIP but says his essay will take the theory on its own terms. I appreciate Mearsheimer’s effort to do this, thereby moving us beyond disagreements that we have already engaged. However, I want to note here that Mearsheimer’s essay does include some characterizations with which I disagree: he overstates my conclusion on possibilities for cooperation and peace, arguing that RTIP holds that “the structural constraints of the international system should discourage great powers from adopting aggressive strategies and especially from starting wars,” when this is true only under certain conditions.

Downloaded by [Florida International University] at 17:24 17 September 2011

474

C. L. Glaser

that results in all states becoming rational, which he correctly does not foresee any time soon. Until then, RTIP cannot explain state behavior, so it “simply provides a baseline for assessing how states should think about” their strategic choices. Here Mearsheimer goes astray by focusing too heavily on the standard criterion for evaluating a theory—its ability to explain past behavior—and consequently overlooks other uses of the theory. First, the value of a rational baseline for contributing to explanations of behavior should not be minimized. Most international relations (IR) theories of suboptimal behavior rely on a rational baseline for determining what policy a state should have chosen. A baseline is necessary because whether a state’s behavior is suboptimal is often not obvious. Bad outcomes may not reflect suboptimal policies; instead, the material and informational constraints facing a state can mean that its best option will lead to a bad outcome. We require a baseline to sort out the bad outcomes from the bad (that is, suboptimal) strategies. In addition, establishing an agreed upon rational baseline has been the central objective of much of the grand IR theory debate, which can be understood as evaluating rational state versus rational state interaction. Disagreement about how to characterize the international situation that states face—that is, which material and information variables to include—combined with the complexity of the interactions that the international system generates has resulted in a decades-long disagreement over the proper rational baseline. This debate, of course, is the debate over realist and rationalist theories of international relations, including Kenneth Waltz’s structural realim, offensive realism (Mearsheimer), defensive realism, arguably neo-institutionalism, and now RTIP. Mearsheimer finds that states are consistently rational, strategic calculators because he has a different theory—that is, a different rational baseline—than defensive realism and RTIP, which he believes explains the vast majority of state behavior. Obviously, reaching consensus on the baseline is no small challenge and the IR theory stakes are high. As I explain in RTIP, resolution cannot be achieved simply by taking the divergent theories to the historical data. Second, a rational baseline will often play an essential role in a theory that explains suboptimal state behavior and therefore makes a significant contribution to our ability to explain state behavior. By identifying which variables are important, the baseline theory defines the types of misperceptions, flawed evaluations, and inefficient behaviors that could underpin suboptimal behavior. For example, Jack Snyder and Stephen Van Evera point to exaggerations of offensive capabilities and of the adversary’s hostility as key sources of unnecessary arms racing and war;3 these exaggerations are 3 Stephen Van Evera, Causes of War (Ithaca: Cornell University Press, 1999); and Jack Snyder, The Ideology of the Offensive: Military Decision Making and the Disasters of 1914 (Ithaca: Cornell University Press, 1984).

Downloaded by [Florida International University] at 17:24 17 September 2011

Defending RTIP, Without Offending Unnecessarily

475

not even relevant to a theory that excludes these variables, such as offensive realism. Related, by exploring how variations in these variables influence states’ choices, the theory enables us to understand whether these mistakes actually matter: not all flawed assessments of relevant variables lead to suboptimal policies; RTIP enables us to evaluate when they do and how much these errors then cost the state. In this role, a rational theory can be combined with theories of suboptimality to generate a compound theory that explains suboptimal behavior. In my view, contributions to either part of this compound theory can stand on their own. As I described above, RTIP, which is a strategic choice theory, lies in the middle layer of a full explanatory theory; theories of suboptimality lie in the third layer and can be combined with it. Because the questions that I wanted to cover in RTIP already constituted an ambitious book project, I decided against developing a compound theory of my own. I was further discouraged from pursuing this route both because a range of theories of suboptimality were already well established and because I am skeptical that there is a theory that explains adequately when the various types of suboptimality are likely to occur. Before leaving this rejoinder I should make one additional point—we should not lose track of the fact that RTIP will by itself explain state behavior when states do act rationally. Although suboptimal behavior is not uncommon, states do often act rationally and in these cases RTIP does explain their behavior. I have sketched a few such cases in chapter 8, where I argue that my theory explains better than the previously established theories. Mearsheimer’s second critique holds that RTIP cannot be safely used to prescribe a state’s strategy because opposing states may be acting recklessly—that is, not in accord with the rational theory—which renders the theory’s prescriptions dangerous instead of helpful. This concern is more serious than his explanation critique, but again is off the mark. First, RTIP provides valuable guidance if only by correcting misunderstandings about how the international system works, which can themselves lead states astray. For example, if a state believes incorrectly that the international system consistently drives all states into competition, then it will overlook situations in which restraint and/or cooperation may be its best option.4 The state may well need to consider the possibility that its adversary may adopt suboptimal policies that would undercut the case for cooperation. But without RTIP (and the closely related defensive realist theories) this question will not even arise. Second, although the possibility of a nonrational state certainly complicates a rational state’s strategic calculations, it does not, as Mearsheimer 4 For an example of how this plays out in a current debate see John J. Mearsheimer, “The Gathering Storm: China’s Challenge to US Power in Asia,” Chinese Journal of International Politics 3, no. 4 (Winter 2010): 318–96; and Charles Glaser, “Will China’s Rise Lead to War? Why Realism Does Not Mean Pessimism,” Foreign Affairs 90, no. 2 (March/April 2011): 80–91.

Downloaded by [Florida International University] at 17:24 17 September 2011

476

C. L. Glaser

holds, as a general rule require states to pursue competitive, offensive policies designed to increase their power and thereby maintain a hedge against reckless states. It is certainly true that a strategy that is best when facing a rational adversary might not be best when facing an irrational one. However, cooperative or restrained policies could remain a state’s best strategy and a state may have some options for exploring the increased risks. A key factor cutting against competition remains the fact that adversaries that adopt flawed policies can be insecure (a point that Mearsheimer touches on at the end of his essay). In fact, an adversary’s misjudgment of its international environment can make it more insecure both by generating military assessments that exaggerate its vulnerability and by leading it to choose military strategies that damage its political relationships and further increase its military vulnerability. The “cult of the offensive” explanations for the First World War lay out this logic.5 Because the adversary’s insecurity is a source of danger, the state has incentives to adopt policies that reduce this insecurity, even though it is exaggerated, as well as to protect itself against attack. This is the familiar security dilemma, except here it is exacerbated by misjudgments. A state’s prospects for effectively dealing with this type of situation will be reasonably good when the actual (to be distinguished from the misperceived) security dilemma is not severe. This would include, for example, a case in which defense has the advantage but the state’s adversary incorrectly believes that offense has the advantage or at least that an offensive doctrine is best matched to its strategic requirements. A number of historical examples help to illustrate this point. Before the First World War, both Germany and France adopted offensive doctrines when their international situation actually favored defensive doctrines. Both would have been more secure with a defensive doctrine, even if the other retained its offensive one.6 Similarly, during the Cold War both the United States and the Soviet Union exaggerated the potential of nuclear offense, pursuing highly accurate multiple-warhead ballistic missile systems designed to destroy each other’s forces, even though they had little chance of using these counterforce systems to reduce the damage they would suffer in an all-out war. The United States would have been better off foregoing nuclear counterforce and focusing on increasing the survivability of its forces, even if the Soviet Union continued to pursue its nuclear counterforce systems.7 In these cases, defensive strategies offer two advantages: they have the potential to signal that the state has benign motives; and they promise to provide equal or greater military capability for a given size investment.

5

Van Evera, Causes of War, esp. chap. 7; and Snyder, The Ideology of the Offensive. Snyder, The Ideology of the Offensive; and Jack Snyder, “Correspondence: The Origins of Offense and the Consequences of Counterforce,” International Security 11, no. 3 (Winter 1986/87): 187–93. 7 On the debate over Cold War nuclear counterforce see Charles L. Glaser, Analyzing Strategic Nuclear Policy (Princeton: Princeton University Press, 1990), chap. 7. 6

Downloaded by [Florida International University] at 17:24 17 September 2011

Defending RTIP, Without Offending Unnecessarily

477

Third, when unilateral defensive policies would be too risky, a state may be able to use arms control—either tacit or negotiated—both to determine whether its adversary is rational and to achieve military capabilities that would otherwise be infeasible. For example, offering restraint in the size of forces to an equally powerful state or offering limits on offense when it has the advantage should under a range of conditions generate a positive response. If the adversary rejects the state’s offer, the state may have reasons to question its adversary’s rationality, but in any event is no worse off, because it will not then limit its own forces and risk the accompanying vulnerabilities. Finally, as I touched on in my response to Jervis, RTIP provides the foundation for developing a theory that can prescribe strategies for dealing with reckless states, that is, states that are pursuing suboptimal policies.

2-2-2 AND 2 OBJECTIONS James Fearon nicely characterizes the core model in RTIP as the 2-2-2 model: two states—the state making the decision and its potential adversary; two types of states—security seeking and greedy; and two actions—cooperation and competition. The theory is, however, significantly richer than this synoptic view might suggest. The types of states are defined along two dimensions, so even RTIP’s most basic formulation identifies four different types, including importantly states that are simultaneously greedy and security seeking. The core of RTIP’s theory deals with these types, with much of its attention going to pure security seekers, but with careful attention also paid to greedy security seekers. In chapter 4, which addresses extensions to the core theory, RTIP identifies the possibility of variation within these types and explores some of the implications. Maybe the most interesting variation is in the value that security seekers place on their security—they could vary in the value they place on protecting their homeland, protecting securityproviding allies, and/or acquiring additional territory that would increase their security. This variation is important not only because it should affect the state’s own behavior, but also its adversary’s behavior. Adversaries, however, may be uncertain about the security seeker’s valuation and this uncertainty adds an entirely new information dimension to the theory. Now a security seeker has incentives not only to communicate that it is not a greedy state, but also to communicate the extent of its security interests, that is, to communicate its resolve. This further complicates the security dilemma and enables RTIP to integrate more fully the deterrence and spiral models. Also potentially quite important is variation within greedy states. Greed can vary in depth (how much a state will pay for nonsecurity expansion) and breadth (whether a greedy state has limited or unlimited aims). Both of

Downloaded by [Florida International University] at 17:24 17 September 2011

478

C. L. Glaser

these types of variation in the adversary’s greed influence the risks of cooperation; the question of limited versus unlimited aims plays a central role in judgments about the advisability of appeasement when limited concessions would increase the adversary’s power. Similarly, RTIP divides cooperation into a variety of different possibilities.8 For example, a state could cooperate via arms control—engaging in mutual restraint in the deployment of arms achieved through either formal or tacit bargaining; defense emphasis—assigning priority to meeting its military requirements with a defensive strategy, even if this costs more than an equally effective offensive strategy; and unilateral restraint—reducing its military capability below what it believes is required for defense and deterrence. The arms control option is further divided into qualitative and quantitative arms control.9 RTIP explains the options’ relative risks and benefits, and how they vary with the material and information conditions that a state faces. In short, although I accept the 2-2-2 characterization, it is important to appreciate that RTIP presents a richer set of possible variations and explores their strategic implications. One of Fearon’s major objections is that pure security seekers rarely if ever exist—that is, all states want to change something about the status quo—in which case RTIP’s heavy attention to this type of state is misplaced. In particular, he worries that small amounts of greed can influence states’ interactions. Although I agree that pure security seekers are likely rare, making them a focus of RTIP is sensible for a couple of reasons. First, whether international structure—traditionally defined in terms of the material conditions facing a state, to which RTIP adds information conditions—can generate competition between security seekers has long been one of the central questions addressed by international relations theorists, who were trying to sort out whether international conflict is rooted in the nature of individual states or instead in the pressures created by the international system. Of course, both could be important sources. Waltz’s answer was that structure is sufficient to generate conflict, but his theory did not provide a fully satisfactory explanation. Work on the security dilemma—as Fearon notes, by Jervis, Andrew Kydd, and me—helped to fill in this important gap. Fearon’s point is not that this is a wasted effort—he sees greater attention to greedy security seekers as an important next step.

8 The latter two options are not strictly cooperation because they do not require the adversary to match the state’s restraint. I note this in RTIP, but nevertheless for ease of presentation decided to use the term cooperation to refer to all forms of restraint (50, n2). 9 In Fearon’s essay in note 3, he argues that RTIP “has little to say about . . . how to think about arms control in theoretical terms, for example, as coordination on one of multiple equilibria in the 2-2-2 model under some parameter conditions.” RTIP does however explore how the incentives for an agreement vary with the offense-defense balance, offense-defense distinguishability, a state’s power, and its information about the adversary’s motives; it also explains the importance of monitoring, reaction time, rearmament rates, and the forces that are allowed by an agreement.

Downloaded by [Florida International University] at 17:24 17 September 2011

Defending RTIP, Without Offending Unnecessarily

479

Second, states that can be reasonably approximated as pure security seekers are likely far more common than Fearon suggests. To illustrate his objection, Fearon focuses on the US preferences toward Iraq prior to 2003, Iran, and North Korea and argues that the United States is at least somewhat greedy toward these regimes—it dislikes these regimes and would be willing to pay some prices for regime change. I do not necessarily disagree with this characterization of the United States, but I think these are easy cases for Fearon’s position because they involve states that face a huge power asymmetry. The harder cases, and the ones that are most important for RTIP’s theory, involve powers that are more closely equal in power, especially major powers. In this situation, pursuit of greedy motives will take a dedicated effort, requiring a major power to invest large amounts in arms and/or to risk fighting a very costly war. If the state is only a little greedy (compared to the value that it places on its security), then it should essentially ignore its greedy motives and focus entirely on its security motives. In these cases, the simplifying assumption of pure security seeking enables RTIP to capture the essential features of the state’s decision calculus. This approach is especially likely to be effective if pursuit of greedy motives would divert scarce resources from military missions dedicated to protecting the state from attack and/or would damage the state’s political relationship by signaling malign motives. This set of considerations captures important aspects of US Cold War policy. Although the United States valued freeing Eastern Europe from the Soviet Union’s grasp, it decided early in the Cold War that using force to roll back Soviet control in Eastern Europe was simply too risky. In addition, acquiring the offensive capabilities required to achieve this goal would have conflicted with NATO’s doctrine, which was restrained partly by the desire not to unnecessarily threaten the Soviet Union. The result was a military doctrine in Europe that did not reflect the United States limited greedy motives. The situation facing the United States after the Cold War is distinctly different, among other reasons because US baseline conventional forces—those that it maintains to perform a wide variety of global missions—are sufficient to severely challenge all small and medium powers and they now lack a major power ally. None of these comments are intended to suggest that states with mixed motives—that is, greedy security seekers—are unimportant. Accordingly, RTIP addresses them—including considering possible trade-offs between pursuing security and nonsecurity goals—although certainly there is more to be said about cases in which states find this a difficult trade-off. Fearon’s second major concern is that RTIP gets off on the wrong foot by assuming that states are motivated by security; instead, he argues, security should be considered among the means available to states that are pursuing consumption as their end. I agree that consumption is more fundamental and

Downloaded by [Florida International University] at 17:24 17 September 2011

480

C. L. Glaser

therefore has the potential to provide a still more general and solid foundation for the theory.10 My choice of security was primarily a pragmatic one: focusing on security is a more direct way to get at most of the key strategy choices that states usually face. Moreover, much of what is lost analytically by not starting with consumption can be captured (albeit less fundamentally) by arguing that states value both security and consumption, and therefore see a trade-off between investing in military forces and consuming these resources. Although I do not run the cost of investing in arms consistently through RTIP’s analysis, I do identify its important role in a variety of decisions, including that states may not maximize their security because they also value consumption (37); that states can communicate resolve by purchasing arms because this entails an opportunity cost with consumption (95); that the security dilemma can leave a state worse off, even if it does not reduce its security, due to the costs of arming (60); and that a policy of defense emphasis demonstrates the value a state places on not threatening its adversary, because the defensive posture is more expensive than the comparable offensive one (67).

MANY RATIONAL THEORIES? Dale Copeland’s primary concern is that although RTIP presents a “plausible rational theory,” it is not set “against alternative rationalist theories that might challenge either the veracity of [Glaser’s] deductive logic or his interpretations of history.” In other words, RTIP is only one of many possible rational theories, so how do we know it is the “right” one? As Copeland notes, relying on the wrong theory as the rational baseline against which states’ behavior is judged could result in concluding that a state pursued suboptimal policies when in fact it was acting rationally. I agree with Copeland’s basic concern.11 However, I think he miscasts the challenge, overlooks some ways in which RTIP responds to it, and at best misreads my short historical cases of key arms races.12

10 I do note the alternative formulation (40). One possible complication with envisioning consumption as a state’s end is that it does not distinguish between the resources a state currently controls and resources it could control through expansion; in other words, the greedy versus not greedy distinction that is captured by security seeking would have to be reestablished through some other type of distinction. 11 I disagree, however, with Copeland’s claim that alternative theories are necessary for exploring the veracity of the deductive logic of RTIP; this is a question of internal validity. 12 In addition, just to avoid confusion, I would like to note that Copeland is wrong on the following. In note 6, he contends that I deal with “state motives” only briefly and that “this extra variable represents a degenerative aspect of Glaser’s research program.” However, my treatment of greedy states runs throughout RTIP and is intertwined with many of its arguments; see for example the discussion of state types in chapter 2 and the discussion of variation within greedy states in chapter 4. It is true that my discussion of greedy states is briefer than of security seekers; this is largely because security seekers present more puzzles, including why they get into conflicts and how they can signal their motives. Notice, however, that

Downloaded by [Florida International University] at 17:24 17 September 2011

Defending RTIP, Without Offending Unnecessarily

481

I address the issue of possible other rational theories in two parts of RTIP. Chapter 7, which I touched on briefly in my response to Jervis’s comments, explores the adequacy of the theory’s assumptions and the completeness of its variables, both of which bear directly on the possibility of alternative rational theories that do not flow from RTIP’s key variables. I argue in that chapter that there are a variety of more general rational theories, with opening the unit-level black box likely being the most important for the key questions RTIP is designed to address. The bigger point in the context of Copeland’s criticism is that RTIP is sensitive to the possibility of other rational theories, and I have in fact identified at least some of the most important possibilities. The second place that RTIP addresses ways of moving beyond its core theory is in the chapter on extensions, and it is these arguments that are of most concern to Copeland. In contrast to the variables discussed above, I term these “within theory” extensions because they build directly on the core theory’s key variables. The extensions I consider include uncertainty about material variables, and changes over time in material and information variables, as well as variations within types of states, which I mentioned in the previous section. My purpose in this chapter was to sketch a variety of possible extensions of the core theory developed in chapters 2 and 3. I did not attempt to explore them as thoroughly as I did RTIP’s core theory. My view is that these within theory extensions can be added to the core theory to create a more complex and far-reaching theory, a fuller theory without changing its defining characteristics. I do not see these extensions, either on their own or combined with the core theory, as competing against the core theory. My point is not that adding variables raises no new issues; it certainly does. But these possibilities flow from the same types of calculations, involve the same types of trade-offs, and involve the same variables that lie at the center of the basic model.13 I think that it is in the same spirit that formal modelers see the step from complete information to incomplete

my discussion of security seekers also necessarily addresses the choices of greedy states—for example, in the discussion of signaling benign and malign motives—which reduces the need to address certain issues when focusing specifically on greedy states. In addition, RTIP demonstrates why its rational theory needs to include greedy states—because there are situations in which greedy states should compete, but pure security seekers should not (see for example, 12–13, and 87); the case for Waltz and offensive realism addressing greedy states is less strong, because they find that security seekers should always compete. Finally, Copeland’s “mop up” comment is misleading. While I do address the possible role of greedy motives in fueling some arms races, in each case I conclude that the competitions reflected suboptimal choices and cannot be explained by greedy motives. My discussion of Adolf Hitler focuses on showing that his expansionist policies cannot be explained by security, which was my key goal in that short discussion, and I leave open whether greed is sufficient to explain his policies or instead whether suboptimality is also required. Contrary to Copeland’s claim in footnote 9, my brief discussion does apply to the early stage of the superpower arms race; I do not think fears of the future are required to explain this competition, but agree they would supplement the risks RTIP identifies. 13 The one obvious exception is the need to discount outcomes across time.

Downloaded by [Florida International University] at 17:24 17 September 2011

482

C. L. Glaser

information as a generalization of a basic model, and adding changes in variables over time can be understood as a generalization of a static model.14 Copeland rejects this formulation, arguing instead that the extensions should be considered alternative rational theories that need to compete against each other and holds that debates between these theories can only be resolved by “in-depth empirical tests.” Copeland focuses on what he terms a “dynamic version of systemic realism,” which allows variation in a state’s future power and in its adversary’s type. In contrast, employing the general logic I described above, I see this dynamic theory as very closely linked to the static theory—the dynamic theory necessarily builds on the static theory. If a state believes that power influences its security today, then it has to believe that power will also matter tomorrow; consequently, if the state expects its power to change in the future, then it must take this into consideration, comparing how its options today compare to its options in the future. How could it be any other way? Copeland disagrees, arguing “none of [Glaser’s] variables are conceptualized in dynamic terms, so it is not clear exactly how his logic could be ‘extended’ to include a state’s expectations of future power and state type.” This is not a problem, however, because the theory does not need to explain why the variables change, but can instead simply assume that they do. Change is exogenous to the extended theory, just like the sources of power are exogenous in the static theory, and future outcomes, which are now influenced by variation across time in key variables, are included in the state’s decision by discounting the future.15 Moreover, it is less clear than Copeland suggests what we would learn about these theories through empirical testing. If power is changing and the dynamic theory explains a state’s choice of strategy, then this finding can provide support for the static version, because it provides the foundation for the dynamic version; in contrast, Copeland sees support for the dynamic theory undermining the static theory. In addition, if the dynamic version explains many cases, while the static version explains few cases, we may simply be learning about the empirical frequency of different international situations, which is not a fundamentally theoretical issue.16 14 Although he is not addressing this issue, I think my point is consistent with Robert Powell’s analysis; he notes that “chapter 4 [which is on bargaining in the shadow of shifting power] generalizes the model developed in chapter 3 [which is on bargaining in the shadow of power] to examine how two states cope with a shift in the distribution of power.” See Robert Powell, In the Shadow of Power (Princeton: Princeton University Press, 1999), 19; see also 116. 15 Copeland argues further that my section on dynamic considerations is designed to protect the conclusions offered by RTIP ’s core theory. However, my purpose was quite different. My goal for this section was to provide a brief survey of how the incentives for preventive war vary with the theory’s central variables. I do point to how certain conditions reduce the incentives for preventive war, and at least implicitly thereby acknowledge that other conditions result in larger incentives. 16 Unfortunately, space limitations prevent me from responding at any length to Copeland’s position on rational arguments that consider commercial access and trade. Although these considerations do raise new issues, their implications are not nearly as dramatic as Copeland suggests. Needing to protect a

Downloaded by [Florida International University] at 17:24 17 September 2011

Defending RTIP, Without Offending Unnecessarily

483

Finally, I feel compelled to respond to Copeland’s assault on my evaluation of specific arms races. He says that I “simply label” US policy on MIRV suboptimal. In fact, I describe how MIRV was going to affect the offensedefense balance and why offense and defense were distinguishable for this technological innovation. I explain that although MIRV would reduce the degree of defense advantage, the United States would not be able to acquire a damage-limitation capability, even though it enjoyed a power advantage. I then review two considerations that complicated the case for arms control: there were a variety of missions for MIRV, each of which needs to be taken into account in assessing the impact of MIRV on the offense-defense balance; and there was the possibility that a MIRV ban could not be verified adequately. While acknowledging some grey areas, I conclude that banning MIRV was the rational choice. My analysis, while brief, draws on a range of authoritative sources and is not weakened by Copeland’s comment about US concerns about decline and Soviet cheating. His assessment of my evaluation of the German army buildup in the years before World War I is no more evenhanded. Contrary to Copeland’s claim, finding that a state should launch a preventive war does not challenge RTIP’s core theory (or its dynamic extension), so I have no stake either way in judging whether Germany’s buildup was suboptimal. In fact my bottom line is that whether Germany should have launched its army buildup depends on a related, but largely separable, debate over whether Germany required an offensive doctrine. Given the large amount of expertise devoted to this question, I decided that a conditional analysis was more productive than taking sides in this unresolved debate.

GIVE UP ON ESTIMATING THE OFFENSE-DEFENSE BALANCE Keir Lieber argues that RTIP is flawed because states are unable to assess the offense-defense balance. More specifically, responding to my argument that if states can perform useful net assessments, they have the analytic tools required to assess the offense-defense balance,17 Lieber argues that “net assessment simply cannot yield the kind of precise knowledge necessary for states to make the strategy and force structure decisions called for in Glaser’s theory.” In addition, Lieber argues that my theory cannot be applied when a state faces more than one adversary. I address these arguments in turn.

distant sea-lane is not very different from needing to defend a distant ally. RTIP notes that a security seeking state could need an offensive doctrine to protect allies, even when defense has the advantage and that this requirement can generate a complex tradeoff (78–79). Also contrary to Copeland, for all these types of cases, the offense-defense balance remains exogenous to state interaction; it is true that the relative importance of different military missions is endogenous to a state’s international environment. 17 I first made this argument in an article coauthored with Chaim Kaufmann. See Charles L. Glaser and Chaim Kaufmann, “What Is the Offense-Defense Balance and Can We Measure It?” International Security 22, no. 4 (Spring 1998): 44–82.

Downloaded by [Florida International University] at 17:24 17 September 2011

484

C. L. Glaser

Lieber’s argument about net assessment starts from the premise that the offense-defense balance needs to be measured precisely, but this is not the case. While more precise and confident estimates would always be helpful, under a range of conditions uncertain estimates of the offensedefense balance will still enable states to make different and better judgments than if their strategic decisions looked only at power. Knowing whether the offense-defense balance is close to or far from 1 is often sufficient for contributing both to explanations of behavior and prescriptions of strategy.18 This will remain true if the estimate of the balance is rather uncertain. Consider for example a state that prefers an offensive strategy and has power that is equal to its adversary’s power. Assume that its modal estimate is that the offense-defense balance equals 2, but has some probability of being anywhere between 1.5 and 2.5, which is a wide spread around the most likely value. This state has very poor prospects for deploying a force that has a decent chance of winning on the offense. In contrast, if this state does not estimate the offense-defense balance and considers only power, its estimate of its prospects for a successful offensive would be greater and misleading. This of course raises the empirical question of whether states actually face situations in which their estimates are this far from 1 and with uncertainty that that does not dwarf this gap. The easiest example is the nuclear balance between the United States and the Soviet Union—numerous detailed reliable net assessments demonstrated that defense (retaliatory capabilities) had a very large advantage over offense (damage-limitation capabilities), with no possibility that balance was close to 1. Nuclear is not the only compelling example, however. There are good examples of net assessments and related historically informed calculations of conventional capabilities that also meet this standard. Before the First World War, German assessments of the requirements for the Schlieffen Plan to succeed suggest an offense-defense balance of 2 or greater. Assessments of the conventional balance in Europe during the Cold War strongly suggested that defense had a substantial advantage, with the tactical balance likely closer to 3 than to 1.19 During the interwar period, Japan relied on a variety of considerations—including the required ratio of the attacking to the defending fleets in a local battle and the possibility of destroying US forces as they crossed the Pacific—to settle 18

The offense-defense balance is defined as the ratio of the cost of forces that the attacker requires to take territory to the cost of the forces that the defender has deployed. Thus the offense-defense balance is 1 when the cost of the offensive and the defensive are equal. 19 The theater-wide balance for a war of attrition may have been less favorable to defense, but influential arguments held that this type of war was deterred by its high costs. See John J. Mearsheimer, Conventional Deterrence (Ithaca: Cornell University Press, 1983). Lieber argues that analysts on all sides of the debate over NATO’s conventional capabilities recognized that the outcome of a conventional war would be sensitive to NATO’s warning and mobilization. The balance, however, should be assessed assuming both sides have the opportunity to mobilize, and first-move advantages are then treated separately. See Glaser and Kaufmann, “What Is the Offense-Defense Balance,” 71–72.

Downloaded by [Florida International University] at 17:24 17 September 2011

Defending RTIP, Without Offending Unnecessarily

485

on a requirement that Japan’s navy needed to be 70 percent of the size of the US Navy.20 Even Lieber’s argument that recent net assessments fail to shed light on the offense-defense balance is unconvincing. Consider Caitlin Talmadge’s assessment of the US ability to counter an Iranian disruption of shipping through the Strait of Hormuz.21 Although complicated by many uncertainties, Talmadge finds that Iran could significantly disrupt the traffic for weeks, if not months. To restore the full flow of traffic within weeks, the United States would likely rely upon a wide array of capabilities, including at least one aircraft carrier, Aegis cruisers, and DSP satellites, in addition to mine-clearing ships. The cost of these forces would dwarf the cost of Iran’s ship and missile forces. For this mission, disruption of traffic through the Strait, which could be part of an Iranian offensive strategy, has a clear advantage over US efforts to open the Strait, which would be defensive. Finally, it is worth noting that Stephen Biddle, whose sophisticated analysis of military effectiveness is often invoked as arguing against offensedefense theory, does in effect conclude both that the offense-defense balance can be estimated and that for high intensity ground warfare it favors defense: “Even with modern-system exposure reduction, offensive preponderance cannot create breakthrough against a modern-system defense . . . only the largest invaders outnumber defenders by enough to advance against a modern-system defense without a boost from differential concentration.”22 Biddle’s argument is that states’ skill—their ability to employ forces as required by modern military technology—is more important in determining the outcome of battles than is the offense-defense balance and that it is also more important than power. In line with other rationalist and structural theories, RTIP assumes that states are highly and equally skilled, and the offense-defense balance is defined in interactions between these types of states. For these states, Biddle finds that the offense-defense balance can be measured and is strategically meaningful. Leiber’s second argument concerning net assessment techniques addresses what he terms the “Goldilocks strategy”—using net assessments to determine the perfectly balanced force, one that deters without provoking—which he seems to think that RTIP prescribes as the force a state should deploy. Unfortunately, on this issue he has misunderstood my argument. As he quotes from RTIP, one way to think about measuring the offense-defense balance is to adjust the size of the offensive force such that, according to the net assessment, it is just large enough to defeat the defense; in contrast, a standard net assessment analyzes the forces that are deployed,

20

For a brief discussion of these examples and citations, see RTIP , 140–42, 197–200, and 244–45. Caitlin Talmadge, “Closing Time: Assessing the Iranian Threat to the Strait of Hormuz,” International Security 33, 1 (Summer 2008): 82–117. 22 Stephen Biddle, Military Power: Explaining Victory and Defeat in Modern Battle (Princeton: Princeton University Press, 2004), 69–70. 21

Downloaded by [Florida International University] at 17:24 17 September 2011

486

C. L. Glaser

asking which side is likely to win. How precisely the offense-defense balance can be measured using this approach will depend on the military situation being analyzed. But the issue for Leiber’s Goldilocks strategy is entirely different—he holds that RTIP calls for deploying forces that are just adequate to defend. This is wrong. The passage that he quotes to support this is not about net assessment and measuring the offense-defense balance, but is instead about what security seekers would like to achieve (53). As I wrote in the next sentence, “the severity of the security dilemma influences the difficulty a state will have in balancing these potentially conflicting strategic objectives.” How to strike this balance depends on a number of variables, including a state’s information about its adversary’s motives and its power, as well as offense-defense variables. RTIP does not call for states to deploy Lieber’s Goldilocks force, and under a wide range of conditions a state will clearly have better options. When offense and defense are distinguishable, a state should not adopt one of these knife-edge force postures; it can deploy defense for highly effective deterrence, while avoiding provocation. When a state enjoys a power advantage, it can improve its deterrent by outbuilding its adversary; especially if its adversary is likely to be a greedy state and defense does not have a significant advantage, a large and more threatening force will be the state’s best option. This said, there are conditions under which a state’s best option could be Lieber’s Goldilocks force—for example, if the offense-defense balance is close to 1 and the states are equally powerful, then the best a state can do may be to deploy forces that are barely adequate for deterrence; its adversary would find itself in the same unsatisfactory situation. Unless their political relations were very good, both states would feel threatened by the other. Depending on a variety of factors, this outcome might be best managed via quantitative arms control or might be produced by ongoing competition. Either way, when the security dilemma takes this form, states do not have a good option, Goldilocks or otherwise. Lieber’s final key argument is that even if the offense-defense balance can be measured and the Goldilocks force determined, RTIP is undermined because states usually face multiple adversaries. Not only would the state’s force requirements for dealing with each adversary be different, but the state could always combine all of its forces for use against a single state and its adversaries would recognize this, which undermines the individual Goldilocks effect of each force. Although there is an element of truth in Lieber’s concern, his position is overstated. To start, multiple adversaries and especially multipolarity create greater complexity and uncertainty in all aspects of international politics—compare alliance dynamics in bipolarity to multipolarity, with the potential problems generated by buck passing, bandwagoning, chain ganging, alliance security dilemmas, and more. We should hardly be surprised, therefore, that the same is true for the offensedefense arguments developed in RTIP.

Downloaded by [Florida International University] at 17:24 17 September 2011

Defending RTIP, Without Offending Unnecessarily

487

In addition, there are many international situations in which the challenges created by multiple adversaries are not as severe as Lieber suggests. Again the easy case is nuclear weapons—a state can maintain the ability to deter many major powers simultaneously without acquiring the ability to undermine their deterrent capabilities. This is both because offense and defense are distinguishable—forces dedicated to retaliation need not pose a threat to adversaries’ retaliatory capabilities—and because defense has the advantage. Bipolarity also creates a relatively easy case—although a superpower will need forces for other contingencies, its force requirements for dealing with the other superpower can dominate its force planning. Especially given that states should not usually be building Goldilocks forces, a state’s ability to concentrate the forces it deployed for dealing with other adversaries will often not add tremendously to its ability to attack the opposing superpower and, therefore, would create at most second-order complications. This was true for the United States facing the Soviet Union during the Cold War. Finally, the most interesting issue may concern how the problems created by multiple adversaries change with variation in the offense-defense balance. Defense advantage can often reduce them. For example, in the years preceding the First World War, defense advantage provided Germany with the option of dealing with its multiple adversaries by defending in both the east and west, instead of adopting an offensive doctrine that threatened France and pressured Russia to mobilize quickly.23 In contrast, offense advantage, which would have created significant dangers against a single adversary, would have left Germany with still less satisfactory and more threatening alternatives for dealing with a two-front war.

THROW RTIP INTO THE DUSTBIN OF HISTORY I appreciate Randy Schweller’s spirited critique, but can accept little of it. This is not because I fully reject his characterization of twenty-first century international politics—territory is less valuable, major power war is less likely, and states are relatively more concerned with economic competition than military competition. However, the radical transformation of international politics that Schweller describes needs to be explained, not simply described. If international politics is finally going to be better characterized by peace than war, then we need international relations theories that provide a full understanding of this revolution, among other reasons to assess its robustness. RTIP is well equipped for this task. Moreover, Schweller overstates his case—the possibility of major power conflict is not so diminished that we can afford to ignore it. Closely related, substantial disagreement continues about whether future major-power relations will be peaceful—with much 23

Snyder, Ideology of the Offensive, 116–22.

Downloaded by [Florida International University] at 17:24 17 September 2011

488

C. L. Glaser

concern focused on the implications of China’s rise. Consequently, even if Schweller is correct, we need theories to engage and advance this debate. RTIP offers three reinforcing explanations for the shift to reduced competition and increased security between major powers. First, defense advantage reduces the threat that major powers pose to each other, which reduces the value they place on gaining territory as a means of achieving security. Nuclear weapons provide all major powers with the potential to meet their requirements for deterring attacks—both nuclear and large conventional attacks—against their homelands, and with a bit more effort against their allies. Geography increases the defense advantage for the US-China dyad, as distance and the Pacific Ocean virtually eliminate the possibility of large direct conventional threats, and water contributes to the US ability to defend Japan. This defense advantage ripples favorably through the states’ interactions, enabling them to avoid damaging their political relations by allowing them to avoid arms competition that would signal malign motives. An alternative argument is that the reduced salience of security concerns since the Cold War reflects US unipolarity, not the offense-defense balance. The distinction matters because unipolarity will pass, while defense advantage remains. RTIP finds that the offense-defense balance is more important than the US power advantage, both because the advantage of defense more than offsets the power advantage and because China is not as insecure as a purely power-focused explanation would suggest.24 This line of argument raises the obvious question of why security concerns played a significantly larger role during the Cold War. The key material variables were similar—both superpowers had capable nuclear arsenals and they were separated by the Atlantic Ocean. Part of the answer lies in the different alliance arrangement—the Warsaw Pact and NATO shared a long land border. More important, however, were flawed assessments. Both the United States and the Soviet Union failed to appreciate the extent of security provided by their nuclear arsenals, because they failed to appreciate fully the implications of the nuclear revolution. Making matters still worse, the Soviet Union maintained an offensive conventional doctrine in Europe, when a less costly and provocative defensive option was feasible, which communicated malign motives and increased NATO’s defense challenges. There is a warning here for US-China relations—even if RTIP and Schweller are correct about the potential for avoiding intense competition, realizing this potential depends on both countries choosing the “right” policies. Second, the fact that territory is less important for wealth is recognized by the major powers, which reduces their insecurity by reducing the probability that other powers are greedy. Although the causes of this reduced value lie outside of RTIP, its implications are captured in states’ expectations 24 For a fuller discussion of these issues see Charles L. Glaser, “Why Unipolarity Doesn’t Matter (Much),” Cambridge Review of International Affairs 20, no. 2 (June 2011).

Downloaded by [Florida International University] at 17:24 17 September 2011

Defending RTIP, Without Offending Unnecessarily

489

about others’ motives. Other drivers of nonsecurity expansion are possible, but the greatly reduced value of territory for wealth certainly reduces the probability of facing a greedy adversary. Third, and related, the major powers began the post-Cold War era either with excellent political relations or at least relationships that were still to be determined. Excellent political relationships between the EU countries and the United States go a long way toward explaining why they have not considered balancing against US power advantages. Less good, but not severely strained political relations between the United States and China—neither believes the other is likely to be a highly greedy state—have reinforced the impact of defense advantages, thereby further increasing security, which in turn reduces the value of territory. Beyond these issues of explanation, Schweller has overstated his case by arguing that major-power war is even less likely than it actually is, or at least overlooking a continuing debate on this critical issue. This is belied by Schweller’s recognition that “control over resources and access to routes (e.g., sea lanes vital to the transport of oil and projecting military power) will continue to be strategic objectives for the great powers. And disputes over territory and regional influence will continue to fuel rivalry among the emerging powers themselves and with the United States.” Of course this could be true even if at the same time the value of territory is greatly reduced, and this is Schweller’s position. But acknowledging the continuing existence of the seeds of traditional interstate competition means also that we need theories that can guide states about how best to manage them. Finally, there is the debate over a rising China. While Schweller and RTIP reach similar optimistic conclusions, the US debate remains deeply divided and promises to continue for decades. As Schweller strolls into his postrealist digital age, I hope that RTIP will contribute to clarifying and narrowing this debate, and guiding US policy.