Taking Preferences Seriously: A Liberal Theory of International Politics Andrew Moravcsik

Taking Preferences Seriously: A Liberal Theory of International Politics Andrew Moravcsik This article reformulates liberal international relations (IR) theory in a nonideological and nonutopian form appropriate to empirical social science. Liberal IR theory elaborates the insight that state-society relations—the relationship of states to the domestic and transnational social context in which they are embedded—have a fundamental impact on state behavior in world politics. Societal ideas, interests, and institutions influence state behavior by shaping state preferences, that is, the fundamental social purposes underlying the strategic calculations of governments. For liberals, the configuration of state preferences matters most in world politics—not, as realists argue, the configuration of capabilities and not, as institutionalists (that is, functional regime theorists) maintain, the configuration of information and institutions. This article codifies this basic liberal insight in the form of three core theoretical assumptions, derives from them three variants of liberal theory, and demonstrates that the existence of a coherent liberal theory has significant theoretical, methodological, and empirical implications. Restated in this way, liberal theory deserves to be treated as a paradigmatic alternative empirically coequal with and analytically more fundamental than the two dominant theories in contemporary IR scholarship: realism and institutionalism. Grounding liberal theory in a set of core social scientific assumptions helps overcome a disjuncture between contemporary empirical research on world politics and the language employed by scholars to describe IR as a field. Liberal hypotheses stressing variation in state preferences play an increasingly central role in IR scholarship. These include explanations stressing the causal importance of state-society relations as shaped by domestic institutions (for example, the ‘‘democratic peace’’), by economic interdependence (for example, endogenous tariff theory), and by ideas about national, political, and socioeconomic public goods provision (for example, theories about the relationship between nationalism and conflict). Liberal hypotheses do not include, for reasons clarified later, functional regime theory. Yet the conceptual language of IR theory has not caught up with contemporary research. IR theorists continue to speak as if the dominant theoretical cleavage in the field were the dichotomy between realism and (‘‘neoliberal’’) institutionalism. The result: liberal IR theory of the kind outlined earlier is generally ignored as a major paradigmatic alternative. Worse, its lack of paradigmatic status has permitted critics to caricature liberal theory as a normative, even utopian, ideology. Postwar realist critics such as Hans Morgenthau and E. H. Carr took rhetorical advantage of liberalism’s historical role as an ideology to contrast its purported altruism (‘‘idealism,’’ ‘‘legalism,’’ ‘‘moralism,’’ or ‘‘utopianism’’) with realism’s ‘‘theoretical concern with human nature as it actually is [and] historical processes as they actually take place.’’1 Forty years later, little has changed. Robert Gilpin’s influential typology in international political economy juxtaposes a positive mercantilist view (‘‘politics determines economics’’) against a narrower and conspicuously normative liberal one (‘‘economics should determine politics’’). Kenneth Waltz, a realist critic, asserts that ‘‘if the aims . . . of states become matters of . . . central concern, then we are forced back to the descriptive level; and from simple descriptions no valid generalizations can be drawn.’’2 Liberals have responded to such criticisms not by proposing a unified set of positive social scientific assumptions on which a nonideological and nonutopian liberal theory can be based, as has been done with considerable success for realism and institutionalism, but by conceding its theoretical incoherence and turning instead to intellectual history. It is widely accepted that any nontautological social scientific theory must be grounded in a set of positive assumptions from which arguments, explanations, and predictions can be derived.3 Yet surveys of liberal IR theory either collect disparate views held by ‘‘classical’’ liberal publicists or define liberal theory teleologically, that is, according to its purported optimism concerning the potential for peace, cooperation, and international institutions in world history. Such studies offer an indispensable source of theoretical and normative inspiration. Judged by the more narrowly social scientific criteria adopted here, however, they do not justify reference to a distinct ‘‘liberal’’IR theory. Leading liberal IR theorists freely concede the absence of coherent microfoundational assumptions but conclude therefrom that a liberal IR theory in the social scien tific sense cannot exist. Robert Keohane, an institutionalist sympathetic to liberalism, maintains that ‘‘in contrast to Marxism and Realism, Liberalism is not committed to ambitious and parsimonious structural theory.’’ Michael Doyle, a pioneer in analyzing the ‘‘democratic peace,’’ observes that liberal IR theory, unlike others, lacks ‘‘canonical’’ foundations. Mark Zacher and Richard Matthew, sympathetic liberals, assert that liberalism should be considered an ‘‘approach,’’ not a theory, since ‘‘its propositions cannot be . . . deduced from its assumptions.’’4 Accurate though this may be as a characterization of intellectual history and current theory, it is secondbest social science. I seek to move beyond this unsatisfactory situation by proposing a set of core assumptions on which a general restatement of positive liberal IR theory can be grounded. In the first section of the article I argue that the basic liberal insight about the centrality of state-society relations to world politics can be restated in terms of three positive assumptions, concerning, respectively, the nature of fundamental social actors, the state, and the international system. Drawing on these assumptions, I then elaborate three major variants of liberal theory—each grounded in a distinctive causal mechanism linking social preferences and state behavior. Ideational liberalism stresses the impact on state behavior of conflict and compatibility among collective social values or identities concerning the scope and nature of public goods provision. Commercial liberalism stresses the impact on state behavior of gains and losses to individuals and groups in society from transnational economic interchange. Republican liberalism stresses the impact on state behavior of varying forms of domestic representation and the resulting incentives for social groups to engage in rent seeking.5 Finally, I demonstrate that the identification of coherent theoretical assumptions is not simply an abstract and semantic matter. It has significant methodological, theoretical, and empirical implications. The utility of a paradigmatic restatement should be evaluated on the basis of four criteria, each relevant to the empirical researcher: superior parsimony, coherence, empirical accuracy, and multicausal consistency. First, a theoretical restatement should be general and parsimonious, demonstrating that a limited number of microfoundational assumptions can link a broad range of previously unconnected theories and hypotheses. This restatement does so by showing how liberalism provides a general theory of IR linking apparently unrelated areas of inquiry. The theory outlined here applies equally to liberal and nonliberal states, economic and national security affairs, conflictual and nonconflictual situations, and the behavior both of individual states (‘‘foreign policy’’) and of aggregations of states (‘‘international relations’’). Liberal theory, moreover, explains important phenomena overlooked by alternative theories, including the substantive content of foreign policy, historical change, and the distinctiveness of interstate relations among modern Western states. Second, a theoretical restatement should be rigorous and coherent, offering a clear definition of its own boundaries. This restatement does so by demonstrating that institutionalist theories of regimes—commonly treated as liberal due to ideological and historical connotations—are in fact based on assumptions closer to realism than to liberalism. This helps to explain why IR theorists have found it difficult to distill a set of coherent microfoundational assumptions for liberal theory. Third, a theoretical restatement should demonstrate empirical accuracy vis-a`-vis other theories; it should expose anomalies in existing work, forcing reconsideration of empirical findings and theoretical positions. This restatement of liberal theory meets this criterion by revealing significant methodological biases in empirical evaluations of realist theories of ‘‘relative gains-seeking’’ and constructivist analyses of ideas and IR due to the omission of liberal alternatives. If these biases were corrected, liberal accounts might well supplant many widely accepted realist and institutionalist, as well as constructivist, explanations of particular phenomena in world politics. Fourth, a theoretical restatement should demonstrate multicausal consistency. By specifying the antecedent conditions under which it is valid and the precise causal links to policy outcomes, a theory should specify rigorously how it can be synthesized with other theories into a multicausal explanation consistent with tenets of fundamental social theory. This restatement does so by reversing the nearly universal presumption among contemporary IR theorists that ‘‘systemic’’ theories like realism and institutionalism should be employed as an analytical ‘‘first cut,’’ with theories of ‘‘domestic’’ preference formation brought in only to explain anomalies—a prescription that is both methodologically biased and theoretically incoherent. In its place, this restatement dictates the reverse: Liberal theory is analytically prior to both realism and institutionalism because it defines the conditions under which their assumptions hold. If this proposed reformulation of liberal IR theory meets these four criteria, as I argue it does, there is good reason to accord it a paradigmatic position empirically coequal with and analytically prior to realism and institutionalism, as well as constructivism, in theory and research on world politics. Core Assumptions of Liberal IR Theory Liberal IR theory’s fundamental premise—that the relationship between states and the surrounding domestic and transnational society in which they are embedded critically shapes state behavior by influencing the social purposes underlying state preferences—can be restated in terms of three core assumptions. These assumptions are appropriate foundations of any social theory of IR: they specify the nature of societal actors, the state, and the international system. Assumption 1: The Primacy of Societal Actors The fundamental actors in international politics are individuals and private groups, who are on the average rational and risk-averse and who organize exchange and collective action to promote differentiated interests under constraints imposed by material scarcity, conflicting values, and variations in societal influence. Liberal theory rests on a ‘‘bottom-up’’ view of politics in which the demands of individuals and societal groups are treated as analytically prior to politics. Political action is embedded in domestic and transnational civil society, understood as an aggregation of boundedly rational individuals with differentiated tastes, social commitments, and resource endowments. Socially differentiated individuals define their material and ideational interests independently of politics and then advance those interests through political exchange and collective action.6 Individuals and groups are assumed to act rationally in pursuit of material and ideal welfare.7 For liberals, the definition of the interests of societal actors is theoretically central. Liberal theory rejects the utopian notion that an automatic harmony of interest exists among individuals and groups in society; scarcity and differentiation introduce an inevitable measure of competition. Where social incentives for exchange and collective action are perceived to exist, individuals and groups exploit them: the greater the expected benefits, the stronger the incentive to act. In pursuing these goals, individuals are on the average risk-averse; that is, they strongly defend existing investments but remain more cautious about assuming cost and risk in pursuit of new gains. What is true about people on the average, however, is not necessarily true in every case: some individuals in any given society may be risk-acceptant or irrational. Liberal theory seeks to generalize about the social conditions under which the behavior of self-interested actors converges toward cooperation or conflict. Conflictual societal demands and the willingness to employ coercion in pursuit of them are associated with a number of factors, three of which are relevant to this discussion: divergent fundamental beliefs, conflict over scarce material goods, and inequalities in political power. Deep, irreconcilable differences in beliefs about the provision of public goods, such as borders, culture, fundamental political institutions, and local social practices, promote conflict, whereas complementary beliefs promote harmony and cooperation. Extreme scarcity tends to exacerbate conflict over resources by increasing the willingness of social actors to assume cost and risk to obtain them. Relative abundance, by contrast, lowers the propensity for conflict by providing the opportunity to satisfy wants without inevitable conflict and giving certain individuals and groups more to defend. Finally, where inequalities in societal influence are large, conflict is more likely. Where social power is equitably distributed, the costs and benefits of actions are more likely to be internalized to individuals—for example, through the existence of complex, cross-cutting patterns of mutually beneficial interaction or strong and legitimate domestic political institutions—and the incentive for selective or arbitrary coercion is dampened. By contrast, where power asymmetries permit groups to evade the costs of redistributing goods, incentives arise for exploitative, rent-seeking behavior, even if the result is inefficient for society as a whole. Assumption 2: Representation and State Preferences States (or other political institutions) represent some subset of domestic society, on the basis of whose interests state offıcials define state preferences and act purposively in world politics. In the liberal conception of domestic politics, the state is not an actor but a representative institution constantly subject to capture and recapture, construction and reconstruction by coalitions of social actors. Representative institutions and practices constitute the critical ‘‘transmission belt’’ by which the preferences and social power of individuals and groups are translated into state policy. Individuals turn to the state to achieve goals that private behavior is unable to achieve efficiently.9 Government policy is therefore constrained by the underlying identities, interests, and power of individuals and groups (inside and outside the state apparatus) who constantly pressure the central decision makers to pursue policies consistent with their preferences. This is not to adopt a narrowly pluralist view of domestic politics in which all individuals and groups have equal influence on state policy, nor one in which the structure of state institutions is irrelevant. No government rests on universal or unbiased political representation; every government represents some individuals and groups more fully than others. In an extreme hypothetical case, representation might empower a narrow bureaucratic class or even a single tyrannical individual, such as an ideal-typical Pol Pot or Josef Stalin. Between theoretical extremes of tyranny and democracy, many representative institutions and practices exist, each of which privileges particular demands; hence the nature of state institutions, alongside societal interests themselves, is a key determinant of what states do internationally. Representation, in the liberal view, is not simply a formal attribute of state institutions but includes other stable characteristics of the political process, formal or informal, that privilege particular societal interests. Clientalistic authoritarian regimes may distinguish those with familial, bureaucratic, or economic ties to the governing elite from those without. Even where government institutions are formally fair and open, a relatively inegalitarian distribution of property, risk, information, or organizational capabilities may create social or economic monopolies able to dominate policy. Similarly, the way in which a state recognizes individual rights may shape opportunities for voice.10 Certain domestic representational processes may tend to select as leaders individuals, groups, and bureaucracies socialized with particular attitudes toward information, risk, and loss. Finally, cost-effective exit options, such as emigration, noncompliance, or the transfer of assets to new jurisdictions or uses, insofar as they constrain governments, may be thought of as substitutes for formal representation. Societal pressures transmitted by representative institutions and practices alter ‘‘state preferences.’’ This term designates an ordering among underlying substantive outcomes that may result from international political interaction. Here it is essential— particularly given the inconsistency of common usage—to avoid conceptual confusion by keeping state ‘‘preferences’’ distinct from national ‘‘strategies,’’ ‘‘tactics,’’ and ‘‘policies,’’ that is, the particular transient bargaining positions, negotiating demands, or policy goals that constitute the everyday currency of foreign policy. State preferences, as the concept is employed here, comprise a set of fundamental interests defined across ‘‘states of the world.’’ Preferences are by definition causally independent of the strategies of other actors and, therefore, prior to specific interstate political interactions, including external threats, incentives, manipulation of information, or other tactics. By contrast, strategies and tactics—sometimes also termed ‘‘preferences’’ in game-theoretical analyses—are policy options defined across intermediate political aims, as when governments declare an ‘‘interest’’ in ‘‘maintaining the balance of power,’’ ‘‘containing’’ or ‘‘appeasing’’ an adversary, or exercising ‘‘global leadership.’’12 Liberal theory focuses on the consequences for state behavior of shifts in fundamental preferences, not shifts in the strategic circumstances under which states pursue them. Representative institutions and practices determine not merely which social coalitions are represented in foreign policy, but how they are represented. Two distinctions are critical. First, states may act in either a unitary or ‘‘disaggregated’’ way. In many traditional areas of foreign policy, ‘‘politics stops at the water’s edge,’’ and there is strong coordination among national officials and politicians. In other areas, the state may be ‘‘disaggregated,’’ with different elements—executives, courts, central banks, regulatory bureaucracies, and ruling parties, for example—conducting semiautonomous foreign policies in the service of disparate societal interests.13 Second, domestic decision making may be structured so as to generate state preferences that satisfy a strong rationality condition, such as transitivity or strict expected utility maximization, or so as to satisfy only the weaker rationality criterion of seeking efficient means. Recently, formal theorists have derived specific conditions under which nonunitary state behavior can be analyzed ‘‘as if’’ it were unitary and rational, implying that much superficially ‘‘nonrational’’ or ‘‘nonunitary’’ behavior should actually be understood in terms of shifting state preferences.14 Taken together, assumptions 1 and 2 imply that states do not automatically maximize fixed, homogeneous conceptions of security, sovereignty, or wealth per se, as realists and institutionalists tend to assume. Instead they are, in Waltzian terms, ‘‘functionally differentiated’’; that is, they pursue particular interpretations and combinations of security, welfare, and sovereignty preferred by powerful domestic groups enfranchised by representative institutions and practices.15 As Arnold Wolfers, John Ruggie, and others have observed, the nature and intensity of national support for any state purpose—even apparently fundamental concerns like the defense of political and legal sovereignty, territorial integrity, national security, or economic welfare— varies decisively with the social context.16 It is not uncommon for states knowingly to surrender sovereignty, compromise security, or reduce aggregate economic welfare. In the liberal view, trade-offs among such goals, as well as cross-national differences in their definition, are inevitable, highly varied, and causally consequential.17 Assumption 3: Interdependence and the International System The configuration of interdependent state preferences determines state behavior. For liberals, state behavior reflects varying patterns of state preferences. States require a ‘‘purpose,’’ a perceived underlying stake in the matter at hand, in order to provoke conflict, propose cooperation, or take any other significant foreign policy action. The precise nature of these stakes drives policy. This is not to assert that each state simply pursues its ideal policy, oblivious of others; instead, each state seeks to realize its distinctive preferences under varying constraints imposed by the preferences of other states. Thus liberal theory rejects not just the realist assumption that state preferences must be treated as if naturally conflictual, but equally the institutionalist assumption that they should be treated as if they were partially convergent, compromising a collective action problem.18 To the contrary, liberals causally privilege variation in the configuration of state preferences, while treating configurations of capabilities and information as if they were either fixed constraints or endogenous to state preferences. The critical theoretical link between state preferences, on the one hand, and the behavior of one or more states, on the other, is provided by the concept of policy interdependence. Policy interdependence is defined here as the set of costs and benefits created for foreign societies when dominant social groups in a society seek to realize their preferences, that is, the pattern of transnational externalities resulting from attempts to pursue national distinctive purposes. Liberal theory assumes that the pattern of interdependent state preferences imposes a binding constraint on state behavior. Patterns of interdependence or externalities induced by efforts to realize state preferences can be divided into three broad categories, corresponding to the strategic situation (the pattern of policy externalities) that results.19 Where preferences are naturally compatible or harmonious, that is, where the externalities of unilateral poli cies are optimal for others (or insignificant), there are strong incentives for coexistence with low conflict. Where, by contrast, underlying state preferences are zero-sum or deadlocked, that is, where an attempt by dominant social groups in one country to realize their preferences through state action necessarily imposes costs (negative externalities) on dominant social groups in other countries, governments face a bargaining game with few mutual gains and a high potential for interstate tension and conflict. The decisive precondition for costly attempts at coercion, for example, is neither a particular configuration of power, as realists assert, nor of uncertainty, as institutionalists maintain, but a configuration of preferences conflictual enough to motivate willingness to accept high cost and risk.20 In other words, intense conflict requires that an aggressor or revisionist state advance demands to which other states are unwilling to submit. Revisionist preferences—underlying, socially grounded interests in revising the status quo—are distinct from revisionist ‘‘strategies,’’ that is, a need to alter the status quo to protect enduring interests under new strategic circumstances. Liberals focus on the former, realists (and institutionalists) on the latter. Hence while both theories predict security conflict, they do so under different circumstances. For example, increased military spending in response to an adversary’s arms buildup is a change in strategy with fixed preferences consistent with realism; increased spending initiated by a new ruling elite ideologically committed to territorial aggrandizement is a preference-induced change in strategy consistent with liberalism.21 Where, finally, motives are mixed such that an exchange of policy concessions through coordination or precommitment can improve the welfare of both parties relative to unilateral policy adjustment (i.e., a collective action problem), states have an incentive to negotiate policy coordination. Games like coordination, assurance, prisoner’s dilemma, and suasion have distinctive dynamics, as well as impose precise costs, benefits, and risks on the parties. Within each qualitative category, incentives vary further according to the intensity of preferences. For liberals, the form, substance, and depth of cooperation depends directly on the nature of these patterns of preferences. Hence where ‘‘Pareto-inefficient’’ outcomes are observed—trade protection is a commonly cited example—liberals turn first to countervailing social preferences and unresolved domestic and transnational distributional conflicts, whereas institutionalists and realists, respectively, turn to uncertainty and particular configurations of interstate power. half-century. At the heart of the two leading contemporary IR theories, realism and institutionalism, is the belief that state behavior has ironic consequences.23 Power politics and informational uncertainty constrain states to pursue second- and thirdbest strategies strikingly at variance with their underlying preferences.24 Thus varying state preferences should be treated as if they were irrelevant, secondary, or endogenous. In his classic definition of realism Morgenthau contrasts it to ‘‘two popular fallacies: the concern with motives and the concern with ideological preferences.’’25 Neorealist Waltz’s central objection to previous, ‘‘reductionist’’ theories is that in world politics ‘‘results achieved seldom correspond to the intentions of actors’’; hence ‘‘no valid generalizations can logically be drawn’’ from an examination of intentions.26 Though the interests it assumes are different, Keohane’s institutionalism relies on a similar as if assumption: it ‘‘takes the existence of mutual interests as given and examines the conditions under which they will lead to cooperation.’’27 In short, Powell observes that ‘‘structural theories . . . lack a theory of preferences over outcomes.’’28 What states do is primarily determined by strategic considerations—what they can get or what they know—which in turn reflect their international political environment. In short, variation in means, not ends, matters most.29 Liberal theory reverses this assumption: Variation in ends, not means, matters most. Realists and institutionalists, as well as formal theorists who seek to integrate the two, criticize this core liberal assumption because it appears at first glance to rest on what Waltz terms a ‘‘reductionist’’ rather than a ‘‘systemic’’ understanding of IR. In other words, liberalism appears to be a purely ‘‘domestic’’ or ‘‘unit-level’’ theory that ignores the international environment. In particular, realists are skeptical of this view because it appears at first glance to be grounded in the utopian expectation that every state can do as it pleases. This commonplace criticism is erroneous for two important reasons. First, state preferences may reflect patterns of transnational societal interaction. While state preferences are (by definition) invariant in response to changing interstate political and strategic circumstances, they may well vary in response to a changing transnational social context. In the political economy for foreign economic policy, for example, social demands are derived not simply from ‘‘domestic’’ economic assets and endowments, but from the relative position of those assets and endowments in global markets. Similarly, the position of particular values in a transnational cultural discourse may help define their meaning in each society. In this regard, liberalism does not draw a strict line between domestic and transnational levels of analysis.30 A second and more Waltzian reason why the charge of ‘‘reductionism’’ is erroneous is that according to liberal theory the expected behavior of any single state—the strategies it selects and the systemic constraints to which it adjusts—reflect not simply its own preferences, but the configuration of preferences of all states linked by patterns of significant policy interdependence. National leaders must always think systemically about their position within a structure composed of the preferences of other states. Since the pattern of and interdependence among state preferences, like the distribution of capabilities and the distribution of information and ideas, lies outside the control of any single state, it conforms to Waltz’s own definition of systemic theory, whereby interstate interactions are explained by reference to ‘‘how [states] stand in relation to one another.’’31 Hence the causal preeminence of state preferences does not imply that states always get what they want. One implication of liberalism’s systemic, structural quality is that, contra Waltz, it can explain not only the ‘‘foreign policy’’ goals of individual states but the ‘‘systemic’’outcomes of interstate interactions. That systemic predictions can follow from domestic theories of preferences should be obvious simply by inspecting the literature on the democratic peace.32 In addition, by linking social purpose to the symmetry and relative intensity of state preferences, liberalism offers a distinctive conception of political power in world politics—something traditionally considered unique to realist theory. The liberal conception of power is based on an assumption more consistent with basic theories of bargaining and negotiation than those underlying realism: namely that the willingness of states to expend resources or make concessions is itself primarily a function of preferences, not capabilities. In this view—the foundation of Nash bargaining analysis, which has been extended to IR by Albert Hirshman, Keohane, Joseph Nye, and others—bargaining outcomes reflect the nature and relative intensity of actor preferences.33 The ‘‘win-set,’’ the ‘‘best alternative to negotiated agreement,’’ the pattern of ‘‘asymmetrical interdependence,’’ the relative opportunity cost of forgoing an agreement—all these core terms in negotiation analysis refer to different aspects of the relationship of bargaining outcomes on the preference functions of the actors. The capability-based power to threaten central to realism enters the equation in specific circumstances and only through linkage to threats and side-payments. Even where capability-based threats and promises are employed, preference-based determinants of the tolerance for bearing bargaining costs, including differential temporal discount rates, risk-acceptance, and willingness to accept punishment, remain central.34 The liberal claim that the pattern of interdependence among state preferences is a primary determinant not just of individual foreign policies, but of systemic outcomes, is commonsensical. Nations are rarely prepared to expend their entire economic or defense capabilities, or to mortgage their entire domestic sovereignty, in pursuit of any single foreign policy goal. Few wars are total, few peaces Carthaginian. Treating the willingness of states to expend resources in pursuit of foreign policy goals as a strict function of existing capabilities thus seems unrealistic. On the margin, the binding constraint is instead generally ‘‘resolve’’ or ‘‘determination’’—the willingness of governments to mobilize and expend social resources for foreign policy purposes. Extensive empirical evidence supports this assumption. Even in ‘‘least likely’’ cases, where political independence and territorial integrity are at stake and military means are deployed, relative capabilities do not necessarily determine outcomes. A ‘‘strong preference for the issue at stake can compensate for a deficiency in capabilities,’’ as demonstrated by examples like the Boer War, Hitler’s remilitarization of the Rhineland, Vietnam, Afghanistan, and Chechnya. In each case the relative intensity of state preferences reshaped the outcome to the advantage of the ‘‘weak.’’35 Such examples suggest that the liberal view of power politics, properly understood, generates plausible explanations not just of harmony and cooperation among nations, but of the full range of phenomena central to the study of world politics, from peaceful economic exchange to brutal guerrilla warfare. Variants of Liberal Theory Like their realist and institutionalist counterparts, the three core liberal assumptions introduced earlier are relatively thin or content-free. Taken by themselves, they do not define a single unambiguous model or set of hypotheses, not least because they do not specify precise sources of state preferences. Instead they support three separate variants of liberal theory, termed here ideational, commercial, and republican liberalism. Each rests on a distinctive specification of the central elements of liberal theory: social demands, the causal mechanisms whereby they are transformed into state preferences, and the resulting patterns of national preferences in world politics. Ideational liberalism focuses on the compatibility of social preferences across fundamental collective goods like national unity, legitimate political institutions, and socioeconomic regulation. Commercial liberalism focuses on incentives created by opportunities for transborder economic transactions. Republican liberalism focuses on the nature of domestic representation and the resulting possibilities for rent-seeking behavior. Ideational Liberalism: Identity and Legitimate Social Order Drawing on a liberal tradition dating back to John Stuart Mill, Giuseppe Mazzini, and Woodrow Wilson, ideational liberalism views the configuration of domestic social identities and values as a basic determinant of state preferences and, therefore, of interstate conflict and cooperation. ‘‘Social identity’’ is defined as the set of preferences shared by individuals concerning the proper scope and nature of public goods provision, which in turn specifies the nature of legitimate domestic order by stipulating which social actors belong to the polity and what is owed them.36 Liberals take no distinctive position on the origins of social identities, which may result from historical accretion or be constructed through conscious collective or state action, nor on the question of whether they ultimately reflect ideational or material factors.37 Three essential elements of domestic public order often shaped by social identities are geographical borders, political decision-making processes, and socioeconomic regulation. Each can be thought of as a public or club good; the effectiveness of each typically requires that it be legislated universally across a jurisdiction.38 Recall that for liberals, even the defense of (or, less obvious but no less common, the willing compromise of) territorial integrity, political sovereignty, or national security is not an end in itself, but a means of realizing underlying preferences defined by the demands of societal groups. According to assumption 2, social actors provide support to the government in exchange for institutions that accord with their identity-based preferences; such institutions are thereby ‘‘legitimate.’’ Foreign policy will thus be motivated in part by an effort to realize social views about legitimate borders, political institutions, and modes of socioeconomic regulation. The consequences of identity-based preferences for IR depend, according to assumption 3, on the nature of transnational externalities created by attempts to realize them. Where national conceptions of legitimate borders, political institutions, and socioeconomic equality are compatible, thus generating positive or negligible externalities, harmony is likely. Where national claims can be made more compatible by reciprocal policy adjustment, cooperation is likely.39 Where social identities are incompatible and create significant negative externalities, tension and zero-sum conflict is more likely. Parallel predictions about international politics follow from each of the three essential sources of ideational preferences: national, political, and socioeconomic identity.40 Let us briefly consider each. The first fundamental type of social identity central to the domestic legitimacy of foreign policy comprises the set of fundamental societal preferences concerning the scope of the ‘‘nation,’’ which in turn suggest the legitimate location of national borders and the allocation of citizenship rights. The roots of national identity may reflect a shared set of linguistic, cultural, or religious identifications or a shared set of historical experiences—often interpreted and encouraged by both private groups and state policy. In explaining conflict and cooperation over borders and citizenship, realism stresses the role of relative power, and institutionalism stresses the role of shared legal norms, whereas ideational liberalism stresses the extent to which borders coincide with the national identities of powerful social groups.41 Where borders coincide with underlying patterns of identity, coexistence and even mutual recognition are more likely. Where, however, inconsistencies between borders and underlying patterns of identity exist, greater potential for interstate conflict exists. In such circumstances, some social actors and governments are likely to have an interest in uniting nationals in appropriate jurisdictions, perhaps through armed aggression or secession; other governments may intervene militarily to promote or hinder such efforts. More than twenty years before conflict reemerged in the former Yugoslavia, Myron Weiner termed the resulting disruptive international behavior—a recurrent complex of aggression, exacerbation of nationalist ideologies, offensive alliance formation, and risk acceptance in foreign policy—the ‘‘Macedonian syndrome.’’42 Strong empirical evidence supports the proposition that disjunctures between borders and identities are important determinants of international conflict and cooperation. In early modern Europe, interstate conflict reflected in part the competition between two communal religious identities—each of which, at least until domestic and international norms of tolerance spread, was perceived as a threat to the other.43 Over the last century and a half, from mid-nineteenth-century nationalist uprisings to late-twentieth-century national liberation struggles, the desire for national autonomy constitutes the most common issue over which wars have been fought and great power intervention has taken place; the Balkan conflicts preceding World War I and succeeding the Cold War are only the most notorious examples.44 The post–World War II peace in Western Europe and the reintegration of Germany into Europe were assisted by the reestablishment of borders along ethnic lines in the Saar and AlsaceLorraine, as well as much of Eastern Europe. Even leading realists now concede— though it in no way follows from realist premises—that disputes between ‘‘intermingled or divided nationalities’’ are the most probable catalyst for war in Eastern Europe and the former Soviet Union. A second fundamental type of social identity central to foreign policymaking is the commitment of individuals and groups to particular political institutions. Realism accords theoretical weight to domestic regime type only insofar as it influences the distribution of capabilities, institutionalism only insofar as it contributes to the certainty of coordination and commitment. Ideational liberalism, by contrast, maintains that differences in perceptions of domestic political legitimacy translate into patterns of underlying preferences and thus variation in international conflict and cooperation. Where the realization of legitimate domestic political order in one jurisdiction threatens its realization in others, a situation of negative externalities, conflict is more likely. Where the realization of national conceptions of legitimate decision making reinforce or can be adjusted to reinforce one another, coexistence or cooperation is more likely.46 Plausible examples abound. Thucydides accords an important role to conflict between oligarchs and democrats in alliance formation during the Peloponnesian War. In the seventeenth and eighteenth centuries, absolutist kings fought to establish dynastic claims and religious rule; in the nineteenth century, they cooperated to preserve monarchical rule against societal pressures for reform.47 The twentieth century has witnessed a struggle between governments backing fascist, communist, and liberal ideologies, as well as more recently a resurgence of religious claims and the emergence of a group of developed countries that share democratic norms of legitimate dispute resolution—a plausible explanation for the ‘‘democratic peace’’phenomenon.48 A more complex pattern, consistent with the preceding assumptions, may emerge when individual domestic actors—most often national executives—exploit the legitimacy of particular international policies as a ‘‘two-level’’ instrument to increase their influence over the domestic polity. This is a constant theme in modern world politics, from Bismarck’s manipulation of domestic coalitions to the current use of monetary integration by today’s European leaders to ‘‘strengthen the state’’ at home.49 A third fundamental type of social identity central to foreign policy is the nature of legitimate socioeconomic regulation and redistribution. Modern liberal theory (as opposed to the laissez faire libertarianism sometimes invoked by critics as quintessentially ‘‘liberal’’) has long recognized that societal preferences concerning the nature and level of regulation impose legitimate limits on markets.50 In a Polanyian vein, Ruggie recently reminds us that domestic and international markets are embedded in local social compromises concerning the provision of regulatory public goods.51 Such compromises underlie varying national regulations on immigration, social welfare, taxation, religious freedom, families, health and safety, environmental and consumer protection, cultural promotion, and many other public goods increasingly discussed in international economic negotiations. In the liberal view, state preferences concerning legitimate socioeconomic practices shape interstate behavior when their realization imposes significant transborder externalities. Evidence from the European Community (EC) suggests that substantial prior convergence of underlying values is a necessary prerequisite for cooperation in regulatory issue areas like environmental and consumer protection, many tax and social policies, immigration, and foreign policy, as well as for significant surrenders of sovereign decision making to supranational courts and bureaucracies. Regulatory pluralism limits international cooperation, in particular economic liberalization. Courts, executives, and parliaments mutually recognize ‘‘legitimate differences’’ of policy in foreign jurisdictions.52 Concerns about the proper balance between policy coordination and legitimate domestic regulation are giving rise to even more complex forms of cooperation. Hence regulatory issues play an increasingly important role in international economic negotiations such as the 1992 initiative of the EC, the Uruguay Round of GATT, NAFTA, and the U.S.–Japan Structural Impediments Initiative.53 Commercial Liberalism: Economic Assets and Cross-Border Transactions Commercial liberalism explains the individual and collective behavior of states based on the patterns of market incentives facing domestic and transnational economic actors. At its simplest, the commercial liberal argument is broadly functionalist: Changes in the structure of the domestic and global economy alter the costs and benefits of transnational economic exchange, creating pressure on domestic governments to facilitate or block such exchanges through appropriate foreign economic and security policies. It is tempting, particularly for critics, to associate commercial liberal theory with ideological support for free trade. Yet as theory rather than ideology, commercial liberalism does not predict that economic incentives automatically generate universal free trade and peace—a utopian position critics who treat liberalism as an ideology often wrongly attribute to it—but instead stresses the interaction between aggregate incentives for certain policies and obstacles posed by domestic and transnational distributional conflict.54 The greater the economic benefits for powerful private actors, the greater their incentive, other things being equal, to press governments to facilitate such transactions; the more costly the adjustment imposed by economic interchange, the more opposition is likely to arise. Rather than assuming that market structure always creates incentives for cooperation among social actors as well as states, or focusing exclusively on those issue areas where it does, as do some liberal ideologies, liberal IR theory focuses on market structure as a variable creating incentives for both openness and closure. Accordingly, many commercial liberal analyses start with aggregate welfare gains from trade resulting from specialization and functional differentiation, then seek to explain divergences from foreign economic and security policies that would maximize those gains. To explain the rejection of aggregate gains, commercial liberals from Adam Smith to contemporary ‘‘endogenous’’ tariff theorists look to domestic and international distributional conflicts. The resulting commercial liberal explanation of relative-gains seeking in foreign economic policy is quite distinct from that of realism, which emphasizes security externalities and relative (hegemonic) power, or that of institutionalism, which stresses informational and institutional constraints on interstate collective action.55 One source of pressure for protection is domestic distributional conflict, which arises when the costs and benefits of national policies are not internalized to the same actors, thus encouraging rent-seeking efforts to seek personal benefit at the expense of aggregate welfare. In this view, uncompetitive, monopolistic, or undiversified sectors or factors lose the most from liberalization and have an incentive to oppose it, inducing a systematic divergence from laissez faire policies. Smith himself reminds us that ‘‘the contrivers of [mercantilism are] the producers, whose interest has been so carefully attended to . . . our merchants and manufacturers’’—a view echoed by many liberals since.56 Recent research supports the view that protectionist pressure from rent-seeking groups is most intense precisely where distributional concerns of concentrated groups are strongest, for example, when industries are uncompetitive or irreversible investments (asset specificity) impose high adjustment costs on concentrated interests. Free trade is more likely where strong competitiveness, extensive intra-industry trade, or trade in intermediate goods, large foreign investments, and low asset specificity internalize the net benefits of free trade to powerful actors, thus reducing the influence of net losers from liberalization.57 The distributional consequences of global market imperfections create a second sort of disjuncture between the aggregate benefits of economic interdependence and national policies. Modern trade theory identifies incentives for strategic behavior where increasing returns to scale, high fixed costs, surplus capacity, or highly concentrated sources of supply render international markets imperfectly competitive. Firms hoping to create (or break into) a global oligopoly or monopoly, for example, may have an incentive to engage in predatory dumping abroad while seeking domestic protection and subsidization at home, even though this imposes costs on domestic consumers and foreign producers. Such policies can create substantial international conflict, since government intervention to assist firms can improve welfare for society as a whole, though usually not for all societies involved. Commercial liberalism has important implications for security affairs as well. Trade is generally a less costly means of accumulating wealth than war, sanctions, or other coercive means, not least due to the minimization of collateral damage. Yet governments sometimes have an incentive to employ coercive means to create and control international markets. To explain this variation, domestic distributional issues and the structure of global markets are again critical. Commercial liberals argue that the more diversified and complex the existing transnational commercial ties and production structures, the less cost-effective coercion is likely to be.59 Cost-effective coercion was most profitable in an era where the main sources of economic profit, such as farmland, slave labor, raw materials, or formal monopoly, could be easily controlled in conquered or colonial economies. Yet economic development tends to increase the material stake of social actors in existing investments, thereby reducing their willingness to assume the cost and risk of coercion through war or sanctions.60 As production becomes more specialized and efficient and trading networks more diverse and complex, political extraction (for example, war and embargoes) become more disruptive, and profitable monopolies over commercial opportunities become more difficult to establish. Both cross-cultural anthropological evidence and modern cross-national evidence link warfare to the existence of monopolizable resources; over the past century, it has remained the major determinant of boundary disputes.61 Yet the advent of modern industrial networks, particularly those based on postindustrial informational exchange, has increased the opportunity costs of coercive tactics ranging from military aggression to coercive nationalization.62 Republican Liberalism: Representation and Rent Seeking While ideational and commercial liberal theory, respectively, stress demands resulting from particular patterns of underlying societal identities and economic interests, republican liberal theory emphasizes the ways in which domestic institutions and practices aggregate those demands, transforming them into state policy. The key variable in republican liberalism is the mode of domestic political representation, which determines whose social preferences are institutionally privileged. When political representation is biased in favor of particularistic groups, they tend to ‘‘capture’’ government institutions and employ them for their ends alone, systematically passing on the costs and risks to others. The precise policy of governments depends on which domestic groups are represented. The simplest resulting prediction is that policy is biased in favor of the governing coalition or powerful domestic groups. A more sophisticated extension of this reasoning focuses on rent seeking. When particularistic groups are able to formulate policy without necessarily providing offsetting gains for society as a whole, the result is likely to be inefficient, suboptimal policies from the aggregate perspective—one form of which may be costly international conflict.63 While many liberal arguments are concerned with the seizure of state institutions by administrators (rulers, armies, and bureaucracies), similar arguments apply to privileged societal groups that ‘‘capture’’ the state, according to assumption 2, or simply act independently of it. If, following assumption 1, most individuals and groups in society, while acquisitive, tend also to be risk-averse (at least where they have something to lose), the more unbiased the range of domestic groups represented, the less likely they will support policies that impose high net costs or risks on a broad range of social actors. Thus aggressive behavior—the voluntary recourse to costly or risky foreign policy—is most likely in undemocratic or inegalitarian polities where privileged individuals can easily pass costs on to others.64 This does not, of course, imply the existence of a one-to-one correspondence between the breadth of domestic representation and international political or economic cooperation, for two reasons. First, in specific cases, elite preferences may be more convergent than popular ones. If commercial or ideational preferences are conflictual, for example where hypernationalist or mercantilist preferences prevail, a broadening of representation may have the opposite effect—a point to which I will return. Elites, such as those leaders that constructed the Concert of Europe or similar arrangements among African leaders today, have been attributed to their convergent interests in maintaining themselves in office. Second, the extent of bias in representation, not democratic participation per se, is the theoretically critical point. Direct representation may overrepresent concentrated, organized, short-term, or otherwise arbitrarily salient interests. Predictable conditions exist under which governing elites may have an incentive to represent long-term social preferences more unbiasedly than does broad opinion.65 Despite these potential complexities and caveats, republican liberalism nonetheless generates parsimonious predictions where conflictual policies impose extremely high costs and risks on the majority of individuals in domestic society. With respect to extreme but historically common policies like war, famine, and radical autarky, fair representation tends to inhibit international conflict. In this way, republican liberal theory has helped to explain phenomena as diverse as the ‘‘democratic peace,’’ modern anti-imperialism, and international trade and monetary cooperation. Given the prima facie plausibility of the assumption that major war imposes net costs on society as a whole, it is not surprising that the prominent republican liberal argument concerns the ‘‘democratic peace,’’ which one scholar has termed ‘‘as close as anything we have to a law in international relations’’—one that applies to tribal societies as well as to modern states.66 Liberal democratic institutions tend not to provoke such wars because influence is placed in the hands of those who must expend blood and treasure and the leaders they choose. Often overlooked is the theoretical corollary of ‘‘democratic peace’’ theory: a republican liberal theory of war that stresses abnormally risk-acceptant leaders and rent-seeking coalitions. Substantial evidence shows that the aggressors who have provoked modern great power wars tend either to be risk-acceptant individuals in the extreme or individuals well able to insulate themselves from the costs of war or both. Most leaders initiating twentieth-century great power wars lost them; Adolf Hitler and Saddam Hussein, for example, initiated conflicts against coalitions far more powerful than their own.68 In the same vein, Jack Snyder has recently deepened Hobson’s classic rent-seeking analysis of imperialism—whereby the military, uncompetitive foreign investors and traders, jingoistic political elites, and others who benefit from imperialism are particularly well-placed to influence policy—by linking unrepresentative and extreme outcomes to logrolling coalitions. Consistent with this analysis, the highly unrepresentative consequences of partial democratization, combined with the disruption of rapid industrialization and incomplete political socialization, suggest that democratizing states, if subject to these influences, may be particularly war prone. Such findings may challenge some variants of liberal ideology but are consistent with liberal theory.69 The link between great-power military aggression and small-group interests in nonrepresentative states implies neither unceasing belligerence by autocratic regimes nor unquestioning pacifism by democratic ones. Enlightened despotism or democratic aggression remains possible. The more precise liberal prediction is thus that despotic power, bounded by neither law nor representative institutions, tends to be wielded in a more arbitrary manner by a wider range of individuals, leading both to a wider range of expected outcomes and a more conflictual average. Nonetheless, liberal theory predicts that democratic states may provoke preventive wars in response to direct or indirect threats, against very weak states with no great power allies, or in peripheral areas where the legal and political preconditions for trade and other forms of profitable transnational relations are not yet in place.70 Scholars also often overlook precise analogs to the ‘‘democratic peace’’ in matters of political economy. The liberal explanation for the persistence of illiberal commercial policies, such as protection, monetary instability, and sectoral subsidization, where such policies manifestly undermine the general welfare of the population, is pressure from powerful domestic groups.71 Thus in the liberal view the creation and maintenance of regimes assuring free trade and monetary stability result not primarily from common threats to national security or appropriate international institutions, but from the ability of states to overcome domestic distributional conflicts in a way supportive of international cooperation. This may ultimately reflect the economic benefits of doing so, as commercial liberal theory suggests, but it can also be decisively helped or hindered by biases in representative institutions. Where such biases favor sheltered groups, and substantial misrepresentation of this type is seen as endemic to most contemporary representative institutions, rent-seeking groups are likely to gain protection through tariffs, subsidies, favorable regulation, or competitive devaluation. Where policymakers are insulated from such pressures, which may involve less democratic but more representative institutions, or where free trade interests dominate policy, open policies are more viable.72 Broader Implications of Liberal Theory Do labels matter? I have explored three variants of liberal theory that share a set of assumptions. What is gained by subsuming them under a single rubric, as proposed here? To demonstrate its utility for empirical research and theoretical inquiry, a paradigmatic restatement such as this must meet four criteria. First, its assumptions should highlight unexplored conceptual connections among previously unrelated liberal hypotheses. Second, it should clearly define its own conceptual boundaries in a manner conforming to fundamental social theory, in this case clearly distinguishing liberal hypotheses from ideologically or historically related hypotheses based on different social scientific assumptions. Third, it should reveal anomalies in previous theories and methodological weaknesses in previous testing, creating new presumptions about the proper theories and methods that structure empirical research. Fourth, it should define how the theory in question can be combined rigorously rather than randomly with other theories to form coherent multicausal explanations. Liberalism as a General Theory: Parsimony and Coherence One advantage of this restatement is that it suggests a theory of world politics that parsimoniously connects a wide range of distinctive and previously unrelated hypotheses concerning areas unexplained by existing theories. These hypotheses are not limited to cooperation among liberal states, but subsume liberal and nonliberal polities, conflictual and cooperative situations, security and political economy issues, and both individual foreign policy and aggregate behavior. Its key causal mechanisms can be generalized to many issue areas. Thus liberal theory challenges the conventional presumption that realism is the most encompassing and parsimonious of major IR theories. Although not all liberal theories are easy to specify, hypotheses about endogenous tariff setting, the democratic peace, and nationalist conflict suggest that liberalism generates many empirical arguments as powerful, parsimonious, and ‘‘efficient’’ as those of realism.73 Not only does liberal theory apply across a wide domain of circumstances, but its three variants—ideational, commercial, and republican liberalism—are stronger taken together than separately. Not only do they share assumptions and causal mechanisms, but their empirical implications aggregate in interesting ways. It is widely accepted, for example, that economic development has a strong influence on the viability of democratic governance, with its pacific implications; liberal democratic governments tend in turn to support commerce, which promotes economic development.74 Karl Deutsch, Ernst Haas, and Nye, among many others, have explored how economic interaction can lead to transnational communication and the dissemination of scientific information, which may in turn promote secularizing cognitive and ideological change.75 Liberal theories can be analytically reinforcing even where they do not make parallel predictions. Anomalies within one variant of liberal theory may be resolved by considering other variants. Positive movement along one liberal dimension— patterns of national identity, democratic participation, or transnational economic transactions—may condone or exacerbate the negative distortions along another liberal dimension.76 Norman Angell, whose commercial liberal claims are often parodied by secondhand critics, maintained that his well-known ‘‘unprofitability of war’’ thesis in no way implies ‘‘the impossibility of war,’’ a doctrine he dismissed for republican liberal reasons as a ‘‘ridiculous myth.’’77 Where representative bias permits rentseeking groups to control policy, aggregate incentives for welfare-improving trade are likely to have less effect. Indeed, recent studies reveal that the correlation between economic interdependence and peace holds only (or most strongly) among liberal states.78 Conversely, where democratization heightens socioeconomic inequality, nationalist cleavages, uneven patterns of gains, and losses due to interdependence or extreme heterogeneity of interests—as may have occurred in the former Yugoslavia—it may exacerbate international economic and political conflict.79 Such interaction effects among liberal factors offer a promising area for more detailed analysis. Liberal theory also illuminates at least three major phenomena for which realism and institutionalism offer few, if any, predictions—another indicator of greater parsimony. First, liberal theory provides a plausible theoretical explanation for variation in the substantive content of foreign policy. Neither realism nor institutionalism explains the changing substantive goals and purposes over which states conflict and cooperate; both focus instead on formal causes, such as relative power or issue density, and formal consequences, such as conflict and cooperation per se.80 By contrast, liberal theory provides a plausible explanation not just for conflict and cooperation, but for the substantive content of foreign policy. Major elements of international order emphasized, but not explained, in recent criticisms of realism and institutionalism include the difference between Anglo-American, Nazi, and Soviet plans for the post–World War II world; U.S. concern about a few North Korean, Iraqi, or Chinese nuclear weapons, rather than the greater arsenals held by Great Britain, Israel, and France; the substantial differences between the compromise of ‘‘embedded liberalism’’ underlying Bretton Woods and arrangements under the Gold Standard; divergences between economic cooperation under the EC and the Council for Mutual Economic Assistance; and the greater protectionism of the Organization for Economic Cooperation and Development’s agricultural policy, as compared to its industrial trade policy.81 Liberal IR theory offers plausible, parsimonious hypotheses to explain each of these phenomena.82 Second, liberal theory offers a plausible explanation for historical change in the international system. The static quality of both realist and institutionalist theory— their lack of an explanation for fundamental long-term change in the nature of international politics—is a recognized weakness. In particular, global economic development over the past five hundred years has been closely related to greater per capita wealth, democratization, education systems that reinforce new collective identities, and greater incentives for transborder economic transactions.83 Realist theory accords these changes no theoretical importance. Theorists like Waltz, Gilpin, and Paul Kennedy limit realism to the analysis of unchanging patterns of state behavior or the cyclical rise and decline of great powers. Liberal theory, by contrast, forges a direct causal link between economic, political, and social change and state behavior in world politics. Hence, over the modern period the principles of international order have been decreasingly linked to dynastic legitimacy and increasingly tied to factors directly drawn from the three variants of liberal theory: national self-determination and social citizenship, the increasing complexity of economic integration, and liberal democratic governance.84 Third, liberal theory offers a plausible explanation for the distinctiveness of modern international politics. Among advanced industrial democracies, a stable form of interstate politics has emerged, grounded in reliable expectations of peaceful change, domestic rule of law, stable international institutions, and intensive societal interaction. This is the condition Deutsch terms a ‘‘pluralistic security community’’ and Keohane and Nye term ‘‘complex interdependence.’’85 Whereas realists (and constructivists) offer no general explanation for the emergence of this distinctive mode of international politics, liberal theory argues that the emergence of a large and expanding bloc of pacific, interdependent, normatively satisfied states has been a precondition for such politics. Consider, for example, the current state of Europe. Unlike realism, liberal theory explains the utter lack of com petitive alliance formation among the leading democratic powers today. For example, the absence of serious conflict among Western powers over Yugoslavia—the ‘‘World War I scenario’’—reflects in large part a shared perception that the geopolitical stakes among democratic governments are low. Similarly, liberalism makes more sense of the sudden reversal of East–West relations, a shift made possible by the widespread view among Russian officials (so interview data reveal) that Germany is ethnically satisfied, politically democratic, and commercially inclined.86 The Conceptual Limits of Liberalism: Why Functional Regime Theory Is Not Liberal A second advantage of the reformulation is to clarify the fundamental divergence between theories of state preferences and modern theories of international regimes. This divergence helps explain why liberals have failed to identify a coherent set of social scientific assumptions underlying existing ‘‘liberal’’ IR theory. Those who choose to define liberal theory in terms of its intellectual history naturally conflate the belief in institutions with a concern about the societal sources of state preferences. Liberalism as an ideology and partisan movement has often been associated in the popular mind with advocacy of international law and organization, despite the views of many leading liberals.87 Others link these two arguments ideologically: Both seem to suggest an optimistic, ameliorative trend in modern world politics. Whatever the reason, contemporary ‘‘functional’’ theories of international regimes are often referred to as forms of ‘‘neoliberal institutionalism,’’ though it is fair to note that Keohane, originator of ‘‘functional regime theory,’’ has abandoned the term. Daniel Deudney and G. John Ikenberry’s attempted restatement of liberalism goes furthest, asserting flatly that ‘‘the peace of the West does not derive simply or mainly from the fact that its polities are all democracies,’’ but from international institutions.88 Imre Lakatos reminds us, however, that the coherence of scientific theories is measured not by their conclusions, but by the consistency of their ‘‘hard-core’’ assumptions. By this standard, neoliberal institutionalist theory has relatively little in common with liberal theory as elaborated here, because most of the analytic assumptions and basic causal variables employed by institutionalist theory are more realist than liberal. Like realism, institutionalism takes state preferences as fixed or exogenous, seeks to explain state policy as a function of variation in the geopolitical environment—albeit for institutionalists information and institutions and for realists material capabilities—and focuses on the ways in which anarchy leads to suboptimal outcomes. Liberalism, by contrast, shares none of these assumptions. It permits state preferences to vary while holding power and information constant, explains policy as a function of the societal context, and focuses on how domestic conflict, not international anarchy, imposes suboptimal outcomes. Therefore, contemporary regime theory ought more properly to be termed ‘‘modified structural realism’’ (as it was initially) or ‘‘institutionalism’’ (as some now prefer), rather than ‘‘neoliberal institutionalism.’’89 This division permits us to speak of a coherent set of social scientific assumptions underlying both. Rather than treated as parts of the same theoretical tradition, the two theories should be tested against one another or carefully crafted into explicitly multicausal explanations—options explored in more detail in the next two sections. This is not to imply, however, that liberal theory is of no utility in analyzing international regimes. To the contrary, it contributes to such analysis in at least two distinctive ways. First, liberal theory explains when and why the configuration of state preferences assumed by institutionalists—a mixed-motive collective action problem that can be overcome by the centralized manipulation of information through common rules—is likely to emerge. Since, moreover, particular institutional structures solve specific collective action problems, the configuration of preferences permits us to predict detailed characteristics of international regimes.90 Second, liberal theory deepens the institutionalist account of regime stability. Realists argue that regime stability and expansion are functions of enduring hegemonic power; institutionalists maintain that the high interstate transaction costs of regime creation or renegotiation explain regime stability, even if patterns of functional benefits would recommend renegotiation. Liberal theory suggests an alternative hypothesis: namely that international regimes are stable when societal individuals and groups adjust so as to make domestic policy reversal (or even stagnation) costly—as neofunctionalist regional integration theorists have long argued. This account is consistent with the transaction cost foundations of institutionalist reasoning but grounded in societal ‘‘lock in’’ effects and the resulting stability of state preferences, not the costs of interstate bargaining, monitoring, and sanctioning. Such ‘‘social embeddedness’’ may take the form of fixed investments by private firms, ideological commitments by political parties concerned about their reputation, costly institutional adaptation by domestic bureaucracies, or government investment in military defense.91 The liberal view of regimes as ‘‘socially embedded’’ can be extended to suggest endogenous causes of regime change over time. International regimes that induce greater societal demands for cooperation are more likely to deepen or expand over time, whereas those that do not are likely to be fragile. One example is the liberal account of international law, which suggests that international rules and norms are most effectively implemented as ‘‘horizontal commitments’’ enforced by national courts and parliaments, not ‘‘vertical commitments’’ enforced by supranational ac tors, and that such horizontal commitments can generate self-sustaining momentum over time by empowering particular domestic groups.92 Methodological Implications of Liberal Theory: The Danger of Omission A third potential advantage of reformulating a social scientific theory is to increase its salience, thus compelling empirical studies to give serious consideration to hypotheses drawn from it and discouraging omitted variable bias.93 Powerful liberal hypotheses exist to account for many major phenomena in world politics, yet surprisingly few studies directly confront realist and institutionalist (or constructivist) hypotheses with their liberal counterparts. Instead, empirical studies tend to treat realism (or occasionally institutionalism or ‘‘rationalism’’) as an exclusive baseline. The result is not just incomplete analysis. It is omitted variable bias that inflates the empirical support for new theoretical propositions due to the exclusion of (correlated) liberal ones. Two recent examples—one realist, one constructivist—demonstrate the considerable empirical significance of this bias.94 The first example comes from perhaps the most prominent debate in recent realist theory—namely, that surrounding Joseph Grieco’s ‘‘relative-gains’’ critique of institutionalism. Based on an analysis of the implementation of nontariff barrier (NTB) provisions negotiated in the Tokyo Round of GATT, Grieco seeks to demonstrate that security concerns about relative gains, not fears of future cheating, motivate noncooperation, even in foreign economic policy.95 Yet in focusing on institutionalism, Grieco ignores liberal explanations for noncooperation based on domestic institutions, ideas, and distributional conflict among domestic economic interests.96 Subsequent interventions in the relative-gains debate by formal theorists, which have done much to clarify the strategic conditions under which particular strategies are likely to emerge, exacerbate this neglect by seeking to make a virtue of omission. Emerson Niou and Peter Ordeshook see preferences as ‘‘tangential to a theory of international systems. . . . We can conduct this discussion without references to goals.’’97 As a result, the relative-gains debate has remained extraordinarily narrow. Both Grieco and those he criticizes treat national interests as fixed and seek only to determine which external political constraint—capabilities or information—constitutes the primary determinant of state behavior. This neglect of liberal hypotheses would be of only abstract significance had it not led all participants in the relative-gains debate to overlook the explanation of noncooperation that most analysts of international trade policy, not to mention nearly all who actually conduct negotiations of this kind, consider decisive—namely, pressure from particularistic domestic groups with intense distributional concerns. Liberal preference-based explanations dominate the specialized economic, political science, and policy literature on trade, particularly in precisely those three areas where Grieco finds ‘‘relative gains’’: government procurement, industrial standard-setting, and administrative protection. Yet Grieco codes these three critical cases of interstate bargaining failure as confirming his account, without considering alternative motivations nor, except in one minor case, providing any direct evidence of national security concerns. Studies in other areas that do test liberal theories against realist alternatives reveal that pressure from economic special interests tends to dominate security concerns, even in ‘‘least likely’’ cases like military procurement.98 Since there is good reason to suspect omitted variable bias, our theoretical understanding of relative gains seeking would have been far more reliable (but also surely far less realist!) if the initial research design had included liberal hypotheses. A second example of omitted variable bias is drawn from recent efforts to develop a constructivist approach to IR. Constructivism, though not yet formulated as a theory, is a welcome effort to broaden IR debates by focusing on ideational socialization. Yet, like realist claims about relative gains, constructivist arguments are generally employed so as to prevent confrontation with preexisting liberal theory. The theoretical introduction to a recent collection of constructivist essays, The Culture of National Security, for example, identifies ‘‘two major analytical perspectives on IR’’: Waltzian neorealism and the ‘‘neoliberal’’ regime theory of Keohane and Robert Axelrod. With only a few exceptions, recent constructivist work employs this dichotomy, therefore neglecting liberal theories focusing on the relationship between conflict and democratic government, economic interdependence, and domestic coalitions—theories recognized as among the most powerful in contemporary security studies.99 This is unfortunate. There are good a priori reasons to suspect that omitted variable bias is inflating the empirical support for any constructivist claim that remains untested against a liberal hypothesis. Not only do both liberal and constructivist arguments focus on variation in state preferences, but we know that the receptiveness to particular ideas is closely correlated with authoritative domestic institutions, patterns of interdependence, and existing patterns of cultural identity. ‘‘Systemic’’ constructivist claims—the view that national ideas and identities result from the socializing ‘‘feedback’’ effects of previous international political interactions—are particularly vulnerable to such bias, because domestic preferences are the critical causal link between systemic socialization and state policy. Without a theory of domestic preference formation, how can a constructivist specify which feedback processes of socialization matter, let alone when and how they matter? Sociologists have long since concluded that ‘‘new institutionalist’’ analyses of this kind are crippled unless con joined with a reliable theory of actors and agency.100 In short, in order to theorize rigorously about systemic social construction, we first require a liberal theory. Existing liberal hypotheses, moreover, offer a general account of variation in socialization—a theory of when the transnational transmission of ideas matters— something for which ‘‘systemic’’ constructivists as of yet lack an explanation. Socialization effects, liberals predict, will reflect the extent of convergence or divergence among preexisting domestic institutions and ideas. For example, socialization toward convergent norms stems from convergent domestic institutions and ideas. Liberal institutions and norms may be particularly conducive to the promotion of peace and cooperation, but the argument implies that the convergence of certain other sorts of nonliberal values, such as monarchy in the Concert of Europe or ‘‘Asian values’’inASEAN, may also have significant, if generally less striking, effects on world politics. Consider, for example, the current revival of interest among constructivists in Deutsch’s analysis of how transnational communication creates ‘‘pluralistic security communities’’ (PSGs) in which groups of states ‘‘cease to contemplate’’ military conflict. PSGs are said to demonstrate the importance of the socializing power of transnational ideas, the importance of ‘‘common . . . we-feeling’’rather than ‘‘convergent’’ interests.101 Yet Deutsch himself viewed liberal factors—an autonomous civil society with individual mobility, the rule of law, and competitive politics—as preconditions for transformative effects of high levels of international transactions and communication. Is it just coincidence that of the of the twelve successful post-1750 PSGs identified by Deutsch, ten or eleven were composed of liberal or nearly liberal states?102 This analysis poses two general challenges to constructivism. First, it suggests that liberal variables are more fundamental than constructivist ones, because they define the conditions under which high rates of communication and transaction alter state behavior. Second, it raises the possibility that domestic liberal factors may explain both peace and transactions, rendering the correlation between international communication and peace not just secondary, but spurious.103 Without directly confronting liberal theory, we cannot dismiss either possibility. Surely our understanding of world politics would be better served by more rigorous empirical confrontation between constructivism and liberalism. Better yet would be a sophisticated synthesis, as found in the ‘‘liberal constructivist’’ research program advocated by Thomas RisseKappen. This approach—a ‘‘constructivist interpretation of liberal theory’’—backs away from the notion that values result from interstate socialization and argues instead in a liberal vein that ideas and communication matter when they are most congruent with existing domestic values and institutions.104 These examples demonstrate why it is essential to treat liberalism as a constant theoretical baseline against which either realist or constructivist hypotheses are tested—that is, as a fundamental paradigmatic alternative in IR. Failure to control for underlying variation in state preferences has confounded recent attempts, quantitative and qualitative, to test monocausal realist theory in many other areas. These include the study of deterrence, hegemonic influence, alliance formation, international negotiation, international monetary cooperation, multilateral cooperation, economic sanctions, and European integration.105 Similar criticisms could be directed at functional regime theory; baseline predictions about the precise form and the subsequent consequences of international regimes could be derived from liberal theory.106 Failure to do so poses a clear threat to valid empirical inference. We already see realists and constructivists ‘‘borrowing’’ liberal hypotheses, even where it undermines the ‘‘hard core’’ of theories. Realist Stephen Walt suggests that ‘‘intentions’’ should be included alongside power, proximity, and offense dominance in their specification of ‘‘threat.’’ Constructivist Alexander Wendt is in retreat from his ‘‘holistic’’ or ‘‘top-down’’ claim that state identities are ideationally constructed by interaction of states (not societies) within the international system. Now he accepts a view heavily dependent on ‘‘unit-level changes in the structure of statesociety relations,’’embedded in domestic (as well as international) institutions, which leads him to embrace phenomena for which well-established liberal theories have long provided widely accepted explanations, for example, the democratic peace, U.S. fear of nuclear weapons in the hands of rogue states but not democratic allies, and the ‘‘distinctiveness of the West.’’107 The prognosis: Unbiased tests would very likely supplant numerous accepted realist, institutionalist, and constructivist explanations of state behavior with liberal accounts. In the long run, comparative theory testing should be aimed at a clearer definition of the empirical domain within which each major theory performs best. Detailed predictions concerning these empirical domains go beyond the scope of this essay, since they require issue-specific analysis of at least three theories. We can nonetheless conclude that oft-cited generalizations about the scope of realism and liberalism need to be revised fundamentally. Liberal theory remains important, even primary, even in what are currently considered ‘‘least likely’’ cases, for example, where there exist direct threats to national security, high levels of interstate conflict, and large numbers of nonliberal states. The restatement proposed here aims to facilitate empirical research that would move us beyond these simplistic assertions about the limited explanatory domain of liberal theory. Liberalism and Theory Synthesis: The Priority of Preferences The previous section demonstrates that, as a monocausal theory, liberalism offers a theoretically coherent and empirically promising alternative to realism and institu tionalism, as well as to constructivism. Yet it is not always appropriate to employ a monocausal theory. If foreign policymaking is a process of constrained choice by purposive states, a view shared by realist, institutionalist, and liberal theory, there may well be cases in which a combination of preferences and constraints shapes state behavior. In such cases, a multicausal synthesis, one that treats these theories not as substitutes but as complements, is required. If so, what synthetic model should properly be employed? Fundamental theories should be formulated so as to provide rigorous means of defining their proper relationship to other theories. A fourth important advantage of this theoretical restatement is that it offers a clearer and more internally consistent model for multicausal theory synthesis in IR than currently exists. It does so, moreover, by reversing the nearly universal presumption among IR theorists that ‘‘liberalism makes sense as an explanatory theory within the constraints pointed out by . . . Realism.’’108 Waltz, Keohane, and many others recommend that we synthesize theories by employing realism first (with preferences assumed to be invariant) and then introducing competing theories of domestic politics, state-society relations, and preference change as needed to explain residual variance.109 Yet this conventional procedure lacks any coherent methodological or theoretical justification. Methodologically, the procedure overtly introduces omitted variable bias by arbitrarily privileging realist explanations of any phenomena that might be explained by both realist and liberal theories, without ever testing the latter explanation. Theoretically, the procedure is grounded in an incoherent underlying model. The assumption of state rationality, central to realism, institutionalism, and most variants of liberalism, ought to imply precisely the opposite: Once we accept that both preferences and constraints are causally important, liberal theory enjoys analytical priority in any synthesis. To see why this is so and what it implies, one need only note that the assumption of rationality or purposive behavior central to realism (like the ‘‘bounded rationality’’ claims of institutionalism) implies action on the basis of a prior, specific, and consistent set of preferences. Unless we know what these preferences are (that is, unless we know the extent to which states value the underlying stakes), we cannot assess realist or institutionalist claims linking variation in the particular means available to states (whether coercive capabilities or institutions) on interstate conflict or cooperation. Preferences determine the nature and intensity of the game that states are playing and thus are a primary determinant of which systemic theory is appropriate and how it should be specified. Variation in state preferences often influences the way in which states make calculations about their strategic environment, whereas the converse— that the strategic situation leads to variation in state preferences—is inconsistent with the rationality assumption shared by all three theories.110 In short, liberal theory explains when and why the assumptions about state preferences underlying realism or institutionalism hold, whereas the reverse is not the case. In situations where these assumptions do not hold, realism and institutionalism (as well as some variants of constructivism) are not just of limited importance, they are theoretically inappropriate and thus empirically irrelevant. It follows that in any multicausal synthesis with realist and institutionalist theory— that is, any analysis that remains open to the possibility that variation in state preferences, as well as power and institutions, might influence state policy—liberal theory enjoys causal priority. Steven Krasner’s well-known metaphor captures this insight: If institutionalism determines whether governments reach the Pareto-frontier, and realism determines which point on the Pareto-frontier governments select, liberalism defines the shape of the Pareto-frontier itself.111 Surely the latter task is primary. This conclusion should hardly be surprising to political scientists, for it is the unambiguous lesson of the classic literature on the methodology of studying power and influence, whether in local communities or global politics. Robert Dahl’s analysis of power teaches us that we cannot ascertain whether ‘‘A influenced B to do something’’ (that is, influence) unless we know ‘‘what B would otherwise do’’ (that is, preferences).112 The implication for realism is clear: Not only do we need to know what state preferences are, but unless they are arrayed so that substantial interstate conflict of interest exists and the deployment of capabilities to achieve a marginal gain is acceptable, realist theory is powerless to explain state behavior. Similarly, institutionalist explanations of suboptimal cooperation are appropriate only under circumstances in which states have an interest in resolving particular interstate collective action problems. Kenneth Oye draws the implication: ‘‘When you observe conflict, think Deadlock—the absence of mutual interest—before puzzling over why a mutual interest was not realized.’’113 The analytical priority of liberalism is not simply an abstract requirement of theoretical consistency; it is empirically significant.114 Realists and institutionalists alike are retreating to what Keohane terms a ‘‘fall-back position,’’ whereby exogenous variation in the configuration of state interests defines the range of possible outcomes, within which capabilities and institutions explain outcomes.115 This implicitly concedes not just the need for multicausal synthesis, but the analytical priority of liberal theory. The popularity of the ‘‘fallback’’ position also defuses a practical objection often raised against ‘‘societal’’ or ‘‘domestic’’ theories, namely that research into domestic preferences is overly demanding, if not impossible. To be sure, the investigation of national motivations poses particular challenges. State preferences must be clearly distinguished from strategies and tactics and then must be inferred either by observ ing consistent patterns of state behavior or by systematically analyzing stable elements internal to states, as revealed in decision-making documents, trustworthy oral histories and memoirs, patterns of coalitional support, and the structure of domestic institutions.116 Yet the existence of such difficulties does not constitute a valid reason to neglect liberal theory. No respectable philosophy of science recognizes the difficulty of performing relevant empirical research with current techniques as a legitimate reason to abandon a promising scientific paradigm. Instead, scientific technique and training should adjust—an argument for thorough training in languages and primary-source analysis, as well as in rigorous theories of comparative politics. Moreover, the popularity of the fall-back position demonstrates that the difficulty of ascertaining preferences is not unique to liberalism. We have seen that even monocausal empirical tests of realist and institutionalist theories must control reliably for variation in underlying preferences (not just strategies) of states. This requires precisely the same detailed research into domestic politics. Such a baseline control is, moreover, most reliable where backed by an explicit and generalizable theory of domestic preference formation, that is, a liberal theory. In short, research into domestic preference formation is unavoidable. The priority of liberalism in multicausal models of state behavior implies, furthermore, that collective state behavior should be analyzed as a two-stage process of constrained social choice. States first define preferences—a stage explained by liberal theories of state-society relations. Then they debate, bargain, or fight to particular agreements—a second stage explained by realist and institutionalist (as well as liberal) theories of strategic interaction.117 The two-stage model offers a general structure for research design and theoretical explanation. In those cases where liberal factors only influence strategic outcomes directly, through preferences and preference intensities (a in Figure 1), liberalism can be tested as a monocausal hypothesis against alternative realist or institutionalist factors (c in Figure 1). Liberal factors may also influence outcomes indirectly, because the nature of preferences helps determine (b in Figure 1) the nature and strength of the causal relationship between strategic circumstances and actions (c in Figure 1). Recall that preferences do not simply shape outcomes, they tell us which realist or institutionalist factors are important and how they relate to state behavior. In such cases, explaining (or at least controlling for) variation in state preferences is analytically prior to an analysis of strategic interaction. Without a prior analysis of preferences, only monocausal formulations of realist or institutionalist theory can be tested.118 The primacy of liberal theory in such multicausal explanations may appear to be an abstract admonition, yet precisely this two-stage approach has characterized liberal theory and practice from Kant’s philosophy to the practical calculations by the American architects of the post–World War II settlement. Throughout, multicausal or two-stage liberalism makes sense of what have long been considered contradictions and ambiguities in classic liberal thought and modern liberal statecraft. Consider Wilson’s proposal for the League of Nations, often cited as the epitome of liberal ‘‘legalism’’ and ‘‘utopianism.’’ At first glance, Wilson’s proposal seems to reflect a naive confidence in international institutions. Understood as an implicit social science theory, not ideology, we see that it was neither utopian nor fundamentally institutionalist. It rested instead on a pragmatic two-stage liberal view, and its failure actually confirms liberal predictions. From the start, Wilson was skeptical about the autonomous influence of international institutions. He cared little about their precise form, because he viewed them as no more than ‘‘a symbolic affirmation of the ‘rightness’ of democracies in their mutual relations.’’119 Thus, for example, his initial draft of the Covenant included no provisions for international law or a supranational court; both were eventually added only at the insistence of more conservative (and more cynical) foreign and domestic politicians. Instead what he termed the ‘‘first point’’ to remember about the League was not institutionalist but liberal: Its membership was to be restricted to those countries enjoying republican government and national self-determination. Insofar as the League was to rely on public opinion, it was to be solely democratic public opinion. Based on a multicausal liberal analysis, Wilson explicitly identified a set of narrow preconditions under which collective security institutions could succeed. The League, he argued, would function only if nationally self-determining democracy was a nearly universal form of government among great powers, which in turn controlled an overwhelming proportion of global military power. In 1917, Wilson believed this situation to be imminent: ‘‘There are not going to be many other kinds of nations for long. . . . The Hapsburgs and the Hohenzollerns are permanently out of business.’’120 Given Wilson’s underlying theory, is it surprising that the League had become moribund by 1936, after twelve European countries had moved from democracy to dictatorship? Or that this shift isolated democratic France and Britain, exacerbating their oft-noted geopolitical dilemmas in Manchuria and Abyssinia? Here we again see the virtue of defining liberal theory in a nonideological manner: The failure of the League, often cited as a realist refutation of liberal ideology, in fact confirms liberal IR theory. Multicausal liberalism helps to explain not only ambitious schemes for cooperation like collective security, but ‘‘realist’’policy outcomes like power balancing and bipolar conflict. Kant, for example, recognized the balance of power as an unstable, second-best mechanism suitable only to a particular set of circumstances defined by liberal theory, namely relations among nonrepublican states. In theoretical terms, realism was embedded in a deeper and more encompassing transhistorical liberal theory of social development. The balance of power serves to limit the ‘‘vigorous . . . rivalry’’ among states, permitting the progressive emergence of republican government and commerce (as well as, though clearly secondary, international rules), which would in turn steadily diminish the relevance of interstate balancing. Like Wilson, Kant remained skeptical of strong international institutions, focusing instead on the development of societal preferences.121 A form of multicausal liberalism very similar to that espoused by Kant underlay the post–World War II U.S. policy of containment—a policy traditionally treated as the embodiment of realism. Containment was never simply power balancing. It was an integrated multicausal liberal grand strategy, as made explicit after World War I by Wilson and John Dewey, then after World War II by George Kennan. Kennan, in this regard a liberal, linked the European threat to the nature of the Soviet regime; it is often forgotten that nine-tenths of the seminal ‘‘X’’ article was given over to an analysis of Soviet domestic beliefs.122 A Western military deterrent would be required, he argued, only until the Bolshevik revolution had run its course, whereupon the Soviet system would collapse of its own accord. Thus the decisive Western actions in the Cold War, according to Kennan, were the reconstruction of Germany and Japan as capitalist democracies through policies like the Marshall Plan. The goal of the policy was the transformation of social purposes and state preferences in Western countries, neither of which would assume much importance in a purely realist analysis. This multicausal liberal interpretation of containment banishes various ambiguities and tensions in Kennan’s thought that have bedeviled biographers—not least his singular synthesis of balance-of-power thinking and strident antimilitarism.123 The conduct and conclusion of the Cold War proceeded precisely as Kennan’s two-stage liberal model had predicted. Realist power balancing served throughout as a static, interim instrument to maintain the status quo, but shifting state preferences explain the outbreak and eventual passing of the conflict. By 1959, standing in a Moscow exhibit of kitchenware, Richard Nixon and Nikita Khrushchev declared that the Cold War would be won and lost not through relative military capabilities, but through the relative economic prowess and ideological attractiveness of the two superpowers. Economic stagnation and a measure of ideological change in the East predated foreign policy change. If the West, as Khrushchev rashly promised, had been buried under the superior economic performance of the East, the outcome might well have been different.124 These examples demonstrate the ability of multicausal liberal theory to explain critical twentieth-century foreign policy decisions, such as those taken in 1918, 1947, and 1989, even when national security interests are fully engaged.125 In interpreting such cases, the major difference between realists and liberals lies not, as is often claimed, in the observation that states are concerned about security threats and balancing; this finding is consistent with a multicausal liberal explanation. Where the two theories genuinely differ is on the sources of security threats themselves, with realists attributing them to particular configurations of power, institutionalists attributing them to uncertainty, and liberals attributing them to ideological, institutional, and material conflict among state preferences. If liberal theory contributes to explaining core realist cases such as bipolar conflict, there is good reason to believe that the most powerful influences in world politics today are not the deployment of military force or the construction of international institutions, but the transformation of domestic and transnational social values, interests, and institutions. Conclusion: The Virtues of Theoretical Pluralism Liberal IR theory is not simply an ideological foil for more realistic and rigorous theories, as its critics claim, nor an eclectic collection of hypotheses linked only by common intellectual history and normative commitment, as its proponents are currently forced to concede. It is instead a logically coherent, theoretically distinct, empirically generalizable social scientific theory—one that follows from explicit assumptions and generates a rich range of related propositions about world politics that reach far beyond cases of cooperation among a minority of liberal states. By reformulating liberalism as theory rather than ideology, we have repeatedly seen that what are often treated as liberal failures become liberal predictions. Moreover, liberalism exhibits considerable potential for theoretical extension. Aside from the myriad opportunities for empirical testing and theoretical refinement of specific hypotheses, a number of broader areas are poised for theoretical innovation. Relaxing the assumption of unitary state behavior would support a range of ‘‘two level’’ hypotheses about the differential ability of various domestic state and societal actors to pursue semiautonomous transnational activities. Relaxing the assumption that decision making is static would support analyses of change over time. Greater attention to feedback from prior decisions mediated by intervening liberal factors like domestic ideas, institutions, and interests might provide firmer microfoundations for theories of regime stability and change—an area of potential collaboration with constructivists and historical institutionalists. Finally, the rich interaction among domestic and transnational ideas, interests, and institutions is only beginning to be explored. A final word to those readers who object to using the term liberal to distinguish this restatement. Such potential critics fall into two groups. One group is likely to find this formulation of liberal theory too narrow, the other too broad. The first group of critics will protest that this restatement fails to acknowledge the full richness of the intellectual history and, in particular, the normative implications of liberalism. This criticism is correct, but the omission is deliberate. This article does not aim to provide a comprehensive intellectual history of classical liberal international thought, nor a self-sufficient guide to the normative evaluation of policy, but to distill a coherent core of social scientific assumptions for the narrower purpose of explaining international politics.126 The project is best judged on its own terms—the four criteria outlined in the preceding section—not its fidelity to prior usage. The second group of critics will complain that liberalism has too many definitions as it stands, most too vague to be useful. Some reject altogether the use of ‘‘isms’’ to designate foundational theoretical positions in IR. This criticism is semantic rather than substantive. In contrast to other fundamental divisions—for example, those between domestic and systemic ‘‘levels of analysis,’’ optimistic and pessimistic prognoses, or realist, liberal, and Marxist ideologies—the tripartite division among realism, liberalism, and institutionalism is fully consistent with the foundation of rationalist social theory, which divides the determinants of social behavior into three categories: interests, resources, and institutions or information.127 Those who view state behavior as the result of a process of constrained choice would do well to champion rather than criticize efforts to impose greater theoretical coherence and consistency on theories of rational state behavior. Either type of critic may nonetheless prefer to call liberal theory a ‘‘societal,’’ ‘‘state-society,’’ ‘‘social purpose,’’ or ‘‘preference-based’’ theory. The central claims of this article, however, remain intact. First, major IR theories should be divided into those that stress the pattern of state preferences, the distribution of resources, and the institutional provision of information. Second, greater priority should be given to the further development of the first category. This development need not proceed ad hoc, but can be achieved by grounding such efforts in the common assumptions and causal processes proposed here. Only further research can reveal their full empirical power; yet existing studies—from explanations of the democratic peace to endogenous tariff theory to theories relating domestic institutions and ideas to foreign policy—suggest considerable promise. Third, a liberal theory of state preferences is the most fundamental type of IR theory. Hypotheses that endogenize changes in state preferences deserve equal treatment in monocausal explanations and analytical priority in multicausal ones, because liberal theory defines the theoretical and empirical domains in which it is appropriate even to consider realist and institutionalist claims. Thus those who ignore liberal theory do not simply sacrifice comprehensiveness; they undermine valid empirical evaluation of their own theories. Only by building on these three conclusions can liberals and their critics supplant debates over labels with debates over data.