522 International Organization 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 third- best strategies strikingly at variance with their underlying preferences.24 Thus vary- ing state preferences should be treated as if they were irrelevant,secondary,or endog- enous.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 inten- tions.26 Though the interests it assumes are different,Keohane's institutionalism re- lies 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 out- comes."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 inter- state political and strategic circumstances,they may well vary in response to a chang- ing transnational social context.In the political economy for foreign economic policy, 23.What about Marxism?Marxism provides distinctive normative insights(Doyle 1997),but its non- teleological positive assumptions-the centrality of domestic economic interests,the importance of trans- national interdependence,the state as a representative of dominant social forces-are quite compatible with this restatement of liberalism.For examples,see the contribution by Frieden and Rogowski in Keo- hane and Milner 1996. 24.Waltz1979,60-67,9397 25.The resulting "autonomy of the political"in geopolitics gives realism its "distinctive intellectual and moral attitude";see Morgenthau 1960.5-7.The fact that Morgenthau distinguished nonrealist ele- ments of his own thought illustrates a further danger of defining realism not in terms of social scientific assumptions,but in terms of its intellectual history,that is,assuming that everything a"realist"wrote constitutes a coherent realist theory;see Morgenthau 1960,5,227. 26.Waltz follows Morgenthau almost verbatim:"Neo-realism establishes the autonomy of interna- tional politics and thus makes a theory about it possible";see Waltz 1979,29,and also 65-66,79,90, 108-12.196-98.271. 27.See Keohane 1984,6;and Hellmann and Wolf 1993 28.Powell1994.318. 29.Ruggie1983,107-10
522 International Organization half-century. At the heart of the two leading contemporary IR theories, realism and institutionalism, is the belief that state behavior has ironic consequence^.^^ Power politics and informational uncertainty constrain states to pursue second- and thirdbest strategies strikingly at variance with their underlying preference^.^^ 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 preference^."^^ 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 intention~.~~ 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 c~operation."~~ In short, Powell observes that "structural theories . . . lack a theory of preferences over outcome~.''~~ 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, 23. What about Marxism? Marxism provides distinctive normative insights (Doyle 1997), but its nonteleological positive assumptions-the centrality of domestic economic interests, the importance of transnational interdependence, the state as a representative of dominant social forces-are quite compatible with this restatement of liberalism. For examples, see the contribution by Frieden and Rogowski in Keohane and Milner 1996. 24. Waltz 1979,60-67,93-97. 25. The resulting "autonomy of the political" in geopolitics gives realism its "distinctive intellectual and moral attitude"; see Morgenthau 1960, 5-7. The fact that Morgenthau distinguished nonrealist elements of his own thought illustrates a further danger of defining realism not in terms of social scientific assumptions, but in terms of its intellectual history, that is, assuming that everything a "realist" wrote constitutes a coherent realist theory; see Morgenthau 1960,5,227. 26. Waltz follows Morgenthau almost verbatim: "Neo-realism establishes the autonomy of international politics and thus makes a theory about it possible"; see Waltz 1979, 29, and also 65-66, 79, 90, 108-12, 196-98,271. 27. See Keohane 1984,6; and Hellmann and Wolf 1993. 28. Powell 1994,318. 29. Ruggie 1983, 107-10
Liberal Theory of International Politics 523 for example,social demands are derived not simply from "domestic"economic assets and endowments,but from the relative position of those assets and endow- ments 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 errone- ous 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 sim- ply 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 sys- temic 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 "sys- temic"outcomes of interstate interactions.That systemic predictions can follow from domestic theories of preferences should be obvious simply by inspecting the litera- ture on the democratic peace.32 In addition,by linking social purpose to the symme- try and relative intensity of state preferences,liberalism offers a distinctive concep- tion 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 primar- ily 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 inten- sity of actor preferences.33 The "win-set,"the "best alternative to negotiated agree- ment,"the pattern of"asymmetrical interdependence,"the relative opportunity cost of forgoing an agreement-all these core terms in negotiation analysis refer to differ- ent 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 equa- tion 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 tem- 30.For example,see Gourevitch 1976. 31.Ruggie1983,90-91. 32.For a more general argument,see Elman 1996,especially 58-59. 33.See Harsanyi 1977;Hirshman 1945;and Keohane and Nye 1987,733
Liberal Theory of International Politics 523 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 an~ther."~' 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 preference^.^^ 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 tem- 30. For example, see Gourevitch 1976. 31. Ruggie 1983,90-91. 32. For a more general argument, see Elman 1996, especially 58-59. 33. See Harsanyi 1977; Hirshman 1945; and Keohane and Nye 1987,733
524 International Organization poral 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 out- comes,is commonsensical.Nations are rarely prepared to expend their entire eco- nomic or defense capabilities,or to mortgage their entire domestic sovereignty,in pursuit of any single foreign policy goal.Few wars are total,few peaces Carthagin- ian.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 mar- gin,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 capabili- ties,"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,gener- ates 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 sepa- rate 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 funda- mental collective goods like national unity,legitimate political institutions,and socio- economic regulation.Commercial liberalism focuses on incentives created by oppor- tunities for transborder economic transactions.Republican liberalism focuses on the nature of domestic representation and the resulting possibilities for rent-seeking be- havior. 34.See Raiffa 1982;Sebenius 1991;Evans,Jacobson,and Putnam 1993;and Keohane and Nye 1977. 35.See Morrow 1988,83-84;and Mack 1975
524 International Organization poral 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 "determinationn--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. 34. See Raiffa 1982; Sebenius 1991; Evans, Jacobson, and Putnam 1993; and Keohane and Nye 1977. 35. See Morrow 1988,83-84; and Mack 1975
Liberal Theory of International Politics 525 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 so- cial 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 prefer- ences shared by individuals concerning the proper scope and nature of public goods provision,which in turn specifies the nature of legitimate domestic order by stipulat- ing 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 histori- cal 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 de- mands 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,politi- cal institutions,and modes of socioeconomic regulation. The consequences of identity-based preferences for IR depend,according to as- sumption 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 exter- nalities,harmony is likely.Where national claims can be made more compatible by reciprocal policy adjustment,cooperation is likely.39 Where social identities are in- compatible and create significant negative externalities,tension and zero-sum con- flict is more likely.Parallel predictions about international politics follow from each of the three essential sources of ideational preferences:national,political,and socio- economic identity.40 Let us briefly consider each. 36.This concept is similar but narrower than Ruggie's "legitimate social purpose"and Katzenstein's "collective identity";see Ruggie 1983;Katzenstein 1996a,6. 37.Here is a point of tangency with recent constructivist work;see Katzenstein 1996a,5:Finnemore 1996,27-28;and Wendt 1996,7.Whether the fundamental sources of societal preferences are ideational is the focus of a debate among general social theorists for which IR theorists lack any distinctive compara- tive advantage. 38.Fearon 1995. 39.0ye1986. 40.Liberal theory need not and in general does not claim that shared identities emerge from chance interactions among "atomistic"individuals,or that nationality must reflect "timeless"factors like lan- guage,religion,or ethnicity.Identities need only be translated into political preferences through individual and group commitments;compare Finnemore 1996,147
Liberal Theory of International Politics 525 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 factor^.^' 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 juri~diction.~~ 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. 36. This concept is similar but narrower than Ruggie's "legitimate social purpose" and Katzenstein's "collective identity"; see Ruggie 1983; Katzenstein 1996a, 6. 37. Here is a point of tangency with recent constructivist work; see Katzenstein 1996a, 5; Finnemore 1996,27-28; and Wendt 1996,7. Whether the fundamental sources of societal preferences are ideational is the focus of a debate among general social theorists for which IR theorists lack any distinctive comparative advantage. 38. Fearon 1995. 39. Oye 1986. 40. Liberal theory need not and in general does not claim that shared identities emerge from chance interactions among "atomistic" individuals, or that nationality must reflect "timeless" factors like language, religion, or ethnicity. Identities need only he translated into political preferences through individual and group commitments; compare Finnemore 1996, 147