Realism in the Study of World Politics Author(S): Robert Jervis Reviewed Work(S): Source: International Organization, Vol

Home , Robert Jervis, Robert Keohane

Realism in the Study of World Politics Author(s): Robert Jervis Reviewed work(s): Source: International Organization, Vol. 52, No. 4, International Organization at Fifty: Exploration and Contestation in the Study of World Politics (Autumn, 1998), pp. 971-991 Published by: The MIT Press Stable URL: http://www.jstor.org/stable/2601364 . Accessed: 09/03/2012 15:36

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to International Organization.

http://www.jstor.org Realismin the Studyof World Politics RobertJervis

Fiftyyears of International Organization encompass both accomplishments and dis- appointments,as does thisperiod of intemational history. The articlesin thisissue of 10, especially"International Organization and the Study of World Politics" by Peter J. Katzenstein,Robert 0. Keohane,and StephenD. Krasner,also give us a lot to thinkabout, learn from, and criticize. I willnot attempt a full review but, as befitting a commentator,will concentrate on questionsthat the other authors have downplayed orthat I wouldpose or answersomewhat differently, perhaps because of mygreater familiaritywith the work in securitystudies than in intemationalpolitical economy (IPE), a fieldin which10 andmost of the authors in thisissue specialize. ' I beginby discussingsome determinantsof thefates of schoolsof thoughtand researchprograms, noting the relationships between scholarship and contemporary internationalpolitics. I thenturn to thetheme presented by Katzenstein,Keohane, and Krasnerthat rationalism and constructivismwill be thetwo points of contestationin the coming years, which will lead intomy claims for the continuing relevance of realism.In closing,I will commenton thefield of IPE, aboutwhich the other authorsin thisissue have said surprisinglylittle.

ResearchPrograms, Knowledge, and Politics

Ifthe discipline is functioningwell, each school of thought enriches others as power- fulresearch of one kindstrengthens, not weakens, the alternatives. No one approach

I am gratefulfor comments by RichardBetts, Judith Goldstein, Peter Katzenstein, Robert Keohane, StephenKrasner, Helen Milner, Jack Snyder, and Arthur Stein. 1. Anothergood reviewconcentrating on IPE broadlyconceived is Kahler1997. I shouldalso join Katzenstein,Keohane, and Krasner and Wever (articlesin thisissue) in noting that although our subject is internationalpolitics, the relevant community of scholars is almostentirely American, with a fewcontribu- torsfrom the rest of theEnglish-speaking world. We do notknow the impact of thisconditioni, but I suspectit is considerable.To takejust one example,the fact that during the Cold WarAmerican security scholarshipwas centeredmore on realismthan was Europeanis notlikely to havebeen unrelated to the differentroles and policies of theUnited States and WestEurope. For earlierdiscussions of different nationalscholarly perspectives, see Hoffmann1977; Strange1983; Alker and Biersteker 1984; andHolsti 1985.

InternationalOrganization 52, 4, Autumn 1998, pp. 971-991 ? 1998by The 10 Foundationand the Massachusetts Institute of Technology 972 InternationalOrganization consistentlymaintains a leadingposition: each of themcatches important elements of internationalpolitics, and manyof our argumentsare aboutthe relative impor- tanceof and the interrelationships among various factors. Thus, in their article in this issue,Katzenstein, Keohane, and Krasnerusefully point to thedialectical nature of social science.Popular approaches inevitably are taken too farand call up opposing linesof argument;and if any importantapproach is ignoredfor too long,scholars will returnto it as thepicture of internationalpolitics becomes excessively imbal- anced.2 Althoughit is easyto see thatvarious kinds of research wax andwane, explaining thepattern is moredifficult. Indeed, there is an elementof circularity in determining whatconstitutes a successful research program. In theabsence of someexternal and arguablyobjective measure, a researchprogram succeeds when many scholars adopt it.Without claiming that initial incidents set off positive feedback to suchan extent thatsuccess is accidentalor arbitrary,we shouldnot assume that those approaches currentlymost popular necessarily tell us moreabout international politics than do alternativeapproaches. Determining which research programs are "progressive"and whichare "degenerating"is difficultbecause the relevant judgments are influenced byour perspectives and interests (in bothsenses of the term) and because all theories undergochange in lightof empiricalinvestigations.3 Thus, though I wouldnot dis- sentfrom the consensus that the democratic peace has beenan extraordinarilyfruit- fularea of scholarship,there are grounds for arguing that we havelearned relatively littlesince the originalinvestigations. Many of the findingshinge on definitions, debatablecodings affect the results, the causal mechanisms remain unclear and have notproved readily amenable to empiricalresearch, and it is questionablewhether additionalphenomena, such as the purportedtendency of democraciesto resolve differencesshort of war, can be fitunder the same theoretical umbrella. The otherside of thiscoin is thatthe failure of a researchprogram may not be primarilyattributable to its lack of potential.For example,although I join many othersin believingthat there are greatlimitations to theutility of bureaucraticpoli- ticsanalysis because it is difficultto specifyex antethe actors' preferences and even harderto saywhat outcome is likelyto emergefrom the posited pushing and hauling amongthe diversebureaucratic interests, these failings are hardlyunique to this approach.Can one arguethat bureaucratic politics yields fewer testable propositions thanconstructivism, for example? Thus, I wouldnot dismiss four other sources of the approach'scurrent lack ofpopularity. First, it resonateswith, and indeedpartly was developedby, those who had beenin thegovernment. As thepolitical science disci- plinehas separatedfrom the practice of politics,fewer people move back and forth betweenthe academy and thegovernment, thereby reducing the ranks of those who arelikely to producethis kind of work or find it intellectually satisfying. Second, the

2. CitingGoethe, Stephen Jay Gould notes that "some dichotomies must interpenetrate, and not struggle to thedeath of one side,because each oftheir opposite poles capturesan essentialproperty of any intelli- gibleworld"; Gould 1987,19. 3. Lakatos 1970. For contrastingviews of thestatus of realismin theseterms, see thearticles in the "Forum"section of the American Political Science Review 1997, 899-935. Realismin World Politics 973 onlyneighboring field that bureaucratic politics can drawon is organizationtheory fromsociology, which, compared to thefield of economics (which has lentso much weightto the rationalchoice school),is of littleprestige or assistance.Third, al- thoughstudies of foreignpolicymaking often produce explanations involving the detailsof bureaucraticperspectives and interests,the discipline now favorstheories that,while admitting of manyexceptions, are moreparsimonious. Finally, unlike constructivism,bureaucratic politics does nothold out hope fordrastic change. Al- thoughit is usuallycritical of the policies that have been adopted and, even more, of theway theyare implemented,it does notquestion the decision makers' basic out- lookand goals, let alone the prevailing pattern of international politics. Bothrecent events and the contemporary political atmosphere influence the accept- abilityof theories, though this is usuallyclearer in retrospectthan at thetime. Thus, JeraldCombs has shownthat interpretations ofpast events in American foreign policy werestrongly marked by whatwas happeningwhen scholars were writing.4 In the samevein, the changing fortunes of Marxism in political science are better explained by suchevents as theVietnam War (which fits Marxism badly5) and thedisintegra- tionof theSoviet Union coupled with the abandonment of communismin thePeo- ple's Republicof China(which actually says nothingabout the validity of Marxist theory)than by the ability or inability of Marxist approaches to generatepropositions or accountfor evidence. So, notsulprisingly, the relatively uncompelling nature of theinternational environment now facingthe United States has renewedinterest in howstate interests are defined and in the domestic sources of foreign policy, not only currentlybut also in previouseras. Similarly, scholars often argue that a resurgence ofviolence between states would increase the popularity of realism. I haveno doubt thisis correct,but even if we wereto acceptthe questionable equation of realism withviolence, would we wantcurrent events to driveour research, rather than being concernedwith explaining patterns over a longerspan of history?6Contemporary eventscan of course be relevanthere-and waramong the developed countries would underminesome theoriesdiscussed later-but theyusually tell us moreabout the popularityof various approaches than about their scholarly utility. The kindsof theories we findattractive are influenced not only by eventsbut also byour general political orientations. As usual,seeing these forces operating in others is easierthan seeing them in one's self,and though I agreewith the description of the evolutionof the subfield of international organization presented here by Katzenstein, Keohane,and Krasner, I see normativeconsiderations playing a largerrole. To over- simplify,the field moved from analyzing the United Nations, to studyingregional integration,to lookingat therole of transnationaland transgovemmentalorganiza- tions,to examiningthe causes and effectsof foreigneconomic policies and ac-

4. Combs1983. 5. Krasner1978. 6. Duringthe Cold War,the field of securitystudies was criticizedfor being preoccupied with contem- poraryissues at the expense of more abstract arguments that would apply to a widersweep of history. So it is a littleodd tohear many of the same people now say that the end of the Cold Warmakes security studies less important. 974 InternationalOrganization tivities.The firstthree phenomena, and to a lesserextent the fourth, are ones in which theuse or threatof forceis in thebackground, if notabsent, and each appearedto holdout hope for a morepeaceful world, perhaps even a morehumane and just one. The subfieldhas been guidedby the beliefsthat such changesare possible,that greaterareas of world politics can be governedby law ratherthan by force,and that universalisticcriteria should and can becomeincreasingly important. The normativeagenda is evenmore apparent with social constructivism. The obvi- ous desireis to see worldpolitics transformed by thespread of appropriatenorms, identities,and concepts of world politics. Intriguingly, constructivists pay little attentionto normsand ideas that are both revolutionary and evil.This oversight has been at thecost of ignoring perhaps the strongest argument against materialist claims: one cannotunderstand Hitler's Germany and Stalin's Soviet Union, the two most destruc- tivestates of the twentieth century, without understanding the norms, identities, and ideasheld by the dominant elites.7 Both regimes were driven by the desire to remake firsttheir societies and thenthe world.Neither took the stateas the unitwhose interestswere to be served;no narrowself-interest or considerationsof national securityor even national gain could have led to suchdomestic slaughter and danger- ous expansion.8Perhaps constructivists slight these cases becausethey indicate that realismcan be a forcefor moderation, that new ideas can makethe world worse, and thatthose who seek radical change may be monsters.Recent civil strife has increased constructivists'(and everyoneelse's) interestin nationalism,but it will be a chal- lengeto deal withdesired and undesired ideas in analyticallysimilar ways. Normativeimpulses also accountfor much of the passion in the attacks on realism. RobertGilpin's claim that "no one loves a politicalrealist"9 is onlya bitexagger- ated:those who are committedto theneed forand possibilityof radicalchange in internationalpolitics find what they believe to be realism'sdeep pessimismon this scorenot onlyincorrect but also a majorobstacle to progress.10Similarly, liberal reformershave long argued that those who describe the world in termsof theutility of theuse, or eventhe threat, of forceare teachingincorrect and dangerouslessons andthat realism partly is a self-fulfillingprophecy. "I Thisis notto say thatrealists like war, although in earliereras many intellectuals believedthat armed conflict was themain motor of humanprogress-a proposition thatshould not be dismissedout of hand.Indeed, many realists study the causes of warin thehope of reducing the chances of futureconflict. What I findinteresting is

7. Lumsdaine1994. See also Mearsheimer1994, 42-44. 8. Relevanthere is thegeneral topic of passions and interests:Hirschman 1977. In thisconnection, the associationbetween constructivism and "left"political preferences is significant.This commitment gives thework much of itsdrive and, for many, its appeal. But it also raisesdoubts as to whetherthe approach can produceanswers-or even ask questions-thathave unpalatable political implications. 9. Gilpin1996. 10. Althoughrealism sees thepossibility of cooperation and peace, as I willdiscuss later, there is much to K. J.Holsti's comment that "many of the theoretical arguments about the fundamental contours of our disciplineare really debates about optimism and pessimism, our very general outlooks toward the world in whichwe live"; Holsti1986, 356. 11. See, forexample, Rapoport 1973. Realismin World Politics 975 thatnonrealists have been slow to developand testarguments about the conditions theyconsider conducive to peace. A largeliterature exists on theeffects of threats and aims increases,but even leaving aside thequestionable relevance of theanswers to thesecrucial questions for realism and itsalternatives,"2 this research is plaguedby almostinsurmountable endogeneity problems, especially when decision making is notclosely examined. Classical liberalism was foundedon theidea thatunfettered internationalintercourse would produce peace, but only recently have scholars reex- aminedthe correlations between high levels of tradeand conflict,'3and neoliberal institutionalistshave notlooked at whetherpeace has historicallybeen associated witha highdensity of international institutions. (Of course,we needto considerthat causationcan runboth ways and thatpeace and its purportedcauses maybe the productsof thirdfactors. 14) Thus,I thinkit is thehope for a betterworld more than theresults of investigationsof thecauses of conflictsthat has led manyscholars to rejectrealism. In summary,then, we shouldnot adoptthe Whiggish stance that the fateof a researchprogram is predominantlydetermined by theextent to whichit produces propositionsthat anticipate and fitwith empirical facts. Programs-and, even more, theirfirst cousins, paradigms-are notoriously difficult to confirmor disconfirm. Not onlydo theyshape what counts as a factat all, butthere are so manysteps between theassumptions and outlookson theone handand empiricalfindings on theother thatneither in socialnor in naturalscience can theevidence ever be unambiguous.If thiswere not the case, competingtheories of politicallife could not have survived overseveral centuries. Granted, they rise and fall and undergo permutations (and, we hope,improvements), but the basic schoolsof thoughtof realism, liberalism, Marx- ism,and constructivism have been around for a verylong time. Any that could have beendisconfirmed would have gone extinct long ago.

Constructivismand Rationalism:Necessarily Incomplete

Leavingfor the next section the question of whetherrationalism and constructivism willbe themajor points of contestationin thecoming years, here I wantto notethat theyare even less completethan realism, liberalism, and Marxism. As approachesor stylesof thoughtthey need to be filledwith content in orderto becometheoretical statements,and much of their explanatory power must come from auxiliary assumptionsabout the identities of actors,their goals, and theirbeliefs. This is mostobvi- ouslytrue for rationalism, which argues that behavior can be understoodas theac- tors'attempts to maximize some consistent utility function. But this claim says nothing

12. The viewthat arms increases and detetrencecan preservethe status quo is clearlyrealist, but the claim thatthese postures will set offa spiralof increasingconflict is notnecessatily nonrealist: Jervis 1976,chap. 3. 13. See Oneal and Russett1997; and Kim 1998. For moreskeptical views, see Copeland1996; and Ripsmanand Blanchard 1996. 14. Blainey1973, chap. 2. 976 InternationalOrganization aboutwhat actors value and whatbehaviors they believe will producemaximum benefit.To movebeyond a "thin"version of rationalchoice, a greatdeal mustbe added,which means that there can be no singlerational choice theory of politics. Rationalismthen should not be contrastedwith liberalism or realism.Indeed, it needs theorieslike theseto do any explanatorywork. Debate over the meritsof assumingutility maximization has obscuredthe crucial nature of thesuppositions aboutwhat goals actorsseek and how theybelieve they can bestreach them. Gloss- ingover these questions in IPE is perhapseasier than in securitybecause both goals andunderstandings ofmeans-ends relationships are more widely shared in the former areathan in thelatter. But even when applied to IPE, rationalismcannot successfully bracketpeople's ideas abouthow the economyfunctions: actors who believethe theoryof mercantilismwill behave verydifferently from those who have been schooledin neoclassicaleconomics. Thereare several affinities between rationalism and realism, however. Most formu- lationsof the latter rely on an assumptionof rationality,'5 though to a less demanding degreethan does theformer. Self-interest is usually seen as quitenarrow, although rationalistsare more prone than realists to see theinterest being maximized as thatof theindividual rather than the country being led, a topicto which I willreturn. Neither approachpays much attention to differencesamong individuals or biasesin theway peoplethink, leaving them vulnerable to criticisms rooted in individual and cognitive psychology.Furthermore, both approaches are simultaneously normative and descrip- tivebecause they run together what actors should do to servetheir interests and how theyactually behave. Thus, rationalist accounts of institutionstend to resemblethe functionalistapproach that was originallya targetof rationalism;that is, theinstitu- tionalarrangements that have developed are seen as thosethat best serve the interests of powerfulactors.16 But whenrealists and rationalists confront policies they see as misguided,if not disastrous, their theories as wellas thecountry are in trouble:Hans Morgenthaunot only said thatcountries follow their national interests, he also lec- turedAmericans on theneed to do so-a taskthat would have been unnecessary had hisdescriptive argument been without flaws. Constructivism,too, is an approachor a style.It is one thingto arguethat material factorsand the external environment do notdetermine a state's behavior and to point to theimportance of regulativeand constitutivenorms, shared understandings, and commonpractices; it is quiteanother to say how normsare formed,how identities areshaped, and how interests become defined as theydo. Leavingaside the question of whatevidence would bear on theclaims being made, my point is thatalthough constructivismsays somethingabout the processes at workin politicallife, it does not,by itself, tell us anythingabout the expected content of foreign policies or inter- nationalrelations. Constructivismdoes, however, have an importantaffinity with realism in itsskep- ticismabout the universality, ifnot the sincerity, of the ideas and rationales expressed

15. Morgenthaudoes notdo so consistently,however; Morgenthau 1978. 16. Sometimesmicrofoundations can be developedfor this account. Realismin World Politics 977 by nationalleaders. Realists are oftenaccused of too readilyadopting the perspec- tivesof thosein power,but scholars of thispersuasion often argue that the beliefs articulatedby statesmen(and otherpeople as well) arereflections of theirhistorical and personalcircumstances, that elites tend to universalizethe concepts and values thatare particular to their own situation or era, and that leaders are likely to convince themselvesthat worldwide interests are servedby policiesthat mainly benefit their countriesand perhaps only themselves. There is a valuablecynicism to both constructivismand realism in theappreciation of the self-righteousness ofpowerful actors."7 E. H. Carr'sanalysis is bestknown in thisregard, but Morgenthau similarly stressed thatliberal elites attributed objective and universal validity to the ideas that accompa- niedthe rise to powerof theircountries and social strata.'8Subsequent realists have also been quickto pointout thatthe justifications offered by thedominant powers typicallyclothe national interests in broaderterms; Kenneth Waltz continues this traditionwhen he arguesthat Americans fail to recognize that although inhibiting the spreadof nuclearweapons is indeedgood forthe United States, it disadvantages nonnuclearcountries that face threats, including threats from the United States.19 Interestingly,liberal IPE scholarshiphas less in commonwith constructivism on thisscore in thatits arguments parallel those made by dominantelites who espouse thebenefits of an openeconomic system. Because these scholars see greatercommon interestsamong actors and perhapsbecause theyare less focusedon themultiple effectsof power,including its influenceon theintellectual frameworks of power- holders,they tend to accept the claims of liberal leaders to be servinginterests extend- ingbeyond those of the leading actors. In manyof their variants, rationalism and constructivism share a valuablefocus on theimportance of interaction.But, contrary to theclaims of theirproponents, the extentof theirdifference from realism is notclear. For rationalism,interaction is treatedthrough game theory, which leads to an understandingthat outcomes often divergefrom intentions and that actors set their behavior on thebasis oftheir expectationof how othersare goingto act and withthe knowledge that others are doing likewise.This contrasts with the error common to much formal decision theory-that is, theassumption that although the actor that is thefocus of attention will maximize his or herutility, others will notanticipate this and reactaccordingly. But thebasic outlookof gametheory does notcontrast with realism and sophisticateddiplomatic histories,which have been deeply concerned with interactions as wellas withactors' anticipations,strategies, and estimates of others' strategies and anticipations. Constructivismhas a largeplace forthe ways in whichnorms and practicesnot onlyrestrict what actors can do butalso enablethem to act and,indeed, shape their identitiesand constitutethe fundamental nature of theprevailing international relations.By contrast,constructivists argue, neorealism, if not realism, assumes the iden-

17. Thereis a tensionin constructivismhere: to arguethat ideas largely are rationalizations is to locate majorcausality in preexistinginterests, whose existence constructivists usually deny. If ideas areto have anyautonomy, they cannot be merereflections of actors' experiences and interests. 18. See Carr[1946] 1964; andMorgenthau 1946, 20-53, 153-67. 19. Waltzand Sagan 1995. 978 InternationalOrganization titiesand existence of the states and does notsee howthe units and the international systemmutually form each other. But Waltz's conception of the relations between the unitsand thesystem is notso differentfrom theirs: "B's attributesand actionsare affectedby A, and vice versa.Each is notonly influencing the other; both are being influencedby thesituation their interaction creates.... The behaviorof [a pairof unitsin a system]cannot ... be resolvedinto a setof two-way relations because each elementof behavior that contributes to theinteraction is itself shaped by their being a pair."20 Althoughrationalism does notsee theinteractive process operating as profoundly as this,far from being starkly opposed to constructivism,in a relatedarea thetwo needto be combined.Game theory rests on assumptionsabout each actor'sexpecta- tionsabout how the other will behave-expectations that form socially, both through establishingconventions about the meaning of behavior and by actorstrying to con- vinceothers to accepttheir explanations of theirpast behavior.2' The centralityof interactionsand anticipationsleads actorsto seek to shapetheir informational and interpretationalenvironments; meanings and expectationsare sourcesof influence andsites of cooperation and contestation. In understandingthese processes, strategic rationalityand deductive logic can be-indeed, needto be-coupled withan appre- ciationof how actorsattribute meaning to behavior.Rationalism cannot supply this knowledge,but constructivism is one of theapproaches that can guidethe required empiricalresearch. Anydiplomatic history or newspaperprovides examples of theseprocesses and showsthat meanings are bothcentral and problematic.Let me just mentionthree cases thatoccuiTed in a two-dayperiod in January 1998 as I was draftingthis article: theUlster Protestants "wanted any concessions [from Britain and Ireland] to be part of thenegotiating process rather than 'sweeties' passed out by theBritish and Irish governmentsoutside the talks"; Iran'spresident proposed cultural exchanges with theUnited States, whereas the latter would accept only government-to-government talks;and theUnited States, Israel, and Turkey staged joint naval operations to im- provetheir abilities to coordinaterescues at sea, an actthat called up strongprotests fromSyria and Egypt.22 People here are concerned with the political and psychological significancesof theseacts, not their physical consequences. Most behaviorhas influenceonly as itis interpretedby others who hold their own (often implicit) theo- riesabout how the world works and who are trying to discernthe implications of acts forfuture behavior, while keeping in mindthat the actorand otheraudiences are engagingin similarattributions and projectionsof meaning.There is no objective way of specifyingthe significance people will attachto theseacts, and indeedthey maybe readdifferently bydifferent audiences. The concernsare rational and fit well

20. Waltz1979, 74-75. 21. See Jervis[1970] 1989a;and Kreps1990b. 22. WarTenHoge, "Protestant Threat Impeiils Peace Talksin Ulster," Newv York Times, 7 January1998, 4; ElaineSciolino, "Seeking to Opento U.S., IranianProposes Cultural Ties," New YorkTimles, 8 January 1998,1; andSerge Schmemann, "Unusual Naval Alliance Shows Off, and Arabs Glare," New York-Timzes, 8 January,3. Realismin World Politics 979 withstrategic logic, but the crucial interpretations that people are making and trying to getothers to makemust come from elsewhere and can onlybe understoodwithin frameworksthat are social, psychological, and cultural.

Rationalismand Constructivismas Central?

AlthoughI believethat constructivism and rationalismwill play largeroles in the future,to predictthat they will be pointsof contestationmay be an exaggeration. Much researchis too eclecticto be readilyclassified, which I thinkis a sign of healthydiversity. I find it difficult to fitmany of the articles in IO underthe theoretical headingsused in thisissue. Lookingat otherjournals, especially International Studies Quarterly and Journal of ConflictResolution, reinforcesthis impression. Herewe see empiricalresearch using large-scale data sets that is less closelytied to strongtheories. Put anotherway, the articles in thisissue are strongly"bicoastal." Quitedifferent work tends to be donein the rest of the country (Chicago is an honor- arycoastal university, perhaps by virtue of being on Lake Michigan).I do notwant to exaggeratethis distinction, but neither should it be dismissed.Without rehearsing the familiararguments about induction versus deduction, the value and frequencyof approachesthat do less to elaborategeneral theories and moreto detectpatterns in eventsshould not be underestimated. It is also tellingthat two of themajor debates in thefield and one of themajor pointsof focus have been only marginally influenced by rationalism and constructivism.To startwith the latter, the past decade has seena resurgenceof attentionto the domesticsources of foreignpolicy.23 Many of thearticles in thisissue discussthis research,and I needonly note that though some of it is groundedin rationalismor constructivism,most is not.Although constructivism attends to therole of domestic processesin formingidentities, interests, and norms,much work with a domestic focuseither brackets these subjects or analyzes them quite differently. Whereas ratio- nalismlinked to methodologicalindividualism sees foreignpolicy as theproduct of thenarrow self-interests of domesticactors and focuseson how domesticpolitical institutionsinfluence policy outcomes, much of the domesticsources literature is morediverse-and also shareswith bureaucratic politics the difficulty of estimating how thedomestic bargaining processes will workthemselves out, which coalitions willform, and how thenational leaders will balance internal and externalpressures and values.Thus, the renewed attention to "secondimage"24 should not be equated withthe triumph of either constructivism orrationalism. So itis notentirely surprising that the theory of the democratic peace, probably the mostvibrant theory drawing on domesticsources, cannot be readilyclassified as eitherrationalist or constructivist.Both styleshave contributedto thistopic, but

23. Forinteresting discussions, see Sterling-Folker1997; and Fearon 1998. 24. Waltz1959. For recent statements of theliberal version of thedomestic sources of foreignpolicy, see Doyle 1997and Moravcsik 1997, neither of which presents a rationalistor a constructivistaccount. 980 InternationalOrganization neitherhas dominatedit. Whatis arguablythe most important challenge to one of realism'scentral precepts did notarise from a strongmethodological vision but in- steadfrom a renewedsense of thevalidity of theliberal ideas thatdemocracies are differentfrom autocracies and that foreign policy reflects domestic habits and values, coupledwith intensive analysis of data and case studies. Similarly,the contending arguments concerning the purported rise of globalization arenot dominated by rationalism and constructivism.25 Instead, the range of analyti- cal tools and approachesremains broad and thebasic questionsaddressed fit well withtraditional realism, particularly in thefocus on theextent to whichthe external environmentdisciplines the behavior of statesand otheractors. The argumentthat highlevels of economic interaction impose severe limits on states'abilities to choose theireconomic policies parallels Waltz's claim that states that do notconform to the necessarymodes of internationalbehavior will "fallby thewayside".26 The causal mechanismsare a bitdifferent with globalization and involve a greaterrole for nongovernmentalactors, but the underlying relationships between competing actors and theresulting constraints are much the same.

Realism:Alive and Well

Relatedto mydoubts about the centrality of constructivismand rationalismis my beliefthat realism is notlikely to disappear.Neorealism may become less important notonly because many people are becoming tired of it,but also becauseits concern withthe differences between bipolar and multipolarsystems, especially in termsof stability,appears less interestingwith the declining fear of major war and the end of bipolarity.27As Waltznotes, realism, in contrastto neorealism,is moreof an ap- proachthan a theory-itpoints to a set of actorsthat are important,makes claims aboutthe considerations that decision makers weigh, and describes sets of outcomes thatcan resultfrom particular combinations of nationalpolicies.28 Although it does notreadily yield specific propositions, it has continuallygenerated new questions, insights,and arguments. Realismhas manyversions, but the assumptions that states can be consideredthe mainactors and that they focus in thefirst instance on theirown security are central to most.They are, of course, descriptively inaccurate. But, as almostall socialscien- tistsagree, this is notthe point. Rather, we ask whetherthese assumptions yield a widerarray of better confirmed propositions than do alternativeapproaches. Without claimingthat realism is appropriatefor all questions,I wouldlike to advancethe

25. This subjecthas beentreated extensively in thepages ofIO and is examinedin Garrett'sessay in thisissue. 26. Waltz1979, 71, 91, 118-19. 27. For thedifferences between realism and neorealism,see Gilpin1984; Waltz1991; Shimko1992; and Schwellerand Priess 1997. For recentdiscussions of the utilityof both,see the special issues of SecurityStudies, winter 1995 and spring1996; and the "Forum"section of American Political Science Review1997, 899-935. 28. Waltz1991. Realismin World Politics 981 weakerbut still not trivial argument that these assumptions are oftenof greatutility andthat it is unlikelythat we willsee highlyproductive theories that abandon all of themor that start from their opposites. Let me beginwith the assumption that states seek a highmeasure of security.29 Thisis notto claim that security is unambiguousor that it is theonly value.30 Indeed, ratherthan pay the price of destruction in war,states have surrendered in thehope of regainingtheir autonomy later (partly through the efforts of others).They have also peacefully(and not so peacefully)submerged their political units into those of others orjoined together to form larger units in thebelief that doing so wouldbetter serve a varietyof political,social, and ideologicalgoals. But if securityis rarelythe only objective,even morerarely can it be ignored.Of coursesecurity has been defined differentlyby differentactors, and the routes to itcan be multipleand contested, but thedesire for security is partof the bedrock explanation for why international politics existsat all. Thatis, though it is easyto take for granted the fact that no unithas come to dominatethe entire international system, this outcome needs to be explained.The desirefor security, coupled with the knowledge that one's currentallies may be one's adversariesin thefuture and thatcurrent adversaries may providefuture support generatesmany of the constraintsthat maintain the internationalsystem because self-protectiondictates that states do notwant their allies excessively aggrandized or theiradversaries excessively diminished. Is thesecurity assumption relevant today? As thedeveloped states fear each other muchless and have come to formwhat Karl Deutschcalled a pluralisticsecurity community,31other values come to thefore. But thesestates still worry about secu- ritythreats from elsewhere (including nonstate actors), and othercountries remain deeplyconcerned about their neighbors. Nevertheless, if itis truethat the most powerfulstates in the world no longerthink they may have to fight each other,the change inworld politics will be verygreat, a topicto which I willreturn. Note, however, that theimportance of thisdevelopment is apparentonly when we see thatthroughout historystates have focused on theirsecurity concerns. This analysisassumes that states are themain actors, which is mostappropriate whenthe values at stakeare widelyshared or, if they are not,when the top decision makersare motivatedand able to supportpolicies that will servethe country as a whole.But even in othercircumstances the state is notlikely to lack importanceand autonomy.For all our discussionof theroles of bureaucracies,economic sectors, multinationalcorporations, and transnationalinterests, in mostcases it is thestate thatis thetarget of their activities and it is mainlythrough national policies that these groupscan haveinfluence. For further evidence of the continuing centrality of states

29. One can, of course,talk of thesecurity of individuals,societies, or of theinternational system as wellas of states(Buzan 1991),but it is interestingto notethat aggressive states are also likelyto oppress theirown populations. 30. Wolfers1962, chap. 10. 31. Deutschet al. 1957. The countriesI have in mindare the UnitedStates, West Europe, Japan, Australia,and New Zealand.The questionof why these states but not others are so included,important as itis formany analyses, fortunately can be putaside here. 982 InternationalOrganization we need onlylook to countriesundergoing bloody internal struggles in whichthe objectiveis to gaincontrol of whatevernational machinery there is or to secedeand establishan independentstate. Even if statesare never the only actors on thescene, can we writehistory or current analysis without them? Vulgar Marxists, pluralists, or transnationalistswould see thestate as merelyregistering and implementing the views of powerfulsocietal actors. If theseperspectives could be maintained,we should dispensewith the names of the states and their leaders, which serve as ventriloquists' dummies,and write our accounts in termsof the groups and interests determining the behavior. The statehas provenremarkably resilient in theface of multiple social forces and the insistenceof scholarsthat its importanceis rapidlywaning. Assertions to the contraryby realists are less importantthan the actions by national leaders to reassert theircontrol, often supported by nongovernmentalactors who see greatvalue in centralauthority. Of coursethe fact that previous obituaries of the state were prema- turedoes not mean that they are not warranted now. In the1960s the state was indeed obstinaterather than obsolete, as StanleyHoffmann argued,32 but the European Union mayyet supplant its members-in which case it wouldform a stateof itsown, and thoughthe process by whichit formedmay violate some realist assumptions, many of theconstraints and incentives that it wouldface will be familiar.The claimsthat globalizationhas hollowedout statesmay be similarlyoverstated in partbecause theyoverestimate the implications of theeconomic flows and in partbecause they failto appreciate the way in which new forces call up newincentives and instruments forstate action.33 Arguingthat states are the central actors does nottell us whichinterests and policies theypursue. This questionlooms particularly large in thesecurity field: even thoughit maybe truethat all stateswant a highmeasure of security,some strive for othersgoals, especiallyexpansion of variouskinds, in additionto or even at the expenseof security.34Furthermore, even if securityis theprime objective, this does nottell us-or statesmen-whatbehavior will reachit.35 For example,belligerent policiesare likelyto decreaserather than increase the state'ssecurity when other

32. Hoffmann1966. 33. See Huntington1973; Haskel 1980; Gilpin 1996; and Garrett 1998a. For a paralleldiscussion of the relationsbetween states and transnationalactors, see Krasner1995b. For a strongclaim thatwe are witnessing"the retreat of the state," see Strange1996. 34. Disagreementsover the prevalence of expansionism is perhapsthe main issue between "offensive" and "defensive"realism; see, for example, Snyder 1991; Zakaria 1992; Glaser 1994; and Schweller 1996. See also Brooks1997; Labs 1997;and Kydd 1997. For further discussion, see pp. 986-87. 35. See, forexample, Snyder 1984; Van Evera 1984b;and Rhodes 1996. It is also worthnoting that studiesof theconduct of war,a topicthat has beenpushed to themargins of scholarship,have bitterly debatedthe extentto whichthe outcomesof battlesand wars are detenininedby materialforces (for example,equipment, firepower) as opposedto moraleor moralfactors (for example, faith in one's cause andcomrades). Many militaiy leaders have stressedthe latter-and it is notwithout interest that they use theterms mlorcal and nor-le interchangeablywhich is contraryto thecommon academic view thatthe militaryregards anything nonmaterial as inconsequential,if notnonexistent. In fact,nations are rarely literallydestroyed in a war,and victory does notautomatically go to theside that inflicts more deaths than itsuffers. Realismin World Politics 983 statesare satisfiedwith the status quo; conciliatorypolicies, effective under those circumstances,will decreasethe state'ssecurity if othersare strivingto expand.36 This wouldnot be a problemif statesmencould tellwhether others were-or will become-expansionist.But they cannot, in part because realism and other theories of foreignpolicy offer insufficient guidelines on thisscore. It is thereforenot sulprising thatstudents of security policy have been quick to see thatrealism needs to be supple- mentedby an understandingof theideas thatdecision makers use to guidethem to theirgoals.37 Subjectiveas some securityinterests are, realismargues that their importance meansthat they provide the crucialcontext for everything else. Althoughconve- nienceoften dictates bracketing security politics when they remain constant in a periodwhen economic behavior changes,38 I doubt if manyforeign economic policies andoutcomes are untouchedby broadsecurity concerns. Certainly it wouldbe foolishto tryto explainthe economic relations among the advanced industrialized countriesafter 1945 withouttaking into account the Cold War.39Indeed, the need to bolsterthe strength of its allies required the United States to give much consideration totheir economic needs, especially if it was tolimit their trade with the Soviet Union orChina. U.S. policiesto alleviatepoverty and increasestability in theThird World (misguidedor not)also cannotbe seen apartfrom the perceived fear that only eco- nomicprogress could forestall revolutions and the establishment of anti-Americanif notpro-communist regimes. Americansecurity policies also conditionedother countries' economic relations witheach other.European economic integration was facilitatedif not made possible by theAmerican security guarantee that assuaged British and Frenchfears of Ger- manyand allowed these countries to developan unprecedenteddivision of labor. The expectationof a continuedU.S. presencemeant that they did not have to worry about goingto warwith each otherin theforeseeable future and so did nothave to behave in typicalrealist fashion. The otherside of thiscoin is thateconomic relations are oftenset with at leastone eye on theirimplications for security. Recent progress in Europeanintegration is notunrelated to theunification of Germanyas all parties- includingcurrent German elites-want to see thatGermany will not have the incen- tivesor the easy ability to menaceits neighbors in thefuture.40 (This is notto saythat

36. Jervis1976, chap. 3. 37. The situationis different,at least in degree,in IPE: thepursuit of wealth,either by subnational actorsor by states, involves less needto estimatehow others will behave and fewer difficult choices about whichinstruments toemploy. Differences over how to pursue wealth often are easier to trace to differences in interest,and the sources and validity of beliefsabout how theeconomy functions tend to be taken forgranted by actors and observers.Thus, the discovery of theimportance of ideas comesas moreof an innovationin IPE; see Odell 1982;Rothstein 1984; Goldstein 1993; Ikenberry 1993a; andHalpern 1993. For a culturalexplanation of political and economicunderdevelopment, see Harrison1985. For an inter- estingdiscussion of economic cultures, see Rohrlich1987. 38. See, forexample, Keohane 1984, chap. 3, 137. 39. See, forexample, Pollard 1985; Jentelson 1986; and Gowa 1994. 40. A highGerman official explained his country'ssupport for an expandedNATO in similarterms: "We wantedto bind Germany into a structurewhich practically obliges Germany to take the interests of its neighborsinto consideration. We wantedto give our neighbors assurances that we won'tdo whatwe don't 984 InternationalOrganization evenan extremelyhigh degree of economicintegration guarantees continued unity andpeace, as thedivision of Czechoslovakia, the disintegration ofthe Soviet Union, andthe civil war in theformer Yugoslavia remind us.)

Realism,Change, and Cooperation

Realism'sassumptions lead to a focuson continuity,and John Ruggie has arguedthat thisapproach, particularly in securitystudies, has been egregiousin its failureto recognizethe possibilities for "epochal" change.41 But thisis notthe whole picture. Muchdepends, of course,on thekind and degreeof changethat one is lookingfor. Realistshave arguedthat the last half-century has witnessedas manyas threeenor- mousshifts in internationalpolitics-changes, furthermore, that most of themhave seen as makingthe world more peaceful. First, realists argued that the shiftfrom multipolaritytobipolarity in theaftermath ofWorld War II was extremelyimportant. Althoughdisagreeing about the consequences of thischange, they concluded that dangers,opportunities, and statepolicies were all effectedby thenew international structure.Ironically, it was Waltz,who placed greatstress on continuity,who most persuasivelyargued that bipolarity was unprecedented(contrary to thepreviously prevailingview that the pre-1914 system was bipolar)and that it made the world, or at leastthe major states, much safer. To thosewho believed that the bipolarity made theworld more war-prone because the great powers were tied to theirsmaller allies forwhom they might have to fight,42Waltz and those who reasoned similarly replied thatthis dynamic actually characterized multipolarity and that under bipolarity small allies could defectat will because the superpowersrelied on internallygenerated resources,thereby giving them unprecedented freedom of action.43 The substanceof thedebate is importantfor theory and policy,but all thatis crucialhere is thatboth sidesagreed that bipolarity significantly changed world politics. Realistssimilarly differed on whethernuclear weapons made the world more or less safe,but again agreedthat they brought important changes in statepolicies, bargainingtactics, alliance relationships, and opportunities to changethe status quo. In theend, most scholars came to theconclusion that nuclear weapons decreased the chanceof warand coercivechange at thecenter of theinternational system (but not in theperiphery). Bernard Brodie and his colleaguessaw thisas earlyas 1946,arguingthat atomic bombs were "absoluteweapons" that radically changed the funda- mentalcharacter of militarypower, which had previouslybeen relative.44 With rea- intendto do anyway"(quoted in JanePerlez, "Blunt Reasoning for Enlarging NATO: Curbson Ger- many,"New YorkTimes, 7 December1997, 18; see also Feldstein1997, 68-69). Forthe worries of British, French,and Sovietleaders at thetime of Germanunification, see Zelikowand Rice 1995, 137-38,345; Maier1997, 249, 252; andPowell 1992,235. 41. Ruggie1993, 143. 42. See, forexample, Deutsch and Singer 1964; and Hoffmann 1968, chap. 2. 43. See Waltz1979; andSnyder and Diesing 1977, chap. 6. 44. Brodie1946. This point of viewnever convinced decision makers, however. Some U.S. presidents feltthat nuclear weapons had made their country less ratherthan more secure and soughtto abolishthem, Realismin World Politics 985 son,then, many realists referred to the "nuclear revolution." As withbipolarity, what changedhere was theexternal situation: not what states sought, the prevailing norma- tiveprinciples, or theideas people held, but rather incentives, especially the punish- mentsthat statesmen believed would be incurredby dangerousand expansionist policies.45 Morethan this is at workin thethird candidate for major change that realists-as welland somenonrealists-have seen, which is thedecline of war among the devel- opedstates and even the creation of a securitycommunity among them.46 The fearof punishmentoperates here as wellin thateveryone realizes that a majorwar, even if it did notinvolve nuclear weapons, would be extraordinarilydestructive. But rewards maybe equallyimportant, since states can gainmuch of what they want by peaceful means.Of course, high levels of economic interaction would be viewedwith alarm if statesfeared that others would use theirincreased strength to menacethe stateor mightthreaten to severthe valuable ties in theevent of a dispute.Thus, a reciprocal relationshipexists between expectations of peace and the developmentof a high degreeof economic integration. But this dynamic is notforeign to realism. Thiscannot be saidof the third element that many people, myself included, see as creatingthe current security community. This element involves a changein outlooks and evenvalues among general populations, elites, and nationalleaders. Rabid and competitivenationalism has greatlydeclined, war is seenas a brutalnecessity if not a crimerather than a gloriousactivity, and controlof historicallydisputed territories such as Alsace and Lorraineis of greatlydecreased concern, in partbecause the developedcountries are democraticand sharemost values. Realism cannot readily explainthese developments. It can arguethat they were brought about at leastin part bythe increased costs of war, but the causal links involve psychological processes of attitudechange outside the focus of mostinternational politics theories. Realism is notalone in strugglinghere, however: changes in values are beyondthe scope of rationalism;and though they are, in principle, central to constructivism, itis notclear whetherthis approach can providemore than a post hoc redescription.Classical liberalismmight be morerelevant. Even ifchanges in valuesdo notfit well withrealism, the other shifts discussed in theprevious paragraphs do. Althoughthis analysis cuts against the most dramatic and far-reachingclaims for continuity, it shows that realism not only is compatible withsignificant alterations in internationalpolitics but also pointsto thepowerful motorsof change in whatstates can do tohelp and, especially, to hurt others. A focus on thethreat and use offorce does notimply that behavior remains uniform. albeiton termsthat strongly favored their side. (Gorbachev and Reagan were abolitionists, which was one of thebonds between them.) Furthermore, almost no statesmen,with the exception of Gorbachevand at timesKhrushchev and Eisenhower,thought that the weaponswere absolute.Instead they felt it was necessaryto seekgreater capability than the other side and pursued targeting doctrines that were designed to gainmilitary advantage or at leastto denysuch advantage to the adversary. 45. Someanalysts argued that modern technology would have produced this result even without nuclear weapons;Mueller 1989. 46. See Mueller1989; Van Evera 1990;Jervis 1991; and Singer and Wildavsky 1993. For dissents, see Mearsheimer1990; and Layne 1993. 986 InternationalOrganization

Realism similarlydoes not implyunremitting conflict. To conceiveof internationalpolitics as a Hobbesianstate of naturemeans not that warfare is constant,but onlythat it is always a possibilityand thatactors understand this. Although the anticipationof conflictmay make it morelikely, it can also lead actorsto takemea- suresto reducethe danger. Three facets of realistthought are particularlyrelevant hereand lead to policyprescriptions. First, realism is well knownfor arguing that powermust be musteredin orderto reachthe state's possible goals; Morgenthau's "interestdefined in termsof power"is itsmost familiar formulation.47 But forreal- istsit is equallyimportant that goals have to be trimmedto fitwithin the possibilities createdby the configuration ofpower. Thus, the first edition of Carr's Twenty Years' Crisisapplauded Britain's appeasement policy. We now knowthis to have been in errorbecause Hitlercould nothave been satisfiedshort of worlddomination, but realiststatesmen faced by more reasonable adversaries have been able to avoidcon- flictby appeasing them.48 Justas understandingthe limitsof the state'spower can reduceconflict, so in protectingwhat is mostimportant to themstates must avoid the destructive disputes thatwill result from failing to respectthe vital interests of others. Realists have long arguedthat diplomacy and empathyare vitaltools of statecraft;conceptions of the nationalinterest that leave no roomfor the aspirations and values of others will bring ruinto thestate as well as to itsneighbors.49 Realismalso can speakto theconditions under which states are mostlikely to cooperateand the strategies that actors can employto foster cooperation. This line of theorizingis sometimesassociated with neoliberalism, but the two are hard to distin- guishin thisarea.50 Making a distinctionwould be easy if realismbelieved that conflictwas zero-sum,that actors were always on thePareto frontier. This conclusion perhapsflows from the view of neoclassicaleconomics that all alTangementshave evolvedto be maximallyefficient, but realists see thatpolitics is oftentragic in the senseof actorsbeing unable to realizetheir common interests. Although "offensive realists"who see aggressionand expansionismas omnipresent(or whobelieve that securityrequires expansion) stress the prevalenceof extremeconflict of interest, "defensiverealists" believe that much of internationalpolitics is a Prisoners'Di- lemmaor a morecomplex security dilemma.51 The desireto gain mixes with the need forprotection; much of statecraftconsists of structuringsituations so thatstates can maximizetheir common interests. The ever-presentfear that others will take advantage of the state-and the knowledgethat others have reciprocalworries-leads

47. Morgenthau1978, 5-8. 48. See Kennedy1983, chap. 1; andSchroeder 1976. 49. See Morgenthau1978; Calleo 1978; Kennan1967, 127-30; and Kennan 1968. 50. Keohane1984, 67. This is notto say thereare no disputesbetween these two schools of thought: mostcentrally, they disagree about the importance and fungibilityof force,the typical balance between conflictingand commoninterests, and theextent to whichcooperative arrangements among states can unintentionallyalter their preferences; see, forexample, Keohane and Nye 1977; Baldwin1989, chap. 7; Baldwin1993; Nye 1990,chap. 6; andArt 1996. 51. For theformer, see, forexample, Gray 1992; and Schweller1996; forthe latter, see Schellingand Halpeiin1961; Jervis 1978; Waltz 1979; andSnyder and Jervis forthcoming. Realismin World Politics 987 diplomatsto seek arrangementsthat will reduceif not neutralizethese concerns. Evenif international politics must remain a Prisoners'Dilemma, it can oftenbe made intoone thatis morebenign by alteringthe pay-offs to encouragecooperation, for example,by enhancingeach state'sability to protectitself should the other seek to exploitit and increasing the transparency that allows each to see whatthe other side is doingand understandwhy it is doingit. The knowledgethat even if othersare benigntoday, they may become hostile in thefuture due to changesof mind, circumstances,and regimescan similarlylead decisionmakers to createarrangements that bindothers-and themselves, as previouslynoted. But,in parallel to the earlier discussion of the kinds of changes that realism cannot explain,deeper forms of cooperationexist that are moreproblematic from this perspective.Thus, though realism can accountfor the dramatic growth in cooperation whenthe balance of power(narrowly conceived) is transformedinto a concertre- gime,the more profound alterations in attitudes,identities, and valuesthat perhaps surroundedthe original Concert of Europein 1815 is beyondthe reach of thisapproach.52The basis and formsof cooperationafter the Napoleonic Wars may have restedon conceptionsof commoninterests and sharedresponsibilities that are alien to realism,although realists would not be surprisedthat these beliefs and arrange- mentseventually decayed.

Extensionsof Realism

Much criticismhas been leveledat realismfor the linkedassumptions that states existin a conditionof anarchyand that they follow their narrow self-interests. Inter- estingly,rationalism as appliedto international politics and the current work in Ameri- can and comparativepolitics that inspired it adopta similarapproach. As Helen Milner'sarticle in thisissue shows,these studies see domesticactors as uninhibited by norms,worried (with good reason)that unenforceable promises and threatswill proveempty, and seekingways to avoid the suboptimaloutcomes that can result fromunnecessary conflict. Thus, many recent arguments about the U.S. Congressare morefamiliar to studentsof internationalpolitics than to thosewho had seen this institutionas governedby rules serving the interest of theinstitution as a wholeand populatedby individualsseeking the good ofthe collectivity, ifnot of the country at large.Of course,this perspective is notentirely new in thefields of domestic politics and society.But the"amoral familism" that Edward Banfield saw as makinglife in an Italianvillage so miserablewas consideredpathological, and RichardNeustadt's discussionof presidentialpower in termsof theinterests the president did not share withothers was unsettling,if insightful.53Now thiskind of behavioris takenas normativein one ifnot both senses of the term.

52. See Schroeder1992 and 1994; and Jervis1992a. For a realistview of theConcert, see Kagan 1997. 53. See Banfield1958; and Neustadt 1960. 988 InternationalOrganization

Thisperspective not only draws the subfields of political science together but also pointsto a basic tensionin traditionalrealist thought. There, national leaders are ruthlessfor their states but selflessas individuals.The utilitiesthey maximize are thoseof thecountry as a whole,not of themselvespersonally; they respect no re- straintson theacts necessary for their countries but never put their own interests first. People of thistype can existand perhapshave come to power.Realists are fondof quotingCavour's remark to a colleaguethat "if we wereto do forourselves what we have donefor our country, we shouldindeed be verygreat rogues,"54 but they have notexplored why he didnot advance his personal interests in this way. Although such behavioris compatiblewith "thin" rational choice theory that does notspecify people's utilityfunctions, rationalism does directus tothe idea thatpolitical leaders, like manypeople, will put their own interests first. This perspective leads us toexpect that leadersare prone to exploit their societies for their material good and to adopt foreign policiesthat maximize their own power. Such behavior is antitheticalto theprecepts of realism,but follows from many realist assumptions and indeedcan be seen as removinga majorinconsistency in realism'sconception of human nature. Realistspay littleattention to theformation and maintenanceof ethnicand na- tionalidentity, a topic crucial not only for current world politics but also formuch of thepast. This observation,however, does notmean that realism is of no use here. Indeed,consistent with the logic in theprevious paragraph, many scholars are now stressingthe extentto whichidentities are manipulated,if not created,by self- servingelites who see thatpower is tobe gainedby convincing a large segment of the populationthat they form a community-one,furthermore, that is threatenedby people who are differentfrom them. Realism points to thereciprocal relationship betweenidentities and conflict, arguing that conflict both grows out of and stimulates theperception of differences among groups. Thus, a realistwould not be surprisedby thefact that the breakdown of the state is as mucha cause as a consequenceof ethnic conflict.Not only does theabsence of central authority mean that people and groups are less protectedagainst their neighbors, but, through a securitydilemma, fear as wellas rapaciousnessand rationality as wellas psychologylead to thestrengthening ofavailable group identities.55 Theprocess, of course, is notentirely one of free choice. Identities are often forced onpeople, as inNorthern Ireland when Catholics are attacked because they are Catho- lic and Protestantsbecause they are Protestant.As a nineteen-year-oldsaid recently in Sarajevo,"I was sittingin myclassroom the other day and the teacher handed out a formwhere we hadto writedown whether we werea Serb,Muslim, or a Croat.We weretold that we wouldbe segregatedinto different classrooms according to our ethnicity.It's notwhat any of us askedfor."56 Social psychologistshave long known thatperceptions-and misperceptions-of what people have in commonoften grow

54. Quotedin Palmerand Palmer 1976, 37. 55. For theapplication of thesecurity dilemma to civilconflicts, see Posen 1993a; Walter1994; and Walterand Snyder forthcoming. 56. Quotedin ChrisHedges, "In Bosnia's Schools,Three Ways Never to Learnfrom History," New YorkTimes, 11 November 1997, 4. Realismin World Politics 989 outof conflicts as internalunity is gainedby seeing others as theOther.57 Groups and identitiescan be createdthrough such processes: the "Bosnians" did notexist until therecent war, which also led many"Muslims" to becomeMuslims by turningto Islam. Similarly,recent scholarship argues that the identityof Indiansas having "red" skincame notfrom European colonialists but from Indians who were in con- flictwith them or whowanted to setthemselves apart from black slaves. The Native Americanswere not passive recipients of a colordesignation, but formed it through struggleswith others.58 Related processes of differentiation help to explain why black femalehigh school students smoke much less thaneither their white female or their blackmale counterparts: they see smokingas whiteand male.

InternationalPolitical Economy

Oddly,the articles in this issue say little about the state of the IPE subfield.59Striking by theirabsence are the old but stillcentral topics of the existenceand shape of mutualcausation between economic intercourse and peacefulrelations;60 the linkages amongeconomic and politicalissues; the determinants of thedegree of open- ness of internationaleconomic systems;61 the causes and prospectsof thecurrent consensusamong less developedcountries that economic development is positively associatedwith exposure to theinternational economy rather than negatively associated,as once believed;and therelationships between states, both rich and poor,on theone hand,and nonstate actors, especially labor and multinational corporations, on theother. How muchprogress have we made in understandingthe crucialquestions of whether,when, and how economic frictions lead to politicalconflict and vice versa? It was once commonlyaccepted that economic competition drove allies apartand heightened,if notcreated, deadly enmities: "Nations which act as enemiesin the marketplacecannot long be friendsat thecouncil table," as a highgovernment official putit at theend of World War II whenexplaining why the United States needed toreconstruct the international economic system.62 Is thiscorrect? Is therelationship conditionedby people's economic and political beliefs? The sourcesof stabilityand

57. See Sherifand Sherif 1953; Sherif 1966; and Mercer 1995. The basicarguments from sociology are Simmel1955; and Coser 1956. 58. Shoemaker1997. 59. I am also struckby the claim of Katzenstein,Keohane, and Krasner in theirarticle in thisissue that "ideas originallydeveloped in IPE havebeen redeployed to a widerrange of issues, including questions of nationalsecurity." Although this is truefor some concepts recently borrowed from economics, to itscredit IPE has learnedmuch from other subfields of internationalpolitics. Thus, recent discussions of salient solutions,issue linkages,credibility, commitments, and reputationfollow the treatments by studentsof diplomacyand security. Of coursemuch of the credit here goes to an economistwho supplied many of the keyconcepts: Schelling 1960 and 1966. 60. Muchof what recent literature there is on someof these subjects is summarizedin Mastanduno this issue.See also McMillan1997. 61. Discussedin Webb and Krasner 1989; Lake 1993;and Keohane 1997. 62. Clayton1945, 979. 990 InternationalOrganization changein theinternational political economic system also havereceived less atten- tionthan was thecase in the 1970s and 1980s.63Ruggie's important application of Karl Polanyi's"embedded liberalism," though often cited, has notbeen adequately developeddespite some treatment in thecontext of globalization.64 It is also strikingthat we do nothave manyconstructivist accounts of European integration,which is an obviousarena in whichto explorethe causes and effectsof changingidentities.65 Similarly lacking are discussionsof theimplications for indi- viduallives and nationalpolicies of thechanging value and indeedthe meaning of wealth.Ronald Inglehart's research on shiftsin Europeans'values could be a good startingpoint,66 but it has notattracted much attention from constructivists orthe IPE community.More fundamentally,constructivist IPE theoristshave notchallenged economistson theirown ground by exploringwhat it is thatindividuals and collec- tivitiesseek to maximize,how economicwell-being is construed,and theinterrela- tionshipsamong capitalist ideas, individual identities, and the activities of economic actors. To putit bluntly, the IPE subfield,after a marvelousperiod of development in the 1970sand 1980s,seems to be stagnating.At the start, realism, liberalism, and Marx- ism vigorouslycontended in theprocess of carryingout well-groundedempirical research.67This flow has slackenedwithout being replaced by sustainedconstructiv- istand rationalistaccounts. Indeed, most applications of constructivismor rational- ismhave been to securityor generalinternational politics, with less attentionto IPE. The internalworkings, incentives, self-identities, and culturesof theorganizations involvedin internationalpolitical economy similarly have been studiedless than thosein diplomacyand security.68Also noteworthyis that economists more so than politicalscientists have shapedthe public debate about international economic policies.

The Futureof IPE, War,and Realism

It willbe hardto constructpowerful theories of theinternational economy without keepingin mindthe political relationships among countries, the ways economic ties can ameliorateconflict or createexploitable vulnerabilities, the actors' expectations aboutwhat alliances are likely to formand how long they are likely to last,and their beliefsabout which technologies will be mostuseful if armedconflict occurs.69 In

63. See, forexample, Krasner 1983b; and Keohane 1984. 64. Ruggie1983b. For Ruggie's more recent analysis, see Ruggie1996b, chap. 5-6. 65. Forwork partially along these lines, see Englemannet al. 1997;and Spirtas 1998. 66. The mostrecent study is Inglehart1997. 67. See, forexample, Gilpin 1975 and 1987; Katzenstein1984 and 1985; Krasner1976, 1978, and 1985; Keohaneand Nye 1977;and Keohane1984. 68. For examples,see Demchak1991; Posen 1984; Sagan 1993;Avant 1994; Rosen 1991 and 1996; Legro 1995;and Kier 1997. 69. Of course,the pattems here may not be constant,and we needto understandthe changes. In many eras,for example, states sold armsto theirpotential enemies and carTied on economicintercourse during warfare;see Stevenson1996, chap. 1; andKennedy 1983, chap. 3. Realismin World Politics 991 otherwords, the study of IPE mustremain political, and internationalpolitics has alwaystaken place in theshadow of war. But,as notedearlier, I do notthink this is anylonger true for the developed countries.The consensusamong scholars and, more importantly, elites is thatthe most powerfulstates will not fight each other.This situationrepresents a truly revolution- arychange in world politics and makes particularly relevant the path-breaking analysis ofRobert Keohane and Joseph Nye.70 This does notmean the end of conflict and thestruggle for advantage, let alone theend of thestate.] Thus,I do notbelieve it meansthe end of realism; although, since this approach stressed the pervasive influence of thefear of inter-statewar, it will have to be reshapedif it is to explain,let alone guide,a worldin whichsecurity threats are of a verydifferent nature and probablymuch less important.72 It will also be a challengefor other schools of thought to explainand predicthow stateswithin the security community will managetheir relations.Particularly interesting and importantis thequestion of how and whether thesecurity community will be transformedinto a real community-forexample, whetherthe European countries will seek or be able to maintaintheir separate identi- tiesif they are perceived to lose theability to manage their individual economies and, conversely,whether the belief that one countryis sufferingso thatanother can pros- per will decreasethe EU's unityeven if war remainsunthinkable. We shouldnot, however,neglect the relations among countries that are notin thiscommunity (that is, mostof theworld), the relations between states in thecommunity and thoseout- sideit, and the possibility that armed conflict elsewhere will influence relations within thecommunity. In all theseareas the relations between wealth and power are likely to remaincentral.

I wonderwhat the next fifty years of international politics will bring; I also wonder whatwill appear in thecentennial issue of 10 and whatthe relationship between the two will be. The studyof internationalpolitics will be impoverishedif it is totally divorcedfrom contemporary events and hopesand fearsfor the future, but if it is to mature,it will have to developsome distance from them.

70. Keohaneand Nye 1977.See also Rosecrance1986. 71. Forthe argument that the changed international environment will affect different kinds of statesin differentways, see Desch 1996. 72. Fora discussionof realism in sucha world,see Kapsteinand Mastanduno 1998; and Waltz 1998.