The Promise of Institutionalist Theory Author(s): Robert O. Keohane and Lisa L.

Martin Reviewed work(s): Source: International Security, Vol. 20, No. 1 (Summer, 1995), pp. 39-51 Published by: The MIT Press Stable URL: http://www.jstor.org/stable/2539214 . Accessed: 20/11/2011 11:43
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 support@jstor.org.

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

http://www.jstor.org

Ins Theory Institutionalist Th~eory

0. The Promise of Robert Keohane t

~~~andMartin LisaL.

JohnJ.Mearsheimerhas sharpenedthe theoretical issues dividingrealistfrom institutionalist theory, and forthisservicewe are grateful. are also pleased We thathe has read theinstitutionalist literature thoroughly correctly so He asserts thatliberalinstitutionalists treatstatesas rationalegoists operatingin a world in which agreementscannot be hierarchically enforced, and thatinstitutionalists only expect interstate cooperationto occur if states have significant common interests. Hence institutionalist theorydoes not espouse the Wilsonian conceptof collectivesecurity-which Charles and Clifford to Kupchan refer as "ideal collectivesecurity"-critiqued so well by I.L. Claude thirty years ago.1 Nor does institutionalism embrace the aspirationsto transform international relationsput forwardby some criticaltheorists. Like realism,institutionalist theoryis utilitarian and rationalistic.2 However, ProfessorMearsheimer'sversion of realismhas some ratherserious flaws. Among them are its penchant for assertions that turn out to be incorrect; propensity privilegeits own viewpoint,so thatin the absence its to of decisive evidence eitherway it invariablyseems to prevail; its failureto explicatethe conditionsforthe operationof its generalizations; and its logical We contradictions, escaped only throughverbal sleight-of-hand. will begin by pointing out such errorsfromhis own recent articles in this journal, then
Robert Keohane Stanfield 0. is Professor International of Peace,HarvardUniversity, author After and of Hegemony: Cooperation and Discord in the World Political Economy (Princeton University Press, 1984). Lisa L. Martinis John LoebAssociate L. Professor Government, of HarvardUniversity, author and of Coercive Cooperation: Explaining MultilateralEconomic Sanctions (Princeton University Press, 1992). The authorsthankMarc Busch, Chris Gelpi, Andrew Moravcsik,and Celeste Wallanderfortheir valuable commentson an earlierversion of this essay. 1. Inis L. Claude, Powerand International Relations (New York:Random House, 1962). Mearsheimer relies heavily on Claude's critiquein his own discussion of collectivesecurity. 2. See Richard K. Ashley,"The Povertyof Neorealism,"International Vol. Organization, 38, No. 2 (Spring 1984),pp. 225-286. Ashley included Robert0. Keohane as one of the "neorealists"whose "orreryof errors"he rejected.The fact that Mearsheimercriticizedinstitutionalism and critical theoryin the same articleshould not,therefore, lead readersto believe thatthereis an intellectual betweenthesetwo schools of thought. affinity theorists such However,thework of "constructivist" as Alexander Wendteloquentlymakes a numberof argumentsthatmany institutionalists would accept.
International Security, 20, No. 1 (Summer 1995), pp. 39-51 Vol. ? 1995 by the Presidentand Fellows of Harvard College and the MassachusettsInstitute Technology. of

In his usual direct way,

39

International Security 20:1 | 40

We consider the illusory examine his major claims about institutionalism. divide between security and economic issues, the muddled question of "relative gains," and empiricalwork (admittedlyin its early stages) that provides We conclude that evidence of the significanceof internationalinstitutions. and thatit is a worthytask of social science to institutions sometimesmatter, discover how, and under what conditions,this is the case.

TheFallaciousLogicofRealism
the decline of NATO: Five years ago Professor Mearsheimerforecast imminent "It is the Soviet threatthat holds NATO together. Take away that offensive and theUnited Statesis likelyto abandon theContinent, whereuponthe threat At defensivealliance it headed for forty years may disintegrate."3 the same time,he predictedthat "the EC is likely [due to the end of the Cold War] to grow weaker, not strongerwith time."4Yet now that both NATO and the now the European Union (EU), are expanding their European Community, for memberships,and hardly in decline, he abandons specificity the equally falsebut more difficult falsify to that"institutions have minimal generalization stability prospectforpromoting influence statebehaviorand thushold little on in a post-Cold War world."5 institutions matter. Professor Mearsheimerdemands proofthatinternational have us recently Yet he begins his articleby reminding thatmajorgovernments he institutions; could have added been emphasizingthe value of international materialand reputationalresourcesin NATO, the that theyinvest significant and EU, and also in organizationssuch as the General Agreementon Tariffs to Trade (GATT, recentlystrengthened create the World Trade Organization) and the NorthAmericanFree Trade Agreement (NAFTA). Not all international but some do. How institutions command such resourcesfromgovernments, are we to account for the willingness of major states to invest resources in are lacking in sigif expanding internationalinstitutions, such institutions nificance? Mearsheimersuggests that the answer lies in an ideological blindness of Americanpolicymakers, whose hostility toward realismdrivesthemto to the more congenial institutionalist framework(pp. 47-49). It is difficult
3. John Mearsheimer, in the J. "Back to the Future:Instability Europe after Cold War,"International Security, 15, No. 1 (Summer 1990), p. 52. Vol. Vol. 4. JohnJ.Mearsheimer, Back Security, "Correspondemnce: to the Future,Part II," International 15, No. 2 (Fall 1990), p. 199. Vol. Security, 5. John Mearsheimer, "The False Promiseof International Institutions," International J. to 19, No. 3 (Winter 1994/95),p. 7. Subsequent references thisarticleare in parenthesesin the text.

ThePromise Institutionalist of Theory 41 |

square this assertionof a collectivedelusion with the dominantrole of realist theoryin policy discussions,or with realism's own preceptsabout the forces instiin that drive state behavior.In lightof states' investments international tutions,it is fairto turnMearsheimer'squestion around: could we not legitiare deluded or mately demand evidence eitherthat leaders of governments that NATO and the EU are designed to deceive unsophisticatedobservers? Mearsheimerassumes that his view is privileged,in the sense that we must convincingevidence is presentedforan accept realismunless overwhelmingly institutions make alternative view; but thefactthatstatesinvestin international this stance quite problematic. in and realism differ a numberof otherrespects,one of the Institutionalism most significant which concernshow theyapproach social science.A central of theory-ratherthan as rhetoricfaultof Mearsheimer'srealismas a scientific is thatthe conditionsforthe operationof its "grimpictureof world politics" Realism is repletewith global generali(p. 9) typicallyare not well-specified. zations, lacking qualificationsabout the conditionsunder which theymay be valid. Let us consider two examples fromMearsheimer's own article.First, Mearsheimer writes that "states in a realist world . . . must be motivated primarily relative gains concerns when considering cooperation" (p. 12, by emphasis added). But he lateradmits thatthis propositionmay be false when the threat aggressivewar is low-for instance,when defensivetechnologies of nuclearforces)are prevalent(pp. 23-25). Second, (such as secure second-strike in Mearsheimer'srealistworld, "everystatewould like to be the most formidable militarypower in the system" (p. 12). But since no one thinks that or Britain actuallyseeks to become "the Switzerland,Argentina, contemporary most formidablemilitarypower," what Mearsheimerpresumably means to capabilitiesalways pursue this goal. Even argue is that states with sufficient thisstatement oftenfalse:forexample,theUnited Statesduringtheinterwar is period could reasonablyhave expected to become the most powerfulstate in with such contradicthe world, but did not seek such a position.Confronted fromuniversal rhetoric post to tions and anomalies, realism typicallyretreats history, percephoc and ad hoc qualifications, takinginto account geography, tions,and domesticpolitics. in Institutionalism, contrast, seeks to state in advance the conditionsunder which its propositionsapply. Our theorymay therefore have less appeal to theoriesshould but purportedly scientific those who require simple "truths," As specifythe conditionsunder which the theoryis expected to hold a priori. benefits Mearsheimerindicates,when stateelites do not foreseeself-interested fromcooperation,we do not expect cooperationto occur,nor the institutions

International Security 20:1 | 42

that facilitatecooperation to develop. When states can jointlybenefitfrom cooperation, theotherhand, we expectgovernments attempt construct on to to such institutions.Institutionscan provide information, reduce transaction costs,make commitments morecredible, establishfocalpointsforcoordination, the and in general facilitate operationof reciprocity. seeking to specifythe By conditionsunder which institutions can have an impact and cooperationcan occur,institutionalist theoryshows under what conditionsrealistpropositions are valid. It is in this sense thatinstitutionalism claims to subsume realism. Realism's proclivity bold, unqualifiedgeneralizations for not only generates anomalies but gets its proponentsinto logical difficulties. Mearsheimerholds that "institutions have no independent effect state behavior" (p. 7); that on NATO is an institution 13); and that NATO played a role in preventing (p. World War III and helping the West win the Cold War (pp. 13-14). These propositionssound like a classicallyfallacioussyllogism,untilone recognizes that thereis an escape clause: "NATO was basically a manifestation the of bipolar distribution power in Europe duringthe Cold War,and it was that of balance of power, not NATO per se, that provided the key to maintaining who see instituon stability the continent" 14). But liberalinstitutionalists, (p. tions as rooted in the realitiesof power and interest, not argue thatNATO do could have maintained stabilityunder any imaginable conditions.What we in argue is that institutions make a significant difference conjunctionwith power realities.Institutions important"independently"only in the ordiare for the effects power and nary sense used in social science: controlling of it interests, matterswhetherthey exist. They also have an interactive effect, meaning that theirimpact on outcomes varies, depending on the nature of power and interests. Mearsheimeris forcedto admit the truthof institutional with regard to NATO, although for rhetoricalpurposes he shiftshis effects can ground to attacka view thatwe do not hold: thatinstitutions preventwar in regardlessof the structure which theyoperate. Hence Mearsheimer'sversionof realismis repletewith analyticalproblems. However, it is not our duty here to correct realism's copy-book.In the restof thisbriefresponse,therefore, focuson thepromiseof institutionalist we theory, and the researchdirections thatwe hope will help to realize thatpromise.

PoliticalEconomy Security theIssue ofRelative vs. and Gains

AlthoughMearsheimerhas provided an admirablesummaryof several aspects on of institutionalist his theory, version of our argumentrequires correction

ThePromise Institutionalist of Theory 43 |

two majorpoints.First, Mearsheimer assertsthatinstitutionalist theory based is on "theassumptionthatinternational politicscan be divided intotwo realmssecurityand political economy-and that liberal institutionalism mainly applies to the latter"(pp. 15-16). Althoughsome institutionalists have made this assertion,it is not the predominantview of the institutionalist and literature, we certainly not accept it. Secondly, contrast Mearsheimer'sassertion, do in to our focusis not exclusivelyon "cheating."Situationsof coordination, which in cheating is not a problem but distributional issues are serious, are equally important, althoughtheywere underemphasized(but not absent) in the early institutionalist literature.
THE PURPORTED SECURITY VS. POLITICAL ECONOMY DIVIDE

Mearsheimer's assertionthat institutionalism employs a "neat dividing line" to separate politicaleconomyfromsecurity in issues is surprising, view of the attentionthat he devotes to the volume edited by KennethOye, Cooperation UnderAnarchy. major argumentof Cooperation A UnderAnarchy thatinstituis tionalisttheorycan be applied to bothsecurityand political economy issues. As RobertAxelrod and Robert0. Keohane wrote: It has oftenbeen noted thatmilitary-security issues display more of the charones. Charles acteristicsassociated with anarchy than do political-economic has recently observedthatpolitical-economic Lipson, forinstance, relationships are typicallymore institutionalized ones. This does not than military-security mean, however,thatanalysis of these two sets of issues requirestwo separate analyticalframeworks. Indeed, one of the major purposes of the presentcollection is to show that a singleframework throwlighton both[emphasis can added].6 We share Mearsheimer'sview thatthereis no clean analyticalline between economic and securityissues, although we do not base our view on the overarchingrole of relative gains. Institutionalist theory should be highly applicable to security issues because its argumentrevolves around the role of in institutions providinginformation. argument pertinent realistsecuis to This which oftenrelyon worst-caseanalysis. Realists contendthat rityarguments, in an uncertain,anarchic world, states must assume the worst,particularly about others' intentions,when making policy choices. Worst-caseanalysis

6. RobertAxelrodand Robert0. Keohane, "AchievingCooperationUnder Anarchy: and Strategies Institutions," KennethA. Oye, ed.,'Cooperation in Princeton UnderAnarchy (Princeton: University Press, 1986),p. 227.

International Security 20:1 | 44

implies followingpolicies that do not maximize expected utilityforthe sake of avoiding terrible outcomes. But if one can secure more information, may it be possible to followpolicies thatmorenearlymaximizeutility.7 Realistwriters if fromKautilya on have stressedthe significance information of (intelligence); institutions can provide useful information, realists should see them as significant. The logic of institutionalist theoryis directlyapplicable to security problemsas realistsdefinethem. Hence, if Mearsheimermeant to offerus a "loophole" throughwhich to escape his criticism-thatinstitutionalist theory only applicable to non-secuis rityissues-we emphatically refuseto avail ourselves of his generosity the On we theorywill contrary, hope that,to use Axelrod's phrase, institutionalist gradually "invade" the studyof security issues, helping to explain variationin institutional formwithout denying the validityof many realistinsightsinto power and interests.
RELATIVE GAINS AND INTERNATIONAL COOPERATION

The conclusions we draw fromthe "relativegains" debate are different from those of Professor Mearsheimer. is true thatwhen only two statesexist and It will but theyhave perfectly conflicting interests, institutions not be significant, this point is obvious. Two issues are more significant: the conditionsunder 1) which relativegains are important; and 2) the role of institutions when distributional issues are significant-that when relativegains are at stake. is, It is importantto understand the great variation in the extent to which relativegains matter. The major lesson of the recentdebate on relativegains is that theirimportanceis conditional factorssuch as the number of major on actorsin thesystemand whethermilitary or advantage favorsoffense defense.8 Duncan Snidal has shown thatrelativegains are unlikelyto have much impact on cooperationifthe potentialabsolute gains fromcooperationare substantial, or in any contextinvolving more than two states.9A valuable aspect of the relativegains debate is thatit has made distributional and bargainingissues
7. See Celeste A. Wallander,"Balance and Institutions German-Russian in Relationsafter Security theCold War,"manuscript, Harvard University, 1994;Celeste A. Wallanderand Robert0. Keohane, "Toward an Institutional Theoryof Alliances,"paper preparedfordeliveryat the Annual Meeting of the International Studies Association,Chicago, Illinois,February22-25, 1995. 8. See David A. Baldwin, ed., Neorealism and Neoliberalism: Contemporary The Debate (New York: Columbia UniversityPress, 1993), p. 323, especially chaptersby JosephGrieco, Duncan Snidal, RobertPowell, and Robert0. Keohane. 9. Duncan Snidal, "RelativeGains and'the Pattern International of Cooperation,"American Political ScienceReview, Vol. 85, No. 3 (September1991), pp. 701-726.

ThePromise Institutionalist of Theory 45 |

more salient than they were in early neoliberal thinking,10 if the debate but becomes one of "whether"relativegains matter, thatvalue will be dissipated. We need instead to ask under what conditionssuch distributional conflicts are severe. What is the role of institutions when distributional issues are important? Contraryto the assertionthat institutionalist to theoryis irrelevant distributional issues, we argue thatdistributional conflict may renderinstitutions more important. understandthispoint,it is essentialto distinguish To between two problems that states face when they attemptto cooperate. They oftenworry about the potentialfor othersto cheat, as in a Prisoners'Dilemma. But they also face the problem of coordinatingtheir actions on a particular stable cooperative outcome (solving the problem of multiple equilibria, in gametheoretic terminology). Usually morethanone cooperativeoutcomeexists.The statesinvolved may not agree on which of theseoutcomesis preferred, each as has different distributional implications.Disagreementabout the specificform of cooperation is the principal barrierto cooperation in such coordination games. Unless some coordinating mechanismexists,statesmay fail to capture the potential gains from cooperation. Institutionsdo not provide the only possible coordinating mechanism.11 However, in complex situationsinvolving many states,international institutions step in to provide "constructed can focal points" thatmake particularcooperativeoutcomes prominent. Realists interpret the relative-gainslogic as showing that states will not cooperate with one another if each suspects that its potential partnersare can gaining more fromcooperation than it is. However, just as institutions mitigatefears of cheating and so allow cooperation to emerge,so can they alleviate fearsof unequal gains fromcooperation.Liberal theoryargues that institutions provide valuable information, information and about the distribution of gains fromcooperationmay be especiallyvaluable if the relative-gains logic is correct. Institutions can facilitate cooperationby helping to settledisand by assuring states thatgains are evenly divided over tributional conflicts
10. For development of arguments about the relationshipbetween international regimes and distributional problems,see JamesD. Morrow,"Modeling the Formsof International Cooperation: Distribution versus Information," International Organization, 48, No. 3 (Summer1994),pp. 387Vol. 423; and JamesFearon, "Cooperation and BargainingUnder Anarchy," manuscript, of University Chicago, 1993. 11. For example, Stephen Krasner has argued that coordinationproblems can be solved by the unilateralexerciseof power by the strongest state.Stephen D. Krasner,"Global Communications and National Power: Life on the PairetoFrontier,"WorldPolitics,Vol. 43, No. 3 (April 1991), pp. 336-366.

International 20:1 | 46 Security

time, for example by disclosing information about the military expenditures and capacities of alliance members. In our view the successfulfunctioning institutions of depends heavilyon the operation of reciprocity, both specificand diffuse.12States using strategiesof reciprocity engaged in exchangewith one anotherand so requireinformaare tion about the value of theirexchanges. Institutionalized reciprocity and distributionalconcerns are simply two sides of the same coin, reflecting the of and difficulties cooperatingin a system lacking centralized enforcement if pointingto the need forreliablesources of information statesare to achieve are gains fromcooperation.Far fromleading to the conclusionthatinstitutions not significant world politics,the relative-gains in debate has led us to understand yet another pathway throughwhich they substantiallyinfluencethe course of international relations.A crucial step in the institutionalist research can program will be to understand the conditionsunder which institutions provide the information necessary to serve as reliable solutions to distributional problems.

Empirical Work theImpactofInstitutions on

We agree withJohnMearsheimerthat "more empiricalwork is needed before a finaljudgmentis renderedon the explanatory power of liberalinstitutionalism" (p. 26). The point of a new theoryis to generate testable hypotheses: liberal institutionalism, like any other theory, only has value insofar as it generatespropositionsthatcan be testedagainst real evidence. Institutionalist theoryconceptualizes institutions both as independentand dependent variables: "institutions change as a resultof human action,and the on changes in expectationsand process thatresultcan exertprofoundeffects statebehavior."13 Institutional theoryhas a coherentaccount of both the creation of institutions and theireffects: institutions createdby statesbecause are of theiranticipatedeffects patternsof behavior.Early researchby institutionon alists focused on institutions dependentvariables,examiningthe conditions as under which theyare created.Recentresearchhas sought more systematically

Vol. Relations,"International Organization, 40, 12. Robert0. Keohane, "Reciprocity International in No. 1 (Winter1986), pp. 1-27. 13. Robert0. Keohane, International Institutions StatePower(Boulder,Colo.: Westview,1989), and p. 10.

ThePromise Institutionalist of Theory 47 |

for to demonstratethat institutions are sometimes significant political outcomes, and to determinethe conditionsunder which thisis the case.14 In view of this research program, it should be clear that evidence that institutions change in response to underlyingconditions is hardly a blow afterall, posits that international against institutionalist theory. That theory, and is institutions createdin response to stateinterests, thattheircharacter are structured theprevailingdistribution capabilities.The real empiricalissue by of is how to distinguishthe effects underlyingconditionsfromthose of the of betweeninstitutions institutions themselves.One resultof theinterdependence and underlyingforcesis thatresearchdesigned to isolate the impact of institutionsis difficult design and execute. Rarely, ever,will institutions to vary if while the "rest of the world" is held constant.Thus findingthe ideal quasisituationto testthe impact of institutions not possible. is experimental do However, these difficulties not make it impossible to test the argument thatinstitutions since changes in underlying conditionsand in institumatter, tions are not perfectly correlated.Hence it may be worthwhileto search for instancesin which underlyingconditionshave changed rapidlywhile institutions have remained relativelyconstant,or where similar structural changes confront institutional endowments.Anothertactic regions that have different The institutionalist may be to considerthe level of institutional variationitself. perspectiveleads us to expect patternedvariationin the types of institutions will constatesconstruct, since theyanticipatethatinstitutions constructed so strainthem.Analysisof institutional such as variationsin theinstitutionform, alization of alliances or in the legalizationof the international tradingsystem, should thereforeprovide valuable evidence for evaluating institutionalist theory. have only marginaleffects rendersits Realism's insistencethat institutions account of institutional creationincompleteand logicallyunsound, and leaves it withouta plausible accountof theinvestments thatstateshave made in such international institutions theEU, NATO, GATT,and regionaltradingorganias when states act rationally zations. Accordingto the preceptsof realisttheory, will theyconstruct institutions, althoughtheyknow thattheseinstitutions have
14. Since institutionalists not claim thatinstitutions do always have a major impact on outcomes, finding weak institutions hardly constitutesa refutationof institutionalist theory Hence the weakness of the International EnergyAgency during the 1979 oil crisis,describedby Keohane in Princeton UniverAfter Hegemony: Cooperation Discordin theWorld and Political Economy (Princeton: sityPress, 1984), is hardlythe damning evidence thatMearsheimerclaims.

International Security 20:1 148

no impact on patternsof cooperation.But what could be the rationalebehind devoting resources to structures that will make no difference? Rather than assertingthatinstitutions have no impact,realistsmust mean thatinstitutions have some effect otherthan that assumed by liberal institutionalists. Perhaps institutions satisfythe ideological demands of statesmen,or help to pacify inattentive publics. Whateverthe rationale,we challenge realiststo construct an account of institutional variationand effects thatcan be tested against the institutionalist alternative. The difference between realism and liberal institutionalismdoes not lie in whetherinstitutions independentor dependent are variables; it lies in contrasting understandingsof whyinstitutions created are and howtheyexerttheireffects. A number of recent studies establish institutional effects throughcareful empiricalresearch, guided by institutionalist theoryand recognizingpotential problemsof endogeneity and omitted-variable bias.15 Ronald B. Mitchellshows that on three different issues involving oil pollution at sea, whether states complied with institutional regulationsdepended on the nature of the rules. "Clear causal links unambiguously demonstrate that treaty rules independentlyinfluencedbehavior,with otherplausible factorscontrolledfor or absent."16 New rules on the kinds of tanks that ships are allowed to use, for example, have had a dramaticimpact on intentional dischargeof oil into the oceans. The European Court of Justice(ECJ) has also proven a fruitful ground for the study of institutional influence.Anne-MarieSlaughterBurleyand Walter Mattlishow how the ECJhas had an unexpectedly large impacton the politics of European integration, into legal issues withthe aid of transforming political transnationalnetworks of lawyers and judges.17 The ECJ has gone far to convertthe Treatiesof Rome intoa constitution the EU, with the resultthat for EU law now reaches deeply into the domesticlaw of memberstates.Geoffrey Garrett in and BarryWeingast, anotherstudyof the ECJ,show how it resolved problemsof multipleequilibriaforEU memberstatesby providingconstructed
15. On such issues see Gary King, Robert0. Keohane, and Sidney Verba,Designing SocialInquiry: Scientific Inference Qualitative in Research (Princeton: Princeton University Press, 1994). 16. Ronald B. Mitchell,Intentional Pollution Sea: Environmental Oil at Policyand Treaty Compliance (Cambridge,Mass.: The MIT Press, 1994). See also Ronald B. Mitchell,"Regime Design Matters: Intentional Pollutionand Treaty Oil Compliance,"International Organization, 48, No. 3 (Summer Vol. 1994), pp. 425-458. 17. Anne-MarieBurleyand WalterMattli,"Europe beforethe Court: A Political Theoryof Legal Integration," International Organization, 47, No. 1 (Winter Vol. 1993),pp. 41-76. (Anne-MarieBurley now goes by the name Anne-MarieSlaughter.)

ThePromise Institutionalist of Theory 49 |

focal points in coordinationproblems.18 These studies show that institutions have the wide range of effects attributed them by liberal institutionalists. to They change the incentivesfor states to cheat; they also reduce transaction costs,link issues, and provide focal points forcooperation. The institutionalist perspectivehas also been applied with success to the analysis of security regimes.JohnDuffieldhas consideredNATO as a regional to security regime.He findsthatNATO made an independentcontribution the U.S. commit"Long Peace" in Europe by drawingboundaries,demonstrating the ments and making them credible,and facilitating augmentationof NATO allies' military He capabilities."9 also findsthatthe stable norms and rules of NATO led to stability levels of conventionalforceswithinthe regimethat in cannotbe explained by structural theories.20 In Coercive of Lisa Cooperation, Martinshowed thatthe involvement international organizations in economic sanctions is stronglycorrelatedwith high does not establishcausality, she levels of cooperation.21 Since such a correlation also did qualitative work on several cases involving sanctions,includingEC sanctionsagainst Argentinaduringthe Falklands War.Mearsheimerconsiders the Falklands case in isolation fromthe restof this research,and dismisses it on as "less thana ringing endorsement liberalinstitutionalism" thegrounds for that concernsabout cheatingwere not involved (p. 25). In fact,Martin does to findevidence that states used the EC framework reduce fearsof cheating, in the form of taking advantage of the situation to profitfromtrade with institutionof came through Argentina.22 However,themajoreffect institutions a alized linkagesthatwould otherwisehave been nonexistent: linkagebetween of EC budget contributions and the sanctionsissue. Prevention cheatingis not the only mechanismby which institutions facilitatecooperation.By creating issue linkages,they allow for more effective retaliationagainst cheatersand also create scope formutually-beneficial exchanges. Furtherevidence for the

18. Geoffrey Garrettand BarryR. Weingast,"Ideas, Interests, and Institutions: Constructing the European Community'sInternalMarket,"in Judith Goldstein and Robert0. Keohane, eds., Ideas and Foreign Policy:Beliefs, Institutions, PoliticalChange(Ithaca, N.Y: Cornell University and Press, 1993), pp. 173-206. of 19. John Duffield, S. "ExplainingtheLong Peace in Europe: The Contributions RegionalSecurity Regimes,"ReviewofInternational Studies, Vol. 20, No. 4 (October 1994), pp. 369-388. 20. JohnS. Duffield,"International Regimes and Alliance Behavior: Explaining NATO Conventional Force Levels," International Organization, 46, No. 4 (Fall 1992), pp. 819-855. Vol. 21. Lisa L. Martin, CoerciveCooperation: ExplainingMultilateralEconomicSanctions(Princeton: PrincetonUniversity Press, 1992). 22. Ibid., p. 143.

International Security 20:1 | 50

EC's role in coordinatingsanctionscomes fromthe factthat outside the EC, the only othersignificant support Britainreceived came fromCommonwealth nationsand theUnited States.In the U.S. case, supportwas delayed untilafter the outbreak of war, in distinctcontrastto the behavior of EC members.23 Mearsheimer'sdismissal of international institutions implies thatlinkages are easy to forgewhen a statedesires cooperation,and thatcooperationis easy to coordinateeven withoutinstitutions, Britaindid not findeitherto be the yet resultsand othercase studies case. Even in isolationfromthe robuststatistical reportedin Coercive the the centralrole Cooperation, Falklands case illustrates of formalinternational in institutions enabling states to cooperate to impose multilateral economic sanctions.24 Institutions sometimes matterfor state policy,but we do not adequately understand in what domains they mattermost, under what conditions,and how theireffects are exerted. More researchon this subject,by students of world politicscritical institutionalist of theory well as by thoseworkingfrom as it,is essential,and will be most welcome.

Conclusion
the irrelevanceof international MearFar from demonstrating institutions, sheimer's characterization conflict world politicsmakes institutions of in appear essential if states are to have any hope of sustained cooperation,and of reaping its benefits. This necessityforinstitutions does not mean thattheyare always valuable, much less that they operate without respectto power and or interests, constitute panacea forviolentconflict, always reduce the likelia hood of war. Claiming too much forinternational institutions would indeed be a "false promise." But in a world politics constrainedby state power and hierarchicalgoverndivergentinterests, and unlikelyto experienceeffective will be institutions ance, international operating on the basis of reciprocity componentsof any lastingpeace.
refusedBritishpleas to impose sanctions,and took only minorsteps following 23. Japaninitially U.S. impositionof sanctions,much laterthan EC members. a 24. The Falklands case cannot be dismissed on grounds that,as Mearsheimerclaims, striking includingpublic protestsin The historicalrecordshows intenseconflict, deal was "not difficult." some countriesand challenges to the sittinggovernmentin others. The Thatchergovernment believed thatits survival was at stake in the Falklands War.While perhaps not a "core interest" survival is surely a fundamentalconcernof policymakersthat by realiststandards,government could impede cooperation.

ThePromise Institutionalist of Theory 51 |

is relations a promising researchprogramin international The institutionalist withits focuson theinformational role theory, one. The logic of institutionalist should respondto Mearsheimer's of institutions, appears solid. Institutionalists into theirmodels, considerations distributional criticisms betterintegrating by furtherspecifyingthe causal mechanisms by which institutionsexercise influence, and buildingon existingempiricalwork to provide moreconvincing evidence of institutional effects. Both the questions raised and the provisional during the relativelyshort life of this reanswers given by institutionalists, search program,indicate that these tasks may be rewarding.In comparison theory seems bright. withtheextantalternatives, promiseof institutionalist the