The flawed logic of democratic peace theory

8/12/2019 The flawed logic of democratic peace theory

1/19

The Flawed Logic of Democratic Peace Theory

Author(s): Sebastian RosatoSource: The American Political Science Review, Vol. 97, No. 4 (Nov., 2003), pp. 585-602Published by: American Political Science AssociationStable URL: http://www.jstor.org/stable/3593025.

Accessed: 18/10/2013 08:24

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 formsof scholarship. For more information about JSTOR, please contact [email protected].

.

American Political Science Associationis collaborating with JSTOR to digitize, preserve and extend access to

The American Political Science Review.

http://www.jstor.org

This content downloaded from 164.73.224.2 on Fri, 18 Oct 2013 08:24:00 AMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=apsahttp://www.jstor.org/stable/3593025?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/3593025?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=apsa


2/19

AmericanPoliticalScience Review Vol. 97, No. 4 November 2003

h e l a w e d o g i c o emocra t i c P e a c e T h e o r ySEBASTIAN ROSATO The University of ChicagoDemocratic peace theory sprobablythemostpowerful iberalcontributiono thedebateon thecauses of war and peace. In this paper I examine the causal logics that underpin the theory

to determine whether they offer compelling explanations for thefinding of mutual democraticpacifism. Ifind thatthey do not. Democracies do not reliably externalizetheirdomestic norms of conflictresolution and do not trust or respect one another when their interests clash. Moreover, elected leadersare not especially accountable to peace loving publics or pacific interestgroups, democracies are notparticularly slow to mobilize or incapable of surprise attack, and open political competition does notguarantee that a democracy will reveal private information about its level of resolve thereby avoidingconflict. Since the evidencesuggests that the logics do not operateas stipulatedby the theory'sproponents,there are good reasons to believe that while there is certainly peace among democracies, it may not becaused by the democratic natureof those states.

emocratic peace theory-the claim that democ- them.2An evaluation of democratic peace theory,then,racies rarely fight one another because they rests on answeringtwo questions. First,do the data sup-share common norms of live-and-let-live and port the claim that democracies rarelyfighteach other?domestic institutions that constrain the recourse to Second, is there a compelling explanation for why thiswar--is probably the most powerful liberal contribu- should be the case?tion to the debate on the causes of war and peace.1 If Democratic peace theorists have discovered a pow-the theory is correct, it has important implications for erful empirical generalization: Democracies rarely goboth the study and the practice of international poli- to war or engage in militarized disputes with one an-tics. Within the academy it undermines both the realist other. Although there have been several attempts toclaim that states are condemned to exist in a constant challenge these findings (e.g., Farber and Gowa 1997;state of security competition and its assertion that the Layne 1994; Spiro 1994), the correlations remain ro-structure of the international system, rather than state bust (e.g., Maoz 1998; Oneal and Russett 1999; Raytype, should be central to our understanding of state 1995; Russett 1993; Weart 1998). Nevertheless, somebehavior. In practical terms democratic peace theory scholars arguethatwhile there is certainly peace amongprovides the intellectual justification for the belief that democracies, it may be caused by factors other than thespreading democracy abroadwill perform the dual task democratic nature of those states (Farber and Gowaof enhancing American national security and promot- 1997; Gartzke 1998; Layne 1994). Farber and Gowaing world peace. (1997), for example, suggest that the Cold War largelyIn this article I offer an assessment of democratic explains the democratic peace finding. In essence, theypeace theory. Specifically, I examine the causal logics are raising doubts about whether there is a convinc-that underpin the theory to determine whether they ing causal logic that explains how democracies inter-offer compelling explanations for why democracies do act with each other in ways that lead to peace. Tonot fight one another. resolve this debate, we must take the next step inA theory is comprised of a hypothesis stipulating an the testing process: determining the persuasiveness ofassociation between an independent and a dependent the various causal logics offered by democratic peacevariable and a causal logic that explains the connec- theorists.tion between those two variables. To test a theory fully, A causal logic is a statement about how an inde-we should determine whether there is support for the pendent variable exerts a causal effect on a depen-hypothesis, that is, whether there is a correlation be- dent variable. It elaborates a specific chain of causaltween the independent and the dependent variables mechanisms that connects these variables andtakes theand whether there is a causal relationship between following form: A (the independent variable) causesB (the dependent variable) because A causes x, whichcauses y,which causes B (see, e.g., Elster 1989, 3-10). Inthe case at hand, democratic peace theorists maintainSebastian Rosato is Ph.D. Candidate, Department of Political Sci- that democracy has various effects, such as support forence, The Universityof Chicago,5828 South UniversityAvenue, peaceful norms of conflict resolution, which, in turn,Chicago, L 60637 [email protected]).I would like to thank Alexander Downes, John Mearsheimer, increase the prospect for peace.SusanPratt,Duncan Snidal, and three anonymous reviewers for their I adopt two strategies for testing the persuasivenesshelpful comments and suggestions and the Smith Richardson Foun- of the causal logics that underpin democratic peacedation for financial support. A previous version of this paper was theory. First, I take each logic at face value and askpresented at The University of Chicago's Program on InternationalPolitics, Economics and Security (PIPES).1 The democratic peace research programhas generated several ad-ditional empirical regularities. See, for example, Bueno de Mesquita 2 On correlation versus causation see Dessler 1991 and Waltz 1979,et al. 1999, 791. 1-13.

585

http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp


3/19

The FlawedLogic of Democratic Peace Theory November 2003

FIGURE. DemocraticPeace Theory'sCausalLogicsLogic Independent Causal Mechanisms DependentVariable Variable

Normative Democracy - Externalization - Trust and Respect - PeaceInstitutional Democracy - Accountability - Public Constraint -- PeaceInstitutional Democracy Accountability - GroupConstraint - PeaceInstitutional Democracy -- Accountability - Slow Mobilization - PeaceInstitutional Democracy - Accountability -+ No SurpriseAttack - PeaceInstitutional Democracy - Accountability - Information -- Peace

whether the hypothesized causal mechanisms oper-ate as stipulated by the theory's proponents (Georgeand McKeown 1985, 34-41; King, Keohane, and Verba1994, 226-28; Van Evera 1997, 64-66). In other words,does the available evidence support the claims that Acauses x, that x causes y, and that y causes B? If itdoes, then the theory must be considered compellingbecause, as mentioned above, it is widely agreed thatthere is strong correlational support for its main hy-pothesis. If not, there is good reason to be skeptical ofthe theory.Second, I use the logics to generate additionaltestable propositions about the effects of democracy onstate behavior. If we accept that A does cause x, thatx causes y, and that y causes B, then logical deductioncan yield other propositions that should also be true.These too can be checked against the historical record,and the theory will be strengthened or weakened tothe extent that they find empirical support. Before per-forming these tests, however, a brief summary of thecausal logics is in order.

CAUSALLOGICSNormative LogicProponents of the normative logic argue that one im-portant effect of democracyis to socialize political elitesto act on the basis of democratic norms whenever pos-sible. In essence, these norms mandate nonviolent con-flict resolution and negotiation in a spirit of live-and-let-live.3 Because democratic leaders are committedto these norms they try, as far as possible, to adoptthem in the international arena. This in turnmeans that

democraciesboth trustandrespectone anotherwhenaconflictof interestarisesbetween them. Sentimentsofrespectderive roma conviction hattheotherstatead-heres to the same normsand is therefore ustandwor-thyof accommodation.Trustderives romtheexpecta-tion that the otherpartyto the dispute s also inclinedto respecta fellow democracyand will be proscribednormativelyromresortingoforce.Togetherhesetwocausalmechanisms-norm externalizationnd mutualtrustandrespect-make upthenormativeogicandex-plain why democraciesrarelyfightone another(e.g.,Dixon 1994, 16-18; Russett 1993, 31-35; Weart1998,77-78, 87-93) (Fig. 1).Whilemutual rustandrespectgenerallyensurethatconflictsof interestbetweendemocraciesare resolvedamicably,herewillbe some situations nwhichosten-sibly democratic tates do not perceiveeach other tobe democraticandtherefore ightone another.Inpar-ticular,a democracymaynot be recognizedas such ifit is in the early stagesof democratization r if it doesnot meet the criteria hatpolicymakersn another tatehaveadoptedto definedemocracy e.g.,Russett1993,34-35;Weart1998,90-92, 132-34).This logic also explains why democracieshave of-ten been prepared o go to war with nondemocracies.Simplyput,nondemocracies re neither rustednor re-spected.Theyarenotrespectedbecause heirdomesticsystemsareconsideredunjust,andtheyare not trustedbecauseneitherdo they respectthe freedom of self-governingndividuals, oraretheysocialized o resolveconflictsnon-violently.Large-scale iolencemaythere-fore occur for one of two reasons.First,democraciesmaynotrespectnondemocracies ecause heyare con-sidered o be in a stateof waragainst heirown citizens.Warmay thereforebe permissible o free the peoplefromauthoritarianuleandintroducehumanrightsorrepresentativegovernment.Second,because democ-racies are inclined owardpeacefulconflictresolution,nondemocraciesmaybetempted otryandextract on-cessionsfromthemby attackingor threatening o useforce duringa crisis.In suchcircumstances emocra-cies mayeither have to defend themselves romattackor launchpreemptivestrikes(e.g., Doyle 1997, 30-43;Russett1993,32-35).

3 Strictly speaking, liberal and democratic norms are not equivalentand may be contradictory.With some notable exceptions, however,democratic peace theorists have tended to equate the two. I thereforeuse the terms liberal state, democracy, and liberal democracyinterchangeably throughout my discussion of the normative logic tomean states based on both liberal and democratic norms. On liberaltheory and norms see Doyle 1997, 4-7, and Owen 1997, 32-37. Ondemocratic theory and norms as defined by democratic peace theo-rists see Dixon 1994, 15-16; Russett 1993, 31; and Weart 1998,59-61.

586



4/19


5/19


FLAWS N THENORMATIVE OGICThe causal mechanisms that comprise the normativelogic do not appear to operate as stipulated. The avail-able evidence suggests that, contrary to the claims ofdemocratic peace theorists, democracies do not reliablyexternalize their domestic norms of conflict resolution,nor do they generally treat each other with trust andrespect when their interests clash. Moreover, existingattempts to repair the logic are unconvincing.Norm ExternalizationThe historical record indicates that democracies haveoften failed to adopt their internal norms of conflictresolution in an international context. This claim rests,first,on determining what democratic norms say aboutthe international use of force and, second, on estab-lishing whether democracies have generally adheredto these prescriptions.Liberal democratic norms narrowlycircumscribe therange of situations in which democracies canjustify theuse of force. As Doyle (1997, 25) notes, Liberal warsare only fought for popular, liberalpurposes. Thisdoesnot mean that they will go to war less often than otherkinds of states; it only means that there are fewer rea-sons available to them for waging war.Democracies are certainlyjustified in fightingwars ofself-defense. Locke ([1690] 1988), for example, arguesthat states, like men in the state of nature, have a rightto destroy those who violate their rights to life, liberty,and property (269-72). There is considerable disagree-ment among liberal theorists regarding precisely whatkinds of action constitute self-defense, but repulsing aninvasion, preempting an impending militaryattack, andfighting in the face of unreasonable demands all plausi-bly fall under this heading. Wagingwar when the otherparty has not engaged in threatening behavior doesnot. In short, democracies should only go to war whentheir safety and security are seriously endangered bythe expansionist policies of outlaw states (Rawls 1999,90-91).Another justification for the use of force is inter-vention in the affairs of other states or peoples, eitherto prevent blatant human rights violations or to bringabout conditions in which liberal values can take root.For Rawls (1999, 81), as for many liberals,human rightsviolators are to be condemned and in grave cases maybe subjected to forceful sanctions and even to inter-vention (see also Doyle 1997, 31-32, and Owen 1997,34-35). Mill ([1859] (1984)) extends the scope of inter-vention, arguing that barbarous nations can be con-quered to civilize them for their own benefit (see alsoMehta 1990). However, if external rule does not ensurefreedom and equality,it will be as illiberal as the systemit seeks to replace. Consequently, intervention can onlybe justified if it is likely to promote the developmentof conditions in which appropriate principles of justicecan be satisfied (Beitz 1979, 90).The imperialism of Europe's great powers between1815 and 1975 provides good evidence that liberaldemocracies have often waged war for reasons other

than self-defense and the inculcation of liberal values.Although there were only a handful of liberal democra-cies in the international system duringthis period, theywere involved in 66 of the 108 wars listed in the Cor-relates of War (COW) dataset of extrasystemic wars(Singer and Small 1994). Of these 66 wars,33 were im-perial, fought against previously independent peoples,and 33 were colonial, waged againstexisting colonies.It is hard to justify the imperial wars in terms ofself-defense. Several cases are clear-cut: The democ-racy faced no immediate threat and conquered sim-ply for profit or to expand its sphere of influence. Asecond set of cases includes wars waged as a result ofimperial competition: Liberal democracies conquerednon-European peoples in order to create buffer statesagainst other empires or to establish control over thembefore another imperial power could move in. ThusBritain tried to conquer Afghanistan (1838) in orderto create a buffer state against Russia, and France in-vaded Tunisia (1881) for fear of an eventual Italianoccupation. Some commentators describe these warsas defensive because they aimed to secure sources ofoverseas wealth, thereby enhancing national power atthe expense of other European powers. There are threereasons to dispute this assessment. First, these warswere often preventive rather than defensive: Russiahad made no move to occupy Afghanistan and Italyhad taken no action in Tunisia.A war designed to avertpossible action in the future, but for which there is nocurrent evidence, is not defensive. Second, there wasfrequently a liberal alternative to war. Rather thanimpose authoritarian rule, liberal great powers couldhave offered non-European peoples military assistancein case of attack or simply deterred other imperialpowers. Finally,a substantial number of the preventiveoccupations were a product of competition betweenBritain andFrance,two liberal democracies that shouldhave trusted one another and negotiated in good faithwithout compromising the rights of non-Europeans ifdemocratic peace theory is correct.A third set of cases includes wars waged directlyagainst non-Europeans whose territory bordered theEuropean empires. Because non-Europeans some-times initiated these wars contemporaries tended tojustify them as defensive wars of pacification to pro-tect existing imperial possessions. Again, there aregood reasons to doubt the claim that such wars weredefensive. In the first place, non-Europeans often at-tacked to prevent furtherencroachment on their lands;it was they and not the Europeans that were fighting inself-defense. Moreover, there is considerable evidencethat the imperial powers often provoked the attacks oracted preventively and exploited local instabilities as apretext for imposing control on the periphery of theirempires (Table 1).Nor were any of the extrasystemic wars fought toprevent egregious abuses of human rights or with theexpress purpose of replacing autocratic rule with amore liberal alternative. The colonial wars,by defini-tion, were conflicts in which imperial powers sought toperpetuate or reimpose autocratic rule. The imperialwars simply replaced illiberal indigenous government

588



6/19

AmericanPoliticalScienceReview Vol. 97, No. 4

TABLE . ImperialWars nvolvingLiberalDemocraciesWar DescriptionBritish-Zulu, 838 Zulus retaliated gainstterritorialncroachment.Suppressed.British-Afghan, 838 Preventivewar to create bufferagainstRussia.No Russianaction.FirstOpium,1839 British ttempt o forceopen Chinese markets.British-Baluchi,843 Annexationo control outhernroute o Afghanistan nd borderregions.UruguayanDispute,1845 British nterventionnlocalconflict.British-Sikh, 845 Attempto controlSikhs. Massedtroopson border.Sikhspreempted.British-Kaffir,846 Kaffirs etaliated gainstterritorialncroachment.Suppressed.British-Sikh, 848 RevoltagainstBritish ontrol.Suppressedandannexed.British-Burmese,1852 Annexation fterBurmese court nsultedBritishmerchants.Second Opium,1856 British ttempt o forceopen Chinese markets.British-Maori,860 Maorisretaliated gainstterritorialncroachment.Suppressed.British-Bhutanese,1865 Expeditiono eliminateBhutanese threat o controlon empire'speriphery.British-Ethiopian,867 Invasionnretaliationorimprisonmentf British ubjects.Franco-Tonkin,873 Disorder nTonkin sed as pretext orexpanding nfluence.Dutch-Achinese,1873 Dutchdemanded controlof ports.Aceh refused,so Dutch nvaded.British-Afghan, 878 Preventivewarto establish controlbeforeRussiaattempted o do so.British-Zulu, 879 ProvokedZuluattack o establishcontrolandpreventgrowthof Zulupower.Franco-Tunisian,881 Preventivewar: earItalywouldseek control.No Italian ction.Franco-Indochinese,1882 Attempto imposecontrol.Franco-Madagascan,1883 Attempto consolidatesphereof influence.British-Burmese, 885 Preventivewar: ear Francewouldseek control.No Frenchaction.Mandigo,1885 Frenchattempt o establish control.Franco-Dahomeyan, 889 Conquestto provideaccess to NigerRiverandevade British ustoms.Franco-Senegalese, 1890 Attempto controlandexploitresourcesof WestAfrica.Belgian-Congolese,1892 Attempto controlandexploitresourcesof Congo.British-Ashanti, 893 Attempto establishcontrolandpreemptFrance.No Frenchaction.Franco-Madagascan,1894 Conquestto consolidatecontrol.MahdiUprising,1896 British ttempt o controlNile and preemptFrance.No Frenchaction.British-Nigerian,897 Attempto establishcontrol.Punitive xpedition orkillings f Europeans.Boer,1899 Britishpreventivewar to destroygrowingpowerof Boers.FirstMoroccan,1911 Frenchattempt o establish control:eared Germanaction.No such action.British-Afghan, 919 Afghanattempt o escape British ontrol.Franco-Syrian,1920 Attempto establishinfluence.Syriadeclaredindependencein 1918.Note:Iuse Przeworski t al.2000, 18-29 throughouto code states as democratic rnondemocratic.Where heydo notprovide codingIuse theircriteria o determineregime type. (1) The chief executivemust be directly lected or responsible o an elected legislature.(2) The legislaturemustbe elected. (3) There must be more than one party.Iftherewere no parties,there was onlyone party, heincumbents stablishednonpartyorone-partyrule,or the incumbentsunconstitutionallylosed the legislatureand rewrote he rulesintheir avor, henthe regimewas nondemocratic.4) Incumbentsmust allow he possibilityhattheywill ose an electionand allowalawfulalternation f officeifdefeatedinelections. These criteriapreciselyreplicate he featuresthatdemocraticpeace theoristsclaimarecharacteristicf democracy e.g., Dixon1994, 15-16; Russett1993, 14-16, 31;Weart1998, 59-61). Acompletedatasetbased onthese criteria ndcoveringall states from1800 to 1999 is availableuponrequest.Iuse Doyle1997 to code states as liberal r illiberal.To be a liberaldemocracy,hen,a state mustbe both iberal s definedby Doyleand democratic s definedby Przeworskit al.Sources:Farwell 972;Featherstone1973;Haythornthwaite995; Hernon2000.

with authoritarian rule. When imperial rule was notimposed directly, the European powers supported lo-cal elites but retained strict control over their actions,thereby underwriting unjust political systems and ef-fectively implementing external rule. In short, despiteprotestations that they were bearing the white man'sburden, there is little evidence that liberal states' useof force was motivated by respect for human rightsor that imperial conquest enhanced the rights of non-Europeans.s

There are, then, several examples of liberal statesviolating liberal norms in their conduct of foreign pol-icy and therefore the claim that liberal states generallyexternalize their internal norms of conflict resolutionis open to question.Proponents of the democratic peace have down-played the importance of these findings in three ways.First, they have restated their argument and claimedthat democracies remain at peace because they trustand respect one other and fight nondemocracies be-cause they neither trust nor respect them. As Doyle(1997, 32) notes, Extreme lack of public respect ortrust is one of the major features that distinguishes re-lations between liberal and nonliberal societies fromrelations among liberal societies. According to this re-statement, we should not be surprisedto observe Euro-pean democracies fightingnon-Europeans and the nor-mative logic can therefore accommodate the imperialevidence. This alternative presentation of the logic is,

5 An analysisof decolonization is beyond the scope of this paper, butsome preliminarycomments are inorder.According to Russett (1993,35), decolonization came about at least in partbecause Western formsof self-rule took root in the colonies and the European powers there-fore lost confidence in their normative right to rule. The evidencesuggests otherwise. Of the 67 states that gained their independencebetween 1950 and 1980, 50 had autocratic governments (Przeworskiet al. 2000, 59-69).

589



7/19

The Flawed Logic of Democratic Peace Theory November2003

however, ad hoc. A more satisfying logic, and the oneputforwardby most democratic peace theorists, ismorecomplex: Democracies rarely fight each other becausethey trust and respect one another, and they are able todo so because they know that their democratic coun-terparts will act on the basis of democratic norms, thatis, they will only fight in self-defense or to democra-tize others. The key to this logic is that democraciesmust reliably externalize democratic norms. If they do,then trust and respect will prevail; if they do not, thenwe cannot be confident that peace will obtain betweenthem. The history of imperialism suggests that they donot and therefore casts doubt on the normative logic'sexplanatory power.Second, democratic peace theorists have claimedthat Britain, France, and the United States were notsufficiently liberal in the period under review and thuscannot be expected to reliably externalize their internalnorms (e.g., Rawls 1999, 53-54). If this claim is true, thenormative logic cannot tell us a great deal about inter-national politics. Britain, France, and the United Statesare generally considered to be classic liberal democra-cies; if they cannot be expected to behave in a liberalfashion, then few, if any, states can.Finally,democratic peace theorists assertthat they donot claim that liberal norms are the sole determinant ofdecisions for war;factors such as power and contiguitymatter as well (e.g., Russett 1995). This defense wouldbe convincing if I were claiming that liberal normswerenot the only factors that went into decision making orthat they were not as important in the decision makingprocess as other factors.However, the claim made hereis quite different: Liberal states have consistently vio-lated liberal norms when deciding to go to war. It is not

that liberal norms only matter a little; they have oftenmade no difference at all.In sum, there are good reasons to believe that one ofthe normative logic's key causal mechanisms does notoperate as advertised. Liberal democratic great powershave frequently violated liberal norms in their deci-sions for war, thereby casting doubt on the claim thatdemocracies generally externalize their internal normsof conflict resolution.Trustand RespectThe available evidence suggests that democracies donot have a powerful inclination to treat each other withtrust and respect when their interests clash. Instead,they tend to act like any other pair of states, bargaininghard,issuing threats, and, if they believe it is warranted,using military force.Cold War Interventions. American interventions todestabilize fellow democracies in the developing worldprovide good evidence that democracies do not alwaystreat each other with trust and respect when they havea conflict of interest. In each case, Washington's com-mitment to containing the spread of communism over-whelmed any respect for fellow democracies. Althoughnone of the target states had turned to communismor joined the communist bloc, and were led by whatwere at most left-leaning democratically elected gov-ernments, American officials chose neither to trust norto respect them, preferringto destabilize them by forceand replace them with autocratic (but anticommunist)regimes rather than negotiate with them in good faithor secure their support by diplomatic means (Table 2).

TABLE . AmericanColdWar nterventionsAgainstDemocraciesTarget DescriptionIran 1953) Mossadeq'sforeignpolicyaimed at disengagementfromsuperpowerrivalry.Domestically,alliedwithorsuppressedcommunistsas necessary.UnitedStates assisted coupthatoverthrewhim.Guatemala 1954) Fourcommunists ngovernmentand hardlyany ingeneral population.Army,he keyinstitutionnpolitics,was anticommunist.Arbentzundertooka numberof leftistreformprograms.UnitedStates financedand directed nvasion hatreplacedhim.Indonesia 1957-) Sukarno's guideddemocracy nly waysimultaneouslyo democratizeIndonesiaandpreventcivilwar.Communistsperformedwellin 1955 elections.UnitedStates assistedrebelsseekingto oust Sukarno.BritishGuyana(1961-) Jagan consistentlysoughtAmerican upport.Washington onvincedhe was leftistandsponsoredterrorist fforts o subverthim, henchangedelection laws to removehim.Brazil 1961, 1964) American oleinQuadros'sresignation 1961) unclear.Goulart'soreignpolicyneutral.Athome made no effort o legalizecommunistpartyorextend termillegally.AcceptedEastEuropeanaidand undertook ome leftistreforms.UnitedStates assisted inredscareand coupthatoverthrewhim.Chile(1973) Allendea socialist,butlegislature ontrolledbycenter-right.UnitedStates approvedChileanmilitary oup thatoverthrewhim.Nicaragua 1984-) Sandinistaswere moredemocratic hanAmerican-backedomoza dynasty.Heldelectionsin 1984 andbowed to internationalressureinrespectinga numberof civilrights.UnitedStates soughtto rollbackapparentcommunist hreat.Note:Democraticritainssisted heUnited tates nIranndBritish uyana.or egimeodingee Table .Iran adnot etexperienceda peacefultransferof power n 1953. TheAmerican-backedoup meant thatMossadeqwas notgivenan opportunityo provethat hewould and verpowerwerehe toloseanelection.Hewas,however,emocraticallylectedandcommittedo futurelections.Sources:Barnet1968; Bill1988; Forsythe1992;Gardner1997;Gleijeses 1991;Gurtov1974; Leacock1990; Ryan1995;Sater 1990;Tillema1973;Weis 1993.

590



8/19

American PoliticalScienceReview Vol. 97, No. 4Three features of these cases deserve emphasis.First,all the regimes that the United States sought to un-dermine were democratic. In the cases of Guatemala,British Guyana, Brazil, and Chile democratic pro-cesses were fairly well established. Iran, Indonesia, andNicaragua were fledgling democracies but Mossadeq,

Sukarno, and the Sandinistas could legitimately claimto be the first proponents of democracy in their re-spective countries. Every government with the excep-tion of the Sandinistas was replaced by a succession ofAmerican-backed dictatorial regimes.Second, in each case the clash of interests betweenWashington and the target governments was not par-ticularly severe. These should, then, be easy cases fordemocratic peace theory since trust and respect aremost likely to be determinative when the dispute isminor. None of the target governments were commu-nist, and although some of them pursued leftist policiesthere was no indication that they intended to imposea communist model or that they were actively court-ing the Soviet Union. In spite of the limited scope ofdisagreement, respect for democratic forms of govern-ment was consistently subordinated to an expandedconception of national security.Third,there is good evidence that supportfor democ-racy was often sacrificed in the name of Americaneconomic interests. At least some of the impetus forintervention in Iran came in response to the national-ization of the oil industry, the United Fruit Companypressed for action in Guatemala, International Tele-phone and Telegraph urged successive administrationsto intervene in Brazil and Chile, and Allende's effortsto nationalize the copper industry fueled demands thatthe Nixon administration destabilize his government.In sum, the record of American interventions in thedeveloping world suggests that democratic trust andrespect has often been subordinated to security andeconomic interests.Democratic peace theorists generally agree thatthese interventions are examples of a democracy usingforce against other democracies, but they offer two rea-sons why covert interventions should not count againstthe normative logic. The first reason is that the targetstates were not democratic enough to be trusted andrespected (Forsythe 1992; Russett 1993, 120-24). Thisclaim is not entirely convincing. Although the targetstates may not have been fully democratic, they weremore democratic than the regimes that preceded andsucceeded them and were democratizing further. In-deed, in every case American action brought more au-tocratic regimes to power.The second reason is that these interventions werecovert, a fact believed by democratic peace theoriststo reveal the strength of their normative argument. Itwas precisely because these states were democratic thatsuccessive administrations had to act covertly ratherthan openly initiate military operations. Knowing thattheir actions were illegitimate, and fearing a publicbacklash, American officials decided on covert action(Forsythe 1992; Russett 1993, 120-24). This defensefails to address some important issues. To begin with, itignores the fact that American public officials, that is,

the individuals that democratic peace theory claims aremost likely to abide by liberal norms,showed no respectfor fellow democracies. Democratic peace theorists willrespond that the logic holds, however, because theseofficials were restrained from using open and massiveforce by the liberal attitudes of the mass public. Thisis a debatable assertion; after all, officials may haveopted for covert and limited force for a variety of rea-sons other than public opinion, such as operationalcosts and the expected international reaction. Simplybecause the use of force was covert and limited, thisdoes not mean that its nature was determined by publicopinion.But even if it is true that officials adopted a covertpolicy to shield themselves from a potential publicbacklash, the logic still has a crucial weakness: Thefact remains that the United States did not treat fellowdemocracies with trust or respect. Ultimately, the logicstands or falls by its predictive power, that is, whetherdemocracies treat each other with respect. If they do,it is powerful; if they do not, it is weakened. It does notmatter why they do not treat each other with respect,nor does it matter if some or all of the population wantsto treat the other state with respect; all that mattersis whether respect is extended. To put it another way,we can come up with several reasons to explain whyrespect is not extended, and we can always find socialgroups that oppose the use of military force againstanother democracy, but whenever we find several ex-amples of a democracy using military force againstother democracies, the trust and respect mechanism,and therefore the normative logic, fails an importanttest.6Great Powers. Layne (1994) and Rock (1997) havefound further evidence that democracies do not treateach other with trust and respect in their analyses ofdiplomatic crises involving Britain, France, Germany,and the United States. Layne examines four prominentcases in which rival democracies almost went to warwith one another and asks whether the crises were re-solved because of mutual trust and respect. His con-clusion offers scant support for the normative logic:In each of these crises, at least one of the democraticstates involved was prepared to go to war..... In eachof the four crises, war was avoided not because of the'live and let live' spiritof peaceful dispute resolution at

6 We cannot conclusively reject the trust and respect mechanism onthe basis of these cases since the United States may have been sig-nificantly more likely to intervene covertly against nondemocraciesduring the Cold War.Creating a comprehensive dataset of covert in-terventions to test this claim is,however, unlikely to be a simple task.Moreover, a chi-square test indicates that we would have to find inexcess of 30 American covert interventions against nondemocraciesbefore we could claim that it was significantly more likely to inter-vene covertly against nondemocracies than democracies (p < .05).This calculation rests on (a) the fact that there were 1,682 years ofdemocracy and 3,007 years of nondemocracy between 1950 and 1990(Przeworski et al. 2000, 29); (b) the fact that there were eight covertinterventions against democracies in this period; and (c) the assump-tion that the United States had the capacity to intervene anywherein the world in any given year.

591



9/19

The Flawed Logic of Democratic Peace Theory November 2003

TABLE. AmericanPerceptionsof LiberalStatus of ForeignPowersParty Status Party Status Levelof ConsensusBritain 1794-96 Federalists Liberal Republicans Illiberal No across-partyagreementFrance 1796-98 Federalists Illiberal Republicans Liberal No across-partyagreementBritain 1803-12 Federalists Liberal Republicans Illiberal No across-partyagreementBritain 1845-46 Whigs Liberal Democrats Illiberal No across-partyagreementMexico 1845-46 Whigs Liberal Democrats Illiberal No across-partyagreementBritain 1861-63 Republicans Liberal Democrats Illiberal No across-partyagreementSpain 1873-73 Republicans Mixed Democrats Mixed Nowithin-partygreementChile 1891-92 Republicans Mixed Democrats Mixed Nowithin-partygreementBritain 1895-96 Republicans Mixed Democrats Mixed Nowithin-partygreementSpain 1896-98 Republicans Illiberal Democrats Illiberal Consensus-illiberalSource:Owen 1997.democratic peace theory's core, but because of realistfactors (Layne 1994, 38).7Similarly,Rock finds little evidence that shared lib-eral values helped resolve any of the crises betweenBritain and the United States in the nineteenth cen-tury.In addition, his analyses of the turn-of-the-centurygreat rapprochement and naval arms control duringthe 1920sshow that even in cases where liberal states re-solved potentially divisive issues in a spiritof accommo-dation, shared liberal values had only a limited effect.In both cases peace was overdetermined and liberalvalues and democratic institutions were not the onlyfactors inclining Britain and the United States towardpeace, and perhapsnot even the dominant ones (Rock1997, 146).8In sum, the trust and respect mechanism does notappear to work as specified. Shared democratic valuesprovide no guarantee that states will both trust andrespect one another. Instead, and contrary to the nor-mative logic's claims, when serious conflicts of interestarise between democracies there is little evidence thatthey will be inclined to accommodate each other's de-mands or refrain from engaging in hard line policies.Repaired Normative LogicGiven that democracies have not treated each other asthe normative logic predicts, democratic peace theo-rists have tried to repair the logic by introducing a newcausal factor: perceptions. In the revised version of thelogic, democracies will only trust and respect one an-other if they consider each other to be democratic. Thisadjustment can only improve the logic's explanatorypower if we can predict how democracies will catego-rize other states with a high level of confidence and ifthis categorization is relatively stable. The available ev-idence suggests, however, that policymakers' personalbeliefs and party affiliations, or strategic interest, oftenpreclude coherent, accurate, and stable assessments of

regime type, thereby lessening our confidence thatjointdemocracy enables democracies to remain at peace.Elusive Consensus. There is rarely agreement, evenamong well-informed policymakers, about the demo-cratic status of a foreign power and we are, there-fore, unlikely to be able to predict how democracieswill classify other states' regime type with a high levelof confidence.9 Owen (1997) has examined the viewsof liberal elites in 10 war-threatening crises involvingthe United States and another state between 1794 and1898. In six of the cases, the major political parties inthe United States disagreed about the liberal status ofFrance, Britain, Chile, and Spain. In three other cases,these disagreements extended both across and withinparties. In only one case, the Spanish American Crisis,was there a consensus within the American elite regard-ing the liberal status of the foreign power (Table 3).In sum, the evidence from Owen's cases suggests thatwe are unlikely to be able to predict how states willperceive one another's regime type: Opinion is almostalways divided, even for cases that look easy to outsideobservers. This being the case, the repaired normativelogic can only tell us if liberal states will view each otheras such after the fact: If they treat each other with trustand respect, then they must have viewed each other asliberal; if they do not, then they must have viewed eachother as illiberal.In these circumstances,the only way to create a moredeterminate logic is to predict whose opinions will winout in the domestic political game. If, for example, wecan predict that doves, republicans, or business inter-ests will generally get their way, then we may be ableto predict policy outcomes. Suchpredictions have, how-ever, eluded democratic peace theorists (see AutocraticRestraint, below).Inaccurate Assessment. Democracies will also oftensimply get another state's regime type wrong, therebylessening our confidence that objectively democraticstates will not fight one another. In five of the ninecases where Owen evaluates how other states per-ceived America, foreign liberal elites either classifiedthe United States as illiberal or were unsure as to its

7 Layne 1997 examines three further cases and comes to the sameconclusion.8 Rock's analysis of the naval arms control agreements of the 1920smisses an important critique of the normative logic. It is not clear, ifwe accept the logic, why the United States should be so concernedabout a naval alliance between democratic Britain and a democra-tizing Japan. See, for example, P.Kennedy 1983, 267-98.

9 Hartz (1955) argues that although America is a thoroughly liberalstate, there have always been violent disagreements about the mean-ing of liberalism.

592



10/19

American PoliticalScience Review Vol. 97, No. 4status. In 1873, Spanish liberals, most of whom identi-fied with the Spanish Republican party, disagreed overthe status of the United States. All Chilean elites andall Spanish elites, regardless of their party affiliation,regarded the United States as illiberal in the 1890s.Finally, British opinion leaders, who had agreed thatthe United States was liberal for over a century, weredivided over its liberal status in 1895-96. The paradig-matic liberal state was, then, often perceived as any-thing but. Even more surprisingis the fact that as thenineteenth century wore on, and the United States be-came more liberal by most objective standards, otherstates increasingly viewed it as illiberal.Regime Type Redefined Not only are perceptions ofother regimes often contested or inaccurate, but theyare also subject to redefinition, and this redefinitiondoes not always reflect the actual democratic attributesof those states. Oren (1995) conducts an in depth studyof the United States' changing relationship with Im-perial Germany prior to World War I and finds thatAmerican opinion leaders stopped defining Germanyas a democracy as the two countries' strategic relation-ship began to deteriorate. This observation leads him toconclude that democracy is not a determinant as muchas it is a product of America's foreign relations: Thereason we do not to fight 'our kind' is not that 'likeness'has a great effect on war propensity, but rather thatwe from time to time subtly redefine our kind to keepour self image consistent with our friends' attributesand inconsistent with those of our adversaries (Oren1995, 147). In other words,contraryto the expectationsof the normative logic, perception of regime type is anoutcome rather than a causal factor.Liberal states appearespecially prone to this practiceof reinterpreting who should be trusted and respected.In the nineteenth century,non-European peoples couldbe put under autocratic imperial rule for their owngood. In the early twentieth century,as Oren has noted,the bar was raised higher and Imperial Germany wasjudged worthy of neither trust nor respect. By the endof the century,even liberal democratic Japancould notcount on unquestioning American friendship. In eachcase, prestige, security concerns, or economic interestsshaped perceptions of regime type.10These examples raise serious problems for anycausallogic based on perceptions. Discerning whether percep-tions matter inevitably becomes a question of siftingthrough the statements of policymakers and opinionleaders during a crisis or war. At the same time, publicfigures will try to distinguish their own state from theenemy in these situations, both for their own cogni-tive consistency and to rally the public. Since peoplein the modern world generally identify themselves asmembers of a nation state, these distinctions will tendto focus on political structures. Scholars will therefore

always be able to find evidence that the other statewas not perceived to be sufficiently democratic asleaders go about demonizing the enemy. I am not argu-ing that this represents a misreading of the evidence-perceptions of another state are bound to change incrisis situations-I am only suggesting that these per-ceptions are caused by factors other than the objectivenature of foreign regimes.In sum,proponents of the normative logic have donelittle to strengthen their case by introducing percep-tions as an independent variable. Often states do nothave a unified perception of the liberal attributes ofa foreign power and it is therefore difficult to arguethat perceptions of regime type affect policy. More-over, these perceptions may change independently ofthe objective nature of the other regime, suggestingthat it is entirely possible for liberal states to fight oneanother.

FLAWSIN THEINSTITUTIONALOGICThe causal mechanisms that make up the institutionallogic do not appear to operate as stipulated. There aregood reasons to believe that accountability, a mech-anism common to all five variants of the institutionallogic, does not affect democratic leaders any more thanit affects their autocratic counterparts. Nor does theavailable evidence support the claims of the institu-tional logic's other causal mechanisms. Pacific publicsand antiwargroups rarelyconstrainpolicymakers' deci-sions for war, democracies are neither slow to mobilizenor incapable of launching surprise attacks, and openpolitical competition provides no guarantee that a statewill be able to reveal its level of resolve in a crisis.AccountabilityEach variantof the institutional logic rests on the claimthat democratic institutions make leaders accountableto various groups that may, for one reason or another,oppose the use of force. I do not dispute this claim but,instead, question whether democratic leaders are moreaccountable than their autocratic counterparts. Sincewe know that democracies do not fight one anotherand autocracies do fight one another, democrats mustbe more accountable than autocrats if accountabilityis a key mechanism in explaining the separate peacebetween democracies. On the other hand, if autocratsand democrats are equally accountable or autocrats aremore accountable than democrats, then there are goodreasons to believe that accountability does not exert theeffect that democratic peace theorists have suggested.11Following Goemans (2000a) I assume that a leader'saccountability is determined by the consequences aswell as the probability of losing office for adopting anunpopular policy.Thisbeing the case, there is no apriorireason to believe that aleader who is likely to lose officefor fighting a losing or costly war, but unlikely to be0 Oren notes that American perceptions of the democratic natureof Japan and the Soviet Union in the twentieth century have tendedto reflect their behavior rather than their domestic institutions andvalues. Similarly,Blank (2000) argues that strategic factors influencedBritish and American perceptions of each other's liberal status in thenineteenth century. 1 Evaluations of the effects of war on the tenure of leaders includeBueno de Mesquita and Siverson 1995 and Goemans 2000a.

593



11/19


TABLE. Consequences of Engaging nLosingor CostlyWarsWars Removed PunishedDemocraticosers 4 3 (75%) 0 (0%)Autocraticosers 89 31 (35%) 26 (29%)Democrats ncostlywars 15 4 (27%) 1 (7%)Autocratsncostlywars 77 27 (35%) 21 (27%)

exiled, imprisoned, or killed in the process, should feelmore accountable for his policy choices than a leaderwho is unlikely to lose office but can expect to be pun-ished severely in the unlikely event that he is in factremoved.Therefore, determining whether autocrats or demo-crats are more accountable and, consequently, morecautious about going to war rests on answering threequestions: Are losing democrats or losing autocratsmore likely to be removed from power? Are losingdemocrats or losing autocrats more likely to be pun-ished severely? and Are democrats or autocrats morelikely to be removed and/or punished for involvementin costly wars,regardless of the outcome?To answer these questions I have used a modifiedversion of Goemans's (2000b) dataset. Our analysesdiffer in one fundamental respect: While he counts theremoval of leaders by foreign powers as examples ofpunishment, I do not. This decision is theoretically in-formed. The purpose of the analysis is to determinewhether leaders' decisions for war are affected by theirdomestic accountability, that is, if there is something

about the domestic structure of states that affects theirchances of being punished. Punishmentbyforeign pow-ers offers no evidence for or against the claim thatdemocrats or dictators have a higher or lower expecta-tion of being punished by their citizens for unpopularpolicies, and these cases are therefore excluded. I havealso made two minor changes to the data that do notaffect the results: I have added 19 wars that appearin the COW dataset but not in Goemans's dataset andcoded 11 regimes that Goemans excludes.12The resultsappear in Table 4.Although democratic losers are two times morelikely to be removed from power than autocratic losers,this evidence is not strong.This is because there areonlyfour cases of democratic losers in the entire dataset,making it impossible to draw any firm conclusionsabout the likelihood that losing democrats will be re-moved. Prime Minister Menzies of Australia, for exam-ple, resigned early in the Vietnam War, but his resig-nation may have had more to do with the fact that hewas in his seventies than the expectation of defeat inSouth East Asia a decade later. If this case is recoded,as it probably should be, democratic losers have only

been removed from power 50% of the time and thedistinction between democrats and autocrats is small.Losing autocratsare more likely to suffer severe pun-ishment than their democratic counterparts. None ofthe four losing democrats was punished, whereas 29%of autocratic losers were imprisoned, exiled, or killed.

Thus,while democratic and autocratic losers have sim-ilar chances of being removed from office, autocratsseem to be more likely to suffer severe punishment inaddition to removal.The evidence from costly wars,regardless of whetherthe leader was on the winning or losing side, confirmsthese findings. Costly wars are defined as wars in whicha state suffered one battle fatality per 2,000 population,as the United States did in World WarI.13Historically,autocrats have been more likely both to lose office andto be punished severely if they become involved in acostly war. Autocrats have been removed 35% of thetime and punished 27% of the time, while democratshave only been removed 27% of the time and punished7% of the time.14In short, there is little evidence that democratic lead-ers face greater expected costs from fighting losing orcostly wars and are therefore more accountable thantheir autocratic counterparts.This being the case, thereis good reason to doubt each variant of the institutionallogic.Public ConstraintPacific public opinion does not appear to place a fun-damental constraint on the willingness of democraciesto go to war. If it did, then democracies would be morepeaceful in their relations with all types of states, notjust other democracies.However, instead of being morepeaceful, on average democracies are just as likely togo to war as nondemocracies (Farberand Gowa 1995).There are three reasons why publics are unlikely toconstrain democratic war proneness. First, the costs ofwar typically fall on a small subset of the population

12Nondemocracies: Mecklenburg-Schwerin, Hesse Grand Ducal,Hesse Electoral, and Hanover in the Seven Weeks War; Germanyin the Franco-PrussianWar;Greece in the war of 1919 with Turkey;Ethiopia, Bulgaria, and Italy in World War II; and Cyprus in 1974.Democracy: Israel in 1948.

13 The results do not change with alternative definitions of costlywars (one fatality per 1,000 population and one fatality per 500population).14Proponents could still interpret the evidence as supporting demo-cratic peace theory. The very fact that democratic leaders rarely losewars suggests that they know that they will be punished for losingwars and therefore only select themselves into wars they can win.There are good reasons to dispute this selection effects argument.Desch (2002) estimates the probability that a state will start a war,then win it, and finds that democracy has one of the smallest effectsof any variable. Stam (1996) reaches a similar conclusion. Reiter andStam (2002) find that democracies are more likely to win wars theyinitiate but do not report the relative effect of democracy comparedto other variables. Desch also notes that if democratic leaders aremore selective about choosing wars, and only start easy ones, thenthey should engage in fewer wars than autocratic leaders since warisinherently risky and few wars are sure bets. The evidence, however,suggests that democracies are just as war prone as other types ofstates. It is also worth noting that if democrats are more selectiveabout the wars they get involved in, then we should see them en-gage in fewer costly wars since they know that costly wars threatentheir incumbencies. However, there is little difference between thepropensity for democracies and that for autocracies to incur highcosts. Democracies incurhigh costs in 34% of cases,while autocraciesdo so 42% of the time.

594



12/19

AmericanPoliticalScience Review Vol. 97, No. 4that will likely be unwilling to protest government pol-icy. Excluding the two World Wars, democratic fatal-ities in war have exceeded 0.1% of the population inonly 6% of cases. In 60% of cases, losses representedless than 0.01% of the population or one in 10,000people. Most democratic citizens, then, will never bepersonally affected by war or know anyone affectedby military conflict. Adding the many militarized dis-putes involving democracies strengthens this finding.Both the United States and Britain have suffered fewerthan 100 battle casualties in approximately 97% of themilitarized disputes in which they have been involved(Singer and Small 1994). Moreover, modern democra-cies have tended to have professional standing armies.Members of the military, then, join the armed forcesvoluntarily, accepting that they may die in the serviceof their countries. This in turn means that their familiesand friends, that is, those who are most likely to sufferthe costs of war, are unlikely to speak out against agovernment that chooses to go to war or are at leastless likely to do so than are the families and friends ofconscripts.In short, the general public has little at stakein most wars and those most likely to suffer the costsof war have few incentives to organize dissent.Second, any public aversion to incurringthe costs ofwar may be overwhelmed by the effects of nationalism.In addition to the growth of democracy,one of the moststriking features of the modern period is that peoplehave come to identify themselves, above all, with thenation state. This identification has been so powerfulthat ordinary citizens have repeatedly demonstrated awillingness to fight and die for the continued existenceof their state and the security of their co-nationals.There are, then, good reasons to believe that if thenational interest is thought to be at stake, as it is inmost interstate conflicts, calculations of costs will notfigure prominently in the public's decision process.Third, democratic leaders are as likely to lead as tofollow public opinion. Since nationalism imbues peo-ple with a powerful spirit of self-sacrifice, it is activelycultivated by political elites in the knowledge that onlyhighly motivated armies and productive societies willprevail in modern warfare (e.g., Posen 1993). Demo-cratically elected leaders are likely to be well placedto cultivate nationalism, especially because their gov-ernments are often perceived as more representativeand legitimate than authoritarian regimes. Any call todefend or spread ourway of life, for example, is likelyto have a strong resonance in democratic polities, andindeed the historical record suggests that wars have of-ten given democratic leaders considerable freedom ofaction, allowing them to drum up nationalistic fervor,shape public opinion, and suppress dissent despite theobligation to allow free and open discussion.Events in the United States duringboth WorldWarshighlight the strength of nationalism and the abilityof democratic elites to fan its flames. Kennedy (1980,46) notes that during the First World War, PresidentWilson lacked the disciplinary force of quick comingcrisis or imminent peril of physical harm but turnedsuccessfully to thedeliberate mobilization of emotionsand ideas. At the same time his administration turned

a blind eye to, or actively encouraged, the deliberatesubversion of antiwar groups within the United States.The Roosevelt administration was equally successfulat generating prowar sentiment during World War II.Early in the war the president spoke for the nation inasserting that the German firebombing of populationcenters had shocked the conscience of humanity, andyet, remarkably,there was no sustained protest in theUnited States against the bombing of Japanese citiesthat killed almost a million civilians a few years later.This abrupt transformation, notes Dower (1986), wasmade possible by a massive propagandacampaign,con-doned by the political elite, describing the Japanese assubhuman and untrustworthy others. In stark con-trast, America's allies were forgiven all their faultsRussian Communists were transformed into agrarianreformers, Stalin into Uncle Joe... (Ambrose 1997,150).Sentiments like these are not aroused only in thevictims of aggression. Although Lord Aberdeen's gov-ernment was reluctant to go to war with Russia over theCrimea in 1854, There was no doubt whatever aboutthe enthusiasm of British public opinion, as expressedby every conduit open to it. The protests of Cobdenand Bright, leaders of the British Peace Movement,were howled down in the House of Commons, in thePress, and at meeting after public meeting.... [They]were thus the first liberal leaders, and by no means thelast, to discover that peace and democracy do not gohand in hand; that public opinion is not an infalliblespecific against war;and that 'the people,' for whateverreasons, can be very bellicose indeed. The next gener-ation of pacifists,the opponents of the Boer War, werevilified in the popular press, had their meetings brokenup, [and] were subjected to physical attack (Howard1978, 45-46, 68).These are not isolated examples. The world's mostmilitarily active democracies-Britain, France, India,Israel, and the United States-have gone to war 30times since 1815. In 15 cases, they were the victims ofattack and therefore we should not be surprised thatpublics reacted in a nationalistic fashion or were per-suaded to support decisions for war. There are, how-ever, 15 other cases in which one could plausibly arguethat it was not obvious to the public that war was inthe national interest because there was no immediatethreat to the homeland or vital national assets. In 12of these cases, the outbreak of war was greeted by aspontaneous and powerful nationalistic response or,in the absence of such a reaction, policymaking elitessuccessfully persuaded a previously unengaged publicto acquiesce to, and in some cases support, the use offorce. In only three cases-the French and British at-tack on Egypt (1956) and the Israeli attack on Lebanon(1982)-did publics not spontaneously support the warand remain opposed to it despite policymaking elites'best efforts to influence their opinions.15

15 Democratic victims: the United States in World War II; Israel inthe Palestine War,Warof Attrition, and Yom KippurWar;Britain inboth World Warsand the FalklandsWar;France in both WorldWars;

595



13/19

The Flawed Logic of Democratic Peace Theory November 2003

One way to try and rescue the public constraintmechanism would be to combine constraints with re-spect for fellow democratic polities (e.g., Mintz andGeva 1993). This new argument would hold thatdemocracies have formed a separate and joint peacebecause democratic citizens are only averse to costsin their relations with other democracies. There are,however, several cases that belie this claim.16There are, then, good reasons to believe that pacificpublic opinion does not significantly reduce the likeli-hood that democracies will go to war. In the majorityof cases, the public is likely to be unaffected by warand therefore adopt a permissive attitude towards theuse of force. Moreover, in those cases where the na-tional interest or honor is at stake, democratic publicsare as likely as any other to disregard the costs of warand democratic leaders have considerable opportuni-ties both to encourage and to exploit nationalistic fer-vor.GroupConstraintThere are two problems with the group constraintmechanism. First, there is little evidence for the claimthat antiwar groups will, more often than not, capturethe democratic policymaking process. Second, if themechanism is to explain why democracies do not fightone another but also account for wars in other kinds ofdyads,then group constraints must be weaker in autoc-racies than democracies, but this does not appear to bethe case.Capturing the State. States are representative insti-tution[s] constantly subject to capture and recapture,construction and reconstruction by coalitions of so-cietal actors (Moravcsik 1997, 518). Moreover, they

are imperfect representative institutions, more likelyto represent those groups that are better organizedand have more at stake in a given issue. Based on thisinsight, there is no reason to believe that pacific inter-est groups will generally win out over prowar groups.While liberal elites, for example, may be well organizedand have a powerful incentive to avoid war with otherdemocracies, other more bellicose actors such as themilitary industrial complex are likely to have just asmuch at stake and be equally proficient at furtheringtheir own interests.

Indeed, the historical record suggests that propo-nents of foreign aggression can often prevail in domes-tic debates. Owen (1997) examines four cases of theUnited States going to war in the nineteenth century.In three of his cases, one of the two major politicalparties was opposed to war but failed to avert it. In thefourth case, the antiwargroup was smaller and also lostout to the prowar group. Similarly, Snyder (1991) findsthat both Britain and the United States have adoptedaggressive foreign policies in the past as prowar groupshave effectively captured the state. Britain's expan-sionist policy in the middle of the nineteenth centuryowed much to the fact that imperialistgroupswere ableto influence policymaking: Imperial ideologists wereable to have a large impact because of their apparentmonopoly on expertise and effective organization, andbecause of the ambivalent interest of the audience.In the American case, despite a Cold War consensusagainst involvement in high-cost, low benefit endeav-ors, the United States became involved in both Koreaand Vietnam as a result of coalitional logrolling (Snyder1991, 206, 209).17 In sum, there are good reasons tobelieve that pacific interest groups may not generallyinfluence the foreign policies of democratic states.Autocratic Constraint. Autocratic leaders typicallyrepresent themselves or narrow selectorates and thesegroups have powerful incentives to avoid war.The first reason for avoiding war is that wars costmoney and solving the problem of war finance ulti-mately poses a threat to an autocrat's hold on power.The argumenthere is straightforward.The costs of warhave risen exponentially since the middle of the nine-teenth century and governments have had to figure outhow to meet these costs. Although the money can theo-retically be raised with or without the consent of thosefromwhom it is demanded, in practice non-consensualsources of revenue have generally proved less elasticthan taxation based on consent. Participation in warhas, therefore, tended to go hand in hand with expan-sion of the franchise (Ferguson 2001, 32-33, 77, 80; seealso Freeman and Snidal 1982). This being the case,autocrats have a powerful incentive not to go to warfor fear of triggering social and political changes thatmay destroy them.The nature of civil military relations in civilian-led authoritarian states provides another incentive for

India in the Sino-Indian, Second Kashmiri, and Bangladesh Wars;and Britain, France, and the United States in the Boxer Rebellion.Warssupported by public or to which public acquiesced even thoughthey were not clearly in the national interest: the United States in theMexican-American, Spanish-American, Korean, Vietnam, and GulfWars and World War I; Israel in the Sinai and SixDay Wars;Britainin the Crimean and Anglo-Persian Wars;France in the Roman Re-public and Sino-French Wars. I only consider the major protagonistsin any given war and, therefore, exclude cases like Britain's decisionto support the United States in the Gulf War.Also, I only considerpublic opinion early in a war since it is presumablythis initial reactionthat concerns policymakers the most.Democratic peace theorists could still claim that these examplesdo not invalidate the public constraint mechanism because there aremany more examples where democracies have not escalated a crisisor have pulled back from the brinkbecause leaders anticipated publicopposition. These nonevents are difficult to observe but if there area lot of them, then 12 examples of public constraints not operatingdo not provide conclusive proof that the mechanism generally failsto operate. Proponents of the democratic peace have not, however,uncovered a large number of such non-events.16 Britain and France over Belgium (1830-32), the Near East (1838-41), Tahiti and Tangier (1844), and Fashoda (1898); Britain and theUnited States in the Oregon Crisis (1845-46), the TrentAffair (1861),and the Venezuelan Crisis (1895-96); Britain and the Boers in theBoer War (1899-1902); France and Germany in the Ruhr Crisis(1923); arguablyFrance, Britain, and Germany before World War I;Peru and Ecuador in the Amazon in the 1980s and 1990s;and Indiaand Pakistan over Kashmir in the 1990s. See Howard 1978; Layne1994, 1997; and Rock 1997.

17 Snyder argues that democracies are moderate overexpandersrather than extreme overexpanders because open debate encouragesquick learning. The fact remains, however, that while they may besmart about their overexpansion, they are still prone to it.

596



14/19

AmericanPoliticalScienceReview Vol. 97, No. 4

ruling groups to avoid war. Since civilian control of themilitary is often more tenuous in autocracies than indemocracies, nonmilitary leaders of autocratic stateshave a powerful incentive to maintain weak militariesfor fear of domestic coups. The problem, from a foreignpolicy standpoint, is that states with weak militaries arevulnerable to foreign aggression. Thusan absolute rulerfaces a dual problem according to Gordon Tullock(1987, 37): [H]e may be overthrown by his neigh-bor's armies, or by the armies he organizes to defendhim againsthis neighbors. Because they recognize thisproblem, civilian authoritarian leaders will generallyprefer to avoid rather than wage war.A different set of factors can inhibit the war prone-ness of military dictators. First, since they must devoteconsiderable time and energy to repressing populardissent at home, they have fewer military resources todevote to external wars. Second, because the militaryis used for internal repression it is unlikely to have agreat deal of societal support and will be ill equippedto deal with external enemies. Third, leaders who as-sume control of the army run the risk of being heldpersonally responsible for any subsequent failures andmay not be prepared to take that risk. Finally, timespent organizing military campaigns is time away fromother governmental duties on which a dictator's tenurealso depends (Andreski 1980;Tullock 1987,37; see alsoDassel 1997).In sum, it is not clear that states behave as the groupconstraint mechanism suggests. Although democraciesand autocracies have selectorates of differing size andallow social groups different levels of access to the poli-cymaking process, they may nevertheless adopt similarpolicies. Not only are democratic governments able toresist the influence of antiwar groups, but they are infact subject to capture by prowar groups. Autocracies,on the other hand, often represent groups that havea vested interest in avoiding foreign wars (see, e.g.,Peceny, Beer, and Sanchez-Terry 2002).Slow MobilizationThe historical record offers scant support for the claimthat the complexity of mobilizing diverse groups indemocracies slows decisions to use force.American presidents have often circumvented or ig-nored checks andbalances, thereby speeding up the wardecision process.18The United States has taken militaryaction abroad more than 200 times during its history,but only five of these actions were wars declared byCongress, and most were authorized unilaterally by thepresident (Rourke 1993, 11). Circumventingthe demo-craticprocess has taken several forms.Some presidentshave simply claimed that matters of national securityare more important than observing the constitution.Jefferson was the first to assert that obeying the con-stitution was the mark of a good president, but that

the law of necessity, of self preservation, of saving ourcountry when in danger, are of the higher obligation(75). Another common tactic has been to redefine theaction as anything but a war, thereby obviating theneed for consultation or debate. Washington added hotpursuitand preemption to the president's prerogatives,Jackson popularized reprisals, and Wilson unilaterallyauthorized interventions, most notably in Russia af-ter World War I. Alternatively, presidents have usedtheir powers to put troops in harm's way in order toprecipitate wider conflicts. Both Polk's actions priorto the Mexican American War and Roosevelt's tacticsprior to America's official entry into World War II fitthis pattern. Finally, incumbents of the White Househave often simply ignored Congress. Trumanorderedforces into Korea without even asking Congress forretroactive support, and at the height of the ImperialPresidency, Nixon rejected the need for congressionalauthority when he invaded Cambodia.While efforts have been made to ensure that choicesfor war and peace are subject to open debate-notably with the passage of the War Powers Resolution(1972)-checks and balances have generally failed tooperate and there have been frequent violations of thespirit if not the letter of the Resolution (Rourke 1993,119-38). The Gulf Warprovides a recent example. Bushadministration officials decided to launch OperationDesert Shield without consulting Congress and repeat-edly put off a congressional vote fearing that it mightgo against them. The decision for Desert Storm wasalso made unilaterally. Bush argued that he did notneed a congressional resolution and was determinedto avoid asking for authorization lest this imply thatthe Executive did not have the final say on matters ofwar. His reaction to Congress's authorization of theuse of force is instructive: In truth, even had Congressnot passed the resolution I would have acted and or-dered our troops into combat. I know it would havecaused an outcry,but it was the right thing to do. I wascomfortable in my own mind that I had constitutionalauthority. It had to be done (Bush and Scowcroft1998, 446).In sum, the slow mobilization mechanism does notappear to function as claimed. Democratic leaders fre-quently decide that protecting what they deem to bethe national interest requires swift and decisive ac-tion. When they believe such situations have arisenthey have been able and willing simply to bypass thedemocratic imperative of open debate and consensusdecision making.

Surprise AttackDemocracies are no less capable of carrying out sur-prise attacks than other kinds of states.19 The mainreason for this is that an attacker's regime type islargely unrelated to the success or failure of an attack.

18 The focus here is on American foreign policy.Other democraticallyelected leaders have adopted similar tactics to initiate military actionwith only minimal legislative input. This paragraphrelies on Reveley1981, 135-69, and Rourke 1993, 63-106.19A surprise attack is an attack against a target that is not preparedfor it due to mistaken estimates of whether, when, where, and howthe enemy will strike (Betts 1982, 11).

597



15/19

The Flawed Logic of DemocraticPeace Theory November 2003

Analysis of surprise attacks, notes Kam (1988, 37),suggests that the intelligence community seldom failsto anticipate them owing to a lack of relevant informa-tion. In most cases the victim possesses an abundanceof information indicating the imminence of the attack.Instead, the common wisdom holds that attacks achievesurprisebecause defenders cannot identify the relevantsignals amidst the noise, and because of cognitive ororganizational shortcomings (Betts 1982, 87-149; Kam1988, 7-212). In short, regardless of whether attackersare democratic or autocratic, they do not appear tobe able to keep their attacks secret; attacks achievesurprise because defenders are poor at evaluating in-formation.Even if we accept that the achievement of surprise isa function of the transparency of the attacker, there islittle historical support for the claim that democraciesare less able to conceal their intentions or impendingactions. There have been approximately 10 cases of sur-prise attack since the beginning of WorldWarII.20Twoof these attacks, the British-French-Israeli coalition'sattack on Egypt (1956) and the Israeli initiation of theSix Day War (1967), were carried out by democracies.There are not enough cases to make any statisticalclaims but we should note that democracies have madeup approximately one third of state-years since 1939,and therefore, one would expect on the basis of chancealone to see three surprise attacks by democracies inthis period. Therefore, democracies do not appear tobe less likely than nondemocracies to launch surpriseattacks.

Israel, France, and Britain planned the Suez War of1956 in such secrecy that even Eisenhower was sur-prised by the attackwhen it came (Betts 1982, 63-65).21Dayan, the Israeli Chief of Staff, engaged in a success-ful campaign of deliberate deception leading outsideobservers to believe that any attack would merely bean extended reprisal campaign. Meanwhile, the rele-vant decision makers in Britainjustified secrecy in starkterms: It is never agreeable to have to refuse, in the na-tional interest, information to the House of Commons.But it has to be done from time to time (Lloyd 1978,250). If democratic government officials believe thatthe national interest is at stake, they will sacrifice dis-closure to military necessity. Similarly,Israel achieved

surprise through deception in launching the Six DayWar (1967). Dayan, then the defense minister, publiclystated that Israel was in no position to reply to theblockade of the Strait of Tiran, that the Israeli armycould not remain mobilized for an extended period,that the army could fight successfully after suffering afirststrike, and that diplomacy must be given a chance,all in a successful attempt to lull the Arabs into a falsesense of security. Only 38 hours later Israel attacked(Betts 1982, 65-68; Van Evera 1999, 66-67). Nor doesthe ability of democratic governments to maintain se-crecy appear to be restricted to extreme cases of sur-prise attack. The United States kept its decisions forwar from the British before the War of 1812, LordGrey did not publicize his agreement to defend FrenchChannel ports prior to WorldWarI, and Roosevelt didnot reveal his agreements with Churchillprior to WorldWar II.Democratic politics are typically marked by the opendiscussion of differing opinions in multiple public fo-rums, but this characterization does not appear to holdwhen democratic leaders perceive a threat to the na-tional interest. In such circumstances the requirementfor transparency and consensus can be decisively sub-ordinated to the twin requirements of military success:secrecy and speed.InformationThe available evidence suggests that democracies can-not clearly reveal their levels of resolve in a crisis.There are two reasons for this. First, democratic pro-cesses and institutions often reveal so much informa-tion that it is difficult for opposing states to interpretit. Second, open domestic political competition doesnot ensure that states will reveal their private infor-mation.

Transparency may contribute little to peace be-cause a lot of information is not always good informa-tion. Simplybecause democracies provide a substantialamount of information about their intentions from avariety of sources does not mean that their opponentswill focus on the appropriate information or that theinformation will be interpreted correctly. In a crisiswith a democracy, the other state will receive signalsnot only from the democracy's appointed negotiatorsbut also from opposition parties, interest groups,publicopinion, and the media. Deciding which signal is trulyrepresentative is a difficulttask. Moreover, individualsfaced with an overwhelming amount of information arelikely to resort to mental shortcuts based on existingviews of the adversary or analogous situations in thepast to make sense of it. Information contradicting theaccepted wisdom is likely to be ignored and confirma-tory evidence will be highlighted. Additional informa-tion may, then, have a limited impact on perceptions(e.g., Jervis 1976). In short, the mistake has been toequate plentiful information with perfect information.If the information is plentiful, there is no reason tobelieve that states will come to a mutually acceptableagreement. On the other hand, if the information isperfect, then states may avoid war.

20 I have compiled the following list using Betts (1982) and Kam(1988): Germany's attack in Western Europe (1940); Germany's at-tack on the Soviet Union (1941); Japan's attack on Pearl Harbor(1941); North Korea's attack on South Korea (1950); China's entryinto the Korean War (1950); Israel, Britain, and France's attack onEgypt (1956); China's attack on India (1962); Israel'sattack on Egypt(1967); the Soviet attack on Czechoslovakia (1968); and the Arabattack on Israel (1973). I have excluded cases of surpriseattack in thecontext of an ongoing war based on the assumption that, regardlessof their regime type, once they are in a war states will enforce secrecyand try to achieve surprise as a matter of military necessity. Thereare, of course, several instances of democracies achieving surpriseduringwars.These include the British bombing of the Italian fleet inTaranto (1941), the D-day landings (1944), and the American assaultat Inchon (1950).21The fact that three democratic governments were involved in suc-cessful collusion is especially powerful evidence of the ability ofdemocracies to maintain secrecy.

598



16/19

AmericanPoliticalScience Review Vol. 97, No. 4

There is good evidence for these claims. BernardFinel and Kristin Lord (1999) have highlighted thenegative effects of transparency in seven case studiesof interstate crises between 1812 and 1969. They findthat open political systems do indeed provide a greatdeal of information, but its sheer volume either hasconfused those who observe it or has merely served toreinforce their priormisperceptions. In 1967, for exam-ple, Nasser was overwhelmed by the 'noise' of Israelidomestic politics and had enough information to seewhatever he wanted and confirm existing mispercep-tions about Israeli intentions (Finel and Lord 1999,334-35). Democracies may not be better at signalingtheir intentions, and even if they are, these intentionsmay be prone to misperception.In response, proponents of the informational storyargue that it is the signal sent by opposition partiesthat provides the most credible evidence of a state'sintent: If they supportthe administration,then the stateis committed, otherwise it is not (Schultz 2001, 95-97). There are two problems with this argument. First,there is powerful support for the claim that the generalpublic and opposition generally rally round the flagand support governments duringcrises.Kenneth Waltzneatly summarizes this finding: The first effect of aninternational crisis is to increase the President's popu-larstanding.One may wonder if this is so only when theresponse of the President is firm or he otherwise givesthe impression of being able to deal with the situationeffectively.... It is, in fact, not necessary to add suchqualifications to the statement (Waltz 1967, 272).22In-deed, Schultz notes that democratic governments thathave issued deterrent threats have received oppositionsupport 84% of the time (Schultz 2001, 167). More-over, democratic leaders can lead rather than followpublic opinion during international crises by control-ling what information reaches the public and by ex-ploiting the media. Reaching high office in a democ-racy rests, to a large degree, on persuading voters, andone would therefore expect democratic governmentofficials to be especially adept at shaping public opin-ion. What this means is that democracies may oftennot be able to signal their private information. Sincepublics and oppositions generally rally to the govern-ment's side or are persuaded to support the adminis-tration during crises, and hostile states know this tobe the case, opposition support is not an informativesignal.Second, in the few cases where opposition par-ties have spoken out against military action, demo-cratic governments have been prepared to take ac-tion nonetheless. In other words, when oppositionstatements should lead us to expect that a govern-ment would not be resolved on war, they have insteadbeen prepared to escalate disputes. Examples are nothard to find: (1) The Federalists opposed war withBritain in 1812, but Madison went to war nonetheless;

(2) Trumanwent to war in Korea despite the protestsof Senate Republicans; (3) the British Labour Partypublicly opposed action against Egypt in 1956, but theEden government plotted and executed an attack onEgypt with the governments of France and Israel; and(4) several Democrats publicly opposed the Gulf War n1990-91, but the Bush administration was determinedto act. In short, there does not appear to be a strongcorrelation between declarations by opposition partiesand decisions to avoid war.23In sum, the purported informational properties ofdemocratic institutions are unlikely to improve theprospects for peace. It is not clear that democraciescan reveal private information or that it will be inter-preted correctly,and even in cases where signaling andinterpretation are accurate there are reasons to doubtthat this will remove the cause of war.

CONCLUSIONThe causal logics that underpin democratic peace the-ory cannot explain why democracies remain at peacewith one another because the mechanisms that make upthese logics do not operate as stipulated by the theory'sproponents. In the case of the normative logic, liberaldemocracies do not reliably externalize their domesticnorms of conflict resolution and do not treat one an-other with trust and respect when their interests clash.Similarly, in the case of the institutional logic, demo-cratic leaders are not especially accountable to peace-loving publics or pacific interest groups, democraciesare not particularlyslow to mobilize or incapable of sur-prise attack, and open political competition offers noguarantee that a democracywill reveal private informa-tion about its level of resolve. In view of these findingsthere are good reasons to doubt that joint democracycauses peace.Democratic peace theorists could counter this claimby pointing out that even in the absence of a good ex-planation for the democratic peace, the fact remainsthat democracies have rarely fought one another. Inaddition to casting doubt

The flawed logic of democratic peace theory

Documents