International Organization Foundation

Intervention and Democracy

Author(s): Bruce Bueno de Mesquita and George W. Downs
Source: International Organization, Vol. 60, No. 3 (Summer, 2006), pp. 627-649
Published by: Cambridge University Press on behalf of the International Organization Foundation
Stable URL: http://www.jstor.org/stable/3877822 .
Accessed: 16/06/2014 22:59

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.

Cambridge University Press and International Organization Foundation are collaborating with JSTOR to
digitize, preserve and extend access to International Organization.

http://www.jstor.org

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
Interventionand Democracy
BruceBueno de Mesquitaand
GeorgeW. Downs

Abstract Recenteventshaveraisedquestionsabouttheextentto whichmilitary

interventionpromotesdemocracyandthedegreeto whichthisdependson thenature
of theintervener.
We arguethattraction on theseissuesis bestobtainedby focusing
on thepoliciesof thetargetstatethathave thegreatestimplications forthepolitical
survivaloftheintervening state'sleaderandthekindofgovernmental institutions
in
thetargetstatethatare mostlikelyto producethem.This perspectivegenerally-
although notalways-predictsthatthird-partymilitaryinterventionin civilwars,other
intra-or interstate
disputesand warswill lead to littleif anyimprovement, and all
too oftenerosionin thetrajectory of democraticdevelopment. Threehypotheses on
theimpactof third-party intervention by democracies,autocracies,and theUnited
Nationsare thentestedand strongly supportedagainsta counterfactual expectation
of whatthedemocratic trajectorywouldhave been in theabsenceof intervention.

The U.S. invasion of Iraq has rekindleddebate about the impact of militaryinter-
vention on democratic nation-buildingand, particularly,the extent to which it
depends on the characterof the intervener.Some leaders, such as French Presi-
dent Jacques Chirac, have argued thatmilitaryinterventionand democracy build-
ing is best advanced in states such as Kosovo when it is carried out by a broad
multilateralcoalition of democratic-in this case European-states. Others,such
as U.S. President George W. Bush, have implicitlydismissed the necessity of a
broad-based coalition and stressthe importanceof the unityof command thatcan
oftenbe most effectivelyachieved by a single democraticstate acting alone or in
concertwitha small coalition. Leaders fromthe developingworld and those closely
identifiedwithinternationalinstitutions,such as Kofi Annan, the secretarygeneral
of the United Nations (UN), argue that the greaterlegitimacy that the UN pos-

Webenefited from
greatly thewisecounselofFeryalCherif,MichaelGilligan,ShankerSatyanath,
andAlastair eachofwhomreadordiscussed
Smith, in depthearlier
versionsofthisstudy.Patrick
Reganwasalsoextremely helpfulinproviding
dataandguidance intheuseofthedataoninterven-
tionsthathehasmadeavailable.
Thestudywassignificantlyimproved adviceofthe
bytheinsightful
anonymous reviewers
(wewishwecouldthank them byname)andbyLisaMartin's ableandthought-
fulguidance.
Authorsoftencomplainaboutreferees;
wehavenothing butpraiseforthecontributions
theymade.Ofcourse, we aloneareresponsible
foranyerrorsandforall remaining in
shortcomings
thisinvestigation.

International 60, Summer2006, pp. 627-649

Organization
? 2006 by The IO Foundation. DOI: 10.1017/S0020818306060206

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
628 International
Organization

sessesprovidesitwitha variety ofadvantagesin comparison withindividualnation

states.
Surprisingly, giventhepoliticalsalienceof theseissues,thereis relatively little
researchon theextentto whichintervention the
facilitates process of democrati-
zationor the degreeto whichthisdependson the characterof the intervening
state.One groupof studiesdealingwiththeimpactof intervention generallysug-
gestthatdemocraticinterveners succeed in promoting democraticreformin the
shorttermbut thatthe targetstatesend up withunstablepoliticalsystems.'A
secondgroupof studiesfocusingon themorelimitedissue of theimpactof inter-
ventionbytheUnitedStatesis generallycriticalandhas produceda growingcon-
sensusthatAmericaninterventions do nottypicallylead to democracy.2
Less clear in both cases is why interventions have the effectthattheydo.
Much of thebroaderliterature as well as some of thosewho have examinedthe
recordof U.S. intervention play downtheimpactof democraticcommitment on
thepartof theintervener or issuesof strategic and
inadequacy argue that adverse
domesticconditions in thetargetcountry wouldoperateto overwhelm thepositive
efforts of anyintervener.3 This selectioneffectargument is also prominent as an
explanation forthedifficultiesconfronted bytheUN in ending hostilitiesand sus-
tainingpeace.4
Some scholars5pointto U.S. military interestsas a reasonthatAmericaninter-
ventionstendto failto lead to democratization, sometimes evenreversing it.6Oth-
ersmaintain thatthenegativeresults,at leastwithregardto U.S. efforts to promote
democracy, are due to an inherent contradiction betweentheidea of democratiza-
tionanditsimposition fromoutsidebyforce.7Stilla thirdgroupcontendsthatthe
emergingconsensusthatinterventions are an ineffective methodof promoting
democracyis overdrawn and unwarranted. Theyarguethata closerreadingof the
recordsuggeststhatundertherightcircumstances theUnitedStatesis an effective
agentforpromoting democracy or at leastforpromoting generally liberalregimes.8
Although theextantliteraturecontributes importantly tothegrowingunderstand-
ing of thelinkagesamonginterventions, domesticconsiderations in targetcoun-

1. See KegleyandHermann 1997,78-107;andGleditsch, andHegre2004.

Christiansen,
2. Lowenthal 1991.
3. See Huntington1991;andMuravchik 1991.
4. See Fortna2004b;andGilligan andStedman 2003.
5. See Karl1986;Forsythe 1992,385-95;andRueschmeyer, Stephens,andStephens 1992.
6. A consequenceofworking witha smallsetofcasesinwhich theoutcomes havebeendisappoint-
inghasbeenthetendency in theliteratureto emphasize theoverwhelming ofdomestic
significance
conditionson theground in thetarget forthefailure
stateas theexplanation ofdemocracy to take
an understandable
hold.Whilethisis certainly conclusiongiventhesmallnumber ofcasesandrela-
tiveabsenceofpositive outcomes, itis nomoredefensible from a statistical thanthecon-
standpoint
clusionthattheUnited Statesis somehow atfault.Theproblem indeterminate.
is basically
7. See Herman andBroadhead 1984;O'Donnel,Schmitter, andWhitehead 1986;Karl1990,1-23;
andWhitehead 1991.
8. See Meernik 1996,391-402;Hermann andKegley1998;Wantchekon andNeeman2002;and
EnterlineandGreig2003.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
 Intervention
and Democracy 629

tries,andpeacekeeping, we believethatitscontributiontounderstandingtheimpact
of interventionson democracygenerallyhas been limitedby a lack of theorizing
abouthow themotivesof theintervener shapesitsinstitution-building
aspirations
in thetargetstateandbyitsfailureto estimateadequatelythecounterfactualdegree
of democraticchange expectedin the absence of an intervention. We hope to
improveon bothof thesedimensions. In thepages thatfollowwe suggesta theo-
reticalexplanationofthelinkagebetweenthird-party and subsequent
intervention
democratization, identify
specifichypotheses focusedon comparisons to thecoun-
terfactualexpecteddegreeof democratization in theabsenceof an intervention,
and testthemagainsta broaddata set.

Theory:Nation-Buildingand theAgencyProblem

Ourtheoretical pointof departureforexplainingthepostintervention prospectsof

democratization are drawnfromBueno de Mesquitaand colleagues' selectorate
theory.9The authorspresenta theory in whicha leader'smotivation to retainpower
andtheinstitutional contextin whichhe or sheoperatesplaya majorrolein deter-
mininga state'spolicychoices.The governmental structureofeachpolityis defined
by itslocationin a two-dimensional institutional
space. One dimension, theselec-
torate(denotedas S), reflectsthenumberof peoplein a politywho have a say in
choosingleadersand,moreimportantly, have some chanceof gainingaccess to
thespecialprivilegesthateveryregimedolesoutto membersofthewinningcoali-
tion.The winningcoalition(denotedas W), theotherdimension, is thesubsetof
theselectorate whosesupportis essentialif a leaderis to remainin office.
Democraciesare generallycharacterized by a largeselectorate and a largewin-
ningcoalition,althoughthereare systematic differences in thesize ofthewinning
coalitionin different
typesofdemocracies.10Typically,systems withdirectly elected
presidentsfostera largerwinningcoalitionthando British-style parliamentary
democracies.These,in turn,relyon a largercoalitionthanmanyproportional rep-
resentationsystems. Autocraciesandilliberaldemocraciessometimes havesmaller
selectorates and alwayssmallerwinningcoalitionsthando democracies.Military
juntasand monarchies normally relybothon a smallselectorate and a smallwin-
ningcoalition,whilerigged-election autocraciescommonlyhave a smallwinning
coalitiondrawnfroma largeselectorate.
The keyfeatureof thetheoryforourpurposeslies in Bueno de Mesquitaand
colleagues'demonstration thatin equilibriumsmall-coalitionleadersprimarily rely
on theprovisionof privategoods to stayin office,whileleadersin largewinning
coalitionsystemscan mostefficiently maintaintheirhold on powerby providing

9. Buenode Mesquitaetal. 2003.

10. Becauseofdatalimitation,
we arenotabletoexploit
thesewithin-democracy
differences
here.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
630 International
Organization

publicgoods." The authorsalso showthatbecausetheloyaltyof coalitionmem-

bersto theincumbent leaderis greatestwhenthecoalitionis smalland theselec-
torateis large and is weakestwhenbothare large,it is easier forautocratsto
survivein officethanformonarchsandjuntaleaders,and it is hardestfordemo-
craticallyelectedofficials.
It followsthatas agentsof theirown domesticselectorswho can deposethem,
bothautocratic anddemocratic leadersengagedin external nation-buildingaremoti-
vatedto takeactionsthat,at themargin,maximizestheirpoliticalsurvivalpros-
pects back home. This means thatleaders of democracieswill be attractedto
interventions whichwill be perceivedby theirselectorates as directlytiedto the
provisionof a publicgood (forexample,safeguarding nationalsecurity)or thatis
likelyto effecta significant
portionofthepopulation(forexample,increasedaccess
to energyresources, protectingtraderoutes,and so on). Leadersofautocratic states
are morelikelyto focuson theacquisitionof resourcesthatcan be convertedto
privategoods fortheirnarrower selectorate.
Citizensof thecountry targetedforintervention,of course,are notpartof the
intervener's butactionstakenby thetarget-country's
selectorate, leadershipcan,
at themargin,enhanceor diminishtheprospectsthattheintervener's constituents
are satisfiedwiththe consequencesof theirleader's foreignpolicy.This means
thattheleaderof theintervening statepossessesa vestedinterest in ensuringthat
thetargetstatepossessesan institutionalstructurethatwillbe responsiveto his or
herinterests.

DemocraticInterveners

To see how theseincentivesare likelyto affecttheimpactof an intervention on

thedemocratic prospectsof a targetstateconsiderthedomesticpoliticalsurvival
problemof a democraticintervener. Reelection(of the individualincumbent or
theincumbent's party)dependson delivering policybenefitsto coreconstituents-
thewinningcoalition-thatare sufficient to makethecoalitionmemberspreferto
retaintheincumbent ratherthanchoosea rival.This meansthatleadersof democ-
racieswillbe attracted to interventionsthatwillbe perceivedbytheirselectorates
as directlytiedtotheprovisionofa publicgood(forexample,safeguarding national
security) orthatis likelyto effecta significant
portion ofthepopulation(forexam-
ple, increasedaccess to energyresourcesor protecting traderoutes).
Does thecreationofdemocracy in anothercountryas a publicgoodfallin either
of thesecategories?Some mightarguethatbecause thecitizensof a democracy
regardtheirformof government and itsassociatedfreedoms as publicgoods,they
will also regardtheirprovisionto a secondstateas publicgoods. However,it is
clear thatthe two actionsare not remotelyequivalent.The citizensof a demo-

11. Buenode Mesquita

etal. 2003;seeespecially
chaps.3-5.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
 Intervention
and Democracy 631

craticintervener receivedirectconsumption benefits fromlivingundera democ-

racyand fromexercisingthepoliticaland civil rightsthatcome withit. Theydo
notreceivethesedirectbenefits whentheirstatesuccessfully promotesdemocracy
abroad(forexample,theydo notvotein thestate'selection).This does notmean
thatthecitizensofa democratic intervenerplace no valueon thecreationofdemoc-
racyin anotherstate,butit does meanthattheyplace considerably less value on
its existenceabroadthantheydo at home.12The same is true,of course,in the
case of otherpublicgoods as well suchas thequalityof theenvironment, educa-
tion,defense,and publichealth.No statespendsnearlyas muchprovidingthese
goods abroadas theydo withintheirborders.How else are we to explainthedis-
tributionofforeign aid givingto relatively compliant recipient dictatorsratherthan
to themostimpoverished or mostdemocratic ThirdWorldstates?13
Giventheseincentives, we believethattheleadersof intervening democracies
will view thegovernmental structureof a targetstateinstrumentally. Thatis, the
value thattheyassignto it will have less to do withits intrinsic characteristics
thanwiththeimpactthatits policies are likelyto have on theleader'spolitical
survival.Whilethisdoes notruleout thepossibilitythatdemocraticleaderswill
supporta democracyin thetargetstate,it does meanthata necessarycondition
fordoingso is thattheleaderof theintervening statewill have to believethatthe
resultantdemocracyis likelyto adoptpolicies thatare regardedas desirableby
the intervening leader's winningcoalition.Given thatthe intervening stateis a
democracy, thesewill usuallyinvolveissues suchas security, trade,and access to
resources.
Generally,an intervening democraticstateis mostinterested in reversingthe
policiesthatprecipitated its interventionin thefirstplace. In somecases, thecre-
ationof a democracyis viewedas thebestway to accomplishthis.For example,
one of the mainreasonsthatthe designof democraticinstitutions was a major
U.S. goal in connection withitspostwaroccupationsof GermanyandJapanwas a
beliefthattheircitizenswouldnotpoliticallysupportthereemergence of militar-
ismin thesestates.In mostcases, however,establishing a trulyliberaldemocracy
possessesa numberofdrawbacksas a meansforensuring thatthetargetstatewill
adoptpoliciesthatare consistent withthepreferences of theintervener.
The biggestsuchdrawbackis uncertainty. The costsofmilitary intervention are
usuallylargeenoughthata failureto obtainthepromisedpolicygoals in theafter-
mathwillthreaten thepoliticalsurvivaloftheleaderoftheintervening state.Hold-
ing a fairand competitive electionis an inherently uncertainprocessthatplaces
thatpoliticalsurvivalat considerablerisk,especiallyin unstablestatesor those
tornby civil war betweenlarge segmentsof the population.In a freeand fair
election,thereis no guaranteethata candidatesympathetic to thepolicygoals of

12. Goldsmithand Posner2005, 218. Goldsmithand Posnermakemuchthesamepointin connec-

tionwiththeirdiscussionof therelationship
betweendemocracyand cosmopolitancharity.
13. Alesina and Dollar 2000, 33-63.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
632 International
Organization

theintervener will evenbe running muchless be victorious, and evenifthatper-

sonis electedhe or shemaynotbe successfulin pushingthrough thepolicychange
desiredby theintervening leader'swinningcoalition.
Thereare rivalagencyproblemsat workhere.The electedleaderof thenew
democracyis preeminently the agentof his or her domesticwinningcoalition
and can only surviveby implementing somethingclose to its preferred policy
agenda-an agenda thatis likelyto involvepublic goods such as securityand
foreignpolicy,naturalresourcepolicy,and tradepolicy.To the extentthatthe
leaderof thetargetstateoperatesas theagentof theintervening state'swinning
coalition,the leaderundermines his or her survivalunlessthe goals of the two
constituencies areidentical,whichtheyareunlikelyto be becauseeach is primar-
ily interestedin theprovisionof publicgoods withinitsborders.
Knowingtheinherent conflictof interests thatexists,we believethatthedem-
ocraticleaderoftheintervening statewillchoosethesaferand less costlystrategy
ofsupporting theestablishment (orcontinued existence)ofan autocracy ora rigged
electiondemocracy in thetargetstatewitha smallwinning coalition.Sucha regime
willbe less concernedwithdelivering to itscitizenspublicgoods-related policies
thatmightconflictwiththegoals of theintervener and moreconcernedwithpri-
vate goods (forexample,foreignaid and military assistance)thattheintervening
leaderwillbe willingto supplyin exchangeforpermitting it to achievethepolicy
goals thatmotivated theintervention in thefirst place. The regimein thetargeted
statemay well be givensome of the trappings of democracyforsymbolicpur-
poses,butitwilllacktheinstitutional arrangements thatareessentialforauthentic
partycompetition and thebasis forthecreationof a genuineliberaldemocracy
(forexample,freedomof speechand assembly,theexistenceof a freeand inde-
pendentmedia,fairelections).Collectively, theseshortcomings operateto ensure
thatthe leadershipin the targetedstatewill be able to deliverdesirablepolicy
concessions.
To see how thislogic operatesin practice,considerthecharacteristic aftermath
of Frenchmilitary intervention. France,as withall NorthAtlanticTreatyOrgani-
zation(NATO) members,advocatesdemocracyas thepreferred formof govern-
ment.In fact,Frenchcoloniesroutinely participated in FrenchAssemblyelections,
sendingfreelyelecteddelegates.Yetin thepostcolonialera,France'scolonieshave
shifted awayfromdemocracytowardpersonalist, autocratic rule.Francehas been
and continuesto be a frequent military intervener, especiallyin its formercolo-
nies.TheFrench,forinstance, haveactivelycommitted troopstopreserve orrestore
peace in suchcountriesas Lebanon,Chad,Angola,and Laos. Lebanon,once the
mostdemocratic country in theMiddleEast otherthanIsrael,saw peace restored
followinga multilateral democraticintervention (includingFrance,the United
States,Italy,andtheUnitedKingdom)in theearly1980sbutdid notexperiencea
seriouseffort torestoredemocratic sovereignty to thepeopleofLebanonfornearly
a quarterofa century. Instead,withthetacitapprovalof theinternational commu-
nity,Syria,a decidedlynondemocratic state,maintained its freereinoverLeba-
nese affairs.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
 Intervention
and Democracy 633

In Chad throughmuchof the 1980s,Franceacted aggressivelyto protectthe

government of HisseinHabr6againstLibyanincursions.Habr6had declaredhis
intention to promotedemocracyin Chad on numerousoccasionsand yet,as is so
oftenthecase, failedto do so. Instead,Chad made minorstridesin addingsome
trappings of democracy(forexample,expandedfactionalpartycompetition) and
even held a legitimate multipartyelectionin 1992 afterHabr6had been deposed
by themilitary. However,Chad quicklyreturned to themoreauthoritarian end of
thepoliticalspectrum. Habr6'ssuccessor,forinstance,recentlyhad theconstitu-
tionrevisedso thathe may succeedhimselffora thirdterm.Thus Frenchinter-
ventiondid little-perhaps nothing-to promotedemocraticreformin Chad,
althoughit did secureand preservestrongeconomicties.Franceremainsone of
Chad's most importanttradingpartnersand the leading importerof Chadian
products.
Woulda genuinely democratic Chad be beneficialforFrenchinterests? It is dif-
ficultto say,butit is clear thatthereare stronganticolonialsentiments in Chad
and thatethnicallyand culturally muchof thecountryhas a naturalaffinity for
Libya-the verycountryagainstwhichFrance intervenedto protectHabrd's
regime-ratherthanforFrance.French(and American)businessinterestsand
investments are, then,probablymoresecureundera tightly ruledChadianstate
thantheywouldbe if thecountry werea full-fledged democracy.
The storyof Chad can be retolda hundredtimesfordifferent targetsand for
differentdemocratic interveners.It is muchthesame as thestoryof theU.S. lib-
erationofKuwaitfromIraq in 1991,whichwas accompaniedbypromisesofdem-
ocraticreform butonlymodestactualpoliticalchange-again a fewtrappings of
democracyratherthansubstantivechange. Indeed, in circumstancessuch as
Chad's or Kuwait'sitprobablyis notantidemocratic (thoughwhether itis morally
acceptableor notis a separatequestion)fora democratic intervenerto bypassthe
opportunity to erectdemocratic institutionsin thetargetstate,noris thefailureto
do so peculiarlyan attribute of any particularintervener. Quite the contrary, if
democratic-ormoreappropriately, republican government-isexpectedto reflect
thewishesof its voters,thenbuildingdemocracyelsewhereagainsttheinterests
of domesticvotersis bothantidemocratic anditis contrary to thereelectionambi-
tionsofan incumbent leaderorparty.Therefore, if,as theselectorate
theory claims,
leadersaremotivated toretainoffice, we shouldnotexpectdemocratic interveners-
exceptunderthetwo unusualconditionswe mentioned-toadvancedemocratic
institutionswhereverand whenevertheyintrudethemselvesintothe affairsof
anotherstate.

NondemocraticInterveners

Now considerthecalculationsof nondemocratic The survivalof their

interveners.
leadersdependson theirabilityto deliverrent-seeking to themem-
opportunities

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
634 International
Organization

bersof theirsmallwinningcoalitionthatsustainthemin office.Suchrentscan be

extracteddomesticallyfromthoseoutsidetheircoalitionor externally fromthe
citizensofa targetstatefollowing Giventheselimitedgoals,
intervention.14
military
autocraticintervenersareunlikely tobe concerned withthelong-term internalaffairs
of statesin whichtheyintervene exceptto ensuretheautocrat'sopportunities for
materialgain as a resultof intervention.15Consequently,theyare unlikelyto pay
the costs of maintaining a presencein the targetstateforlong and unlikelyto
have muchof an impacton thefutureevolutionof democratization in thetarget
state.To theextentthattheydo concernthemselveswithpoliticalchangewithin
thetargetstate,autocratsare likelyto opposedemocratic changebecauseit could
producea nationalistic movementto recoverthe lost territory or the resources
extractedby theintervener. Therefore, autocraticintervenersare likelyeitherto
as theyfoundthemor to disassembledemocraticinsti-
leave politicalinstitutions
tutionsthatmay interfere withtheirabilityto extractresourcesfromthe target
state.
Thus whiletheselectorateargument tellsus thatthereare fundamental differ-
ences in thedomesticgoals of democraticand nondemocratic leaders,theyshare
an interestin avoidingthecreationof democratic in targetstates.This
institutions
is notbecause theirforeignpolicygoals are thesame.As we have noted,demo-
craticinterveners are usuallypreoccupiedwithensuringthatthetargetstatealter
itsforeignpoliciesso thattheyare compatiblewiththeirown and providepublic
goods to theircitizensat home.In the case of nondemocratic interveners,their
goal is usuallyto delivera one-timesurrender of landor exploitableresources.In
eithercase, theselectorate argument tellsus thatthesegoals can be bestattained
by imposingor maintaining a government in thetargetstatein whichtheleaderis
sufficientlyunconstrainedat home(thatis,W is small,democracy is weakorabsent,
and thereare significant on politicalcompetition)
restrictions thathe or she will
be able to satisfythe policy objectivesof the intervening state(or coalitionof
states)withouthavingtojeopardizinghis or herpoliticalsurvival.
Of course,thevictormay notneed to make any changesin thetargetstate's
institutionsiftheinstitutionsof thedefeatedstatealreadygivethenew leaderthe
latitudeto ignoredomesticpressuresto revisethestatusquo, eitherto gainpolicy
endsor privatebenefits. Because moststatesthatexperienceintervention are not
democratic to beginwith,we can expectthatmuchof thetimethereis littleor no

14. Buenode Mesquitaet al. 2003,chap.9, arguesthatmodern-day autocratsaremorelikelyto

seekterritorial
acquisitions
following thanaredemocrats.
interventions
foreign Wesay"modern-day"
autocrats
becauseinearlier
periods, inthemedieval
especially agesandbefore, autocraticinterveners
weremorelikelyto seekthesustained payment oftribute to territorial
in addition than
acquisition
appearstobe trueforthepost-World WarII world.
15. TheSovietexperiencemaybe anexception tothisclaimgiventheAmerican containmentpol-
icythatencouragedtheSovietleaderstostretch
their zonebymaintaining
security friendlyregimesin
states
thatsharedborderswiththeSovietUnionandwithitsNATOadversaries. In thosecaseswhere
continued of resources
extraction is soughtby theintervenerfromthetarget, autocracy shouldbe
expectedto suppresstheprospectsof democratization.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
 Intervention
and Democracy 635

changein institutions. Butwhentheydo occur,we expectthechangesto be small,

moresymbolicthana dramaticchangein institutional form,and theyfavorfree-
ing the incumbent in the targetstateto make choices thatfavorthe intervener
ratherthana broadportionof his or herdomesticpopulation.
Whileautocratsand democratic sharea commoninterest
interveners in prefer-
ringto deal withnondemocratic targetstates,thefactthattheformer tendsto be
interestedin a one-timeacquisitionof territoryor resourcesratherthanreshaping
theforeignpolicygoals of thetargetstateleads us to believe thatthattheyare
likelyto be less institutionally
ambitiousthandemocratic As a conse-
interveners.
quence,democraticinterveners mayactuallydo morein thelong termto restrict
meaningful democratic reform thanwill nondemocratic interveners.

UN SecurityCouncil as Intervener

The case fortheUN as an intervener is no less bleak. If UN interventions were

authorized by a vote of theGeneral itis
Assembly, possible that interventions
might
embodytheambitions ofthemanyweakstatesagainsttheinterference ofthevery
strong.Thatmightlead theGeneralAssembly'smajority to consistentlypromote
democraticreformfollowingintervention.'6 However, when it comes to choices
regardingintervention or peacekeeping,the appropriate venue in the UN is the
Security Council. Fortna has shown that the centrality theSecurityCouncilis
of
notincompatible witheffective UN peacekeepingbutshouldthislead us to expect
thatthe SecurityCouncil will also promotedemocraticchangein the statesin
whichit intervenes?"17
Given our earlierarguments thatneitherdemocraticnornondemocratic inter-
venersgenerally possessany incentiveto establisha democracy in thetargetstate,
thereis littlereasonto be optimistic thattheywill do so collectivelyas members
of the UN SecurityCouncil.However,Voeteneven suggeststhatin thoserela-
tivelyrarecases whenone or even severalpermanent memberswouldpreferto
establisha democracy, thatthenatureoftheSecurity Council'scollectivedecision-
makingrulesare likelyto undermine theidea.'8
Voetenarguesthatall thingsare rarelyequal in theSecurityCouncil.He con-
tendsthatfora stateto pursueunilateral action-the outsideoption-is politically
costly.19The politicalcosts of unilateralactioncan be avoidedthroughSecurity
Councilendorsement but at thepriceof compromising on policygoals or ideal
points.Because the Security Council consists of a mixof democratic andnondem-
ocraticstatesthiscreatestheexpectation thattheresultant outcomewillprovidea

16. Voeten2004, 729-54.

17. Fortna2004a, 269-92.
18. Voeten2001, 845-58, and 2005, 527-58.
19. Voeten2001 and 2005.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
636 International
Organization

mixture of publicgoodsbenefits fortheconstituencies of democratic membersof

theCounciland privategoods fortheconstituencies of thenondemocratic mem-
bers.The selectoratetheorytellsus thatthissort of outcome is most consistent
witha relatively smallW systemwhoseconstituency will allow it to makepolicy
compromises in exchangeforprivategoods. It is not thesortof outcomethatis
likelyto resultfromthecreationof a liberaldemocratic statesincetheselectorate
theory tellsus thatitscitizensarelikelyto be unwilling supplytheprivategoods
to
thatare necessaryto securetheparticipation of thenondemocratic members.20
Giventhatseveralsubsections ofArticle2 oftheUN Charter precludeUN inter-
vention inthedomesticpoliticalaffairs ofmember states;andgivenVoeten'sresults,
it is difficultto imaginecircumstances in whichthepermanent membersof the
SecurityCouncilwouldbe alignedin strongsupportof usingUN peacekeepersto
promotedemocratization.21

Data and Measurement

To summarize, we hypothesize thatfollowingforeignintervention thedegreeof

democracyin thetargetstate(1) does notimproveand maygetworserelativeto
comparablestatesthatdo notexperienceintervention whentheintervener is the
UN; (2) convergeson promoting symbolic,externalmanifestations of democracy
withoutthearrayoffreedoms foundin all full-fledgeddemocracieswhentheinter-
veneris itselfa liberaldemocracy (meaningthatnearlydemocratic statesthatexpe-
riencedemocratic becomeless democratic
intervention relativeto comparable states
thatdo notexperienceintervention whilestatesthatare notremotely democratic
mayadd democratic trappingsfasteror at thesamerateas comparablestatesthat
do notundergointervention); and (3) is no differentand maybe worsefollowing
interventionby an autocratthanis thedemocratization experienceof comparable
statesthatdo notundergointervention.
We evaluateour theoryby examiningtheimpactof military interventions on
thelevel of democratization thathas evolvedin thewake of civil warsand intra-
statedisputes,militarizedinterstate disputes,and interstate wars between1946
and 2001. We includethirdpartiesas interveners iftheymadea physicalmilitary
commitment in thearenaof fighting whetherin theformof peacekeepersor as
activeparticipants in a dispute.Interveners includeindividualthird-party states,
groupsof states,and theUN between1946 and 2001. We comparetherecordof
subsequentdemocratization of statesthathave experiencedthird-party interven-
tionwiththosethathavenot,thereby creating theopportunity to estimate thecoun-

20. It is worthnotingthatsuchan outcomeis perfectly in keepingwithFortna'sresultsregarding

theUN's peacekeepingsuccess.Civil warscreatesecuritycostsforborderingdemocraticandnondem-
ocraticstatesalikeandfortheinternational to whichtheybothbelong.Successfulpeace-
organizations
keepingefforts providebenefits of both.
fortheconstituencies
21. Voeten2001 and 2005.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
 Intervention
and Democracy 637

terfactualdemocratization expectationsin the absence of intervention forthose

statesthatdid undergoa military intervention.
We includeall countriesin our analyseswhileconstructing a dummyvariable
MILYN, thedetailsof whichare explainedshortly, to distinguish cases thatexpe-
rienceda militaryintervention (MILYN = 1) fromthosethatdid not (MILYN = 0).
We relyon severaldata setsto identify interventions and interveners. Regan has
assembleda widelyused datasetofcivilwarrelatedmilitary interventions.22With
some codingmodifications (such as fillingin missingdata on whenan interven-
tionbegan or ended),we includeall of the cases he classifiedas interventions.
Fortnahas amassed data on UN peacekeepingefforts.23 Using her startand
end dates,we code all cases thatshe coded witha scoreof 1 or 2 as a military
intervention(MILYN = 1). In hercoding,a 1 meansunarmedmonitors and a 2
indicatesarmedforces.24 Additionally, we code as interveners any statewitha
MilitarizedInterstate
Dispute(MID) hostility level scoreabove 1; thatis, anystate
thatactivelyparticipated in a militarized disputeprovidedit is notcoded as the
initiator
in theMIDs data.We use 1946 as thestartdateforour analysesas it is
thefirstfullyearin whichtheUN was in operationand therefore had theoppor-
tunityto intervenein disputes.The nonintervention cases in our data includeall
country-yearsforwhichthenecessarydata are availablebetween1946 and 2001
such thatthecountryyeardoes notinvolvean intervention eventor theflowof
democratization opportunities-defined below-aftersuchan event.
In some instancesof intervention therewas more thanone intervener; for
instance,themajorityofUN military interventions also includedmilitary interven-
tionby one or moremembersof the SecurityCouncil.Our analysescode each
intervenerseparatelyin each event.

DependentVariables
We examinethreeinstitutional variables:(1) DEMOCRACY, measuredas thetwenty-
one-point scale from
resulting calculating the difference betweenthe PolityIV
Democracy scoreand the IV
Polity Autocracy score foreach country,recodedto
fallbetween0 and 1, withhighervaluesindicating of
greaterdegrees democracy;
(2) PARTICIPATION COMPETITION, a six-point scale takenfromPolityIV. We recode
thisvariableso thatits maximumscoreis 1 and 1 indicatesa competitive party
system,withlowervalues indicatingless and less participation and competition;
and (3) Bueno de Mesquitaet al.'s variableW, a five-point indexof coalitionsize
(and here labeled as COALITION SIZE), with 1 being thevalue of systemswiththe
largestcoalition and 0 the smallest.25The details forthe measurement of W are
foundin theAppendix.

22. Regan2000.
23. Fortna,
2004a.
24. Ourresults
do notchangeifwe onlycountas interventions codesas 2.
thosecasesthatFortna
25. Bueno de Mesquitaet al. 2003, 134-35.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
638 International
Organization

TABLE1. Bivariatecorrelations
amongdependentvariables

PARTICIPATION
DEMOCRACY COMPETITION COALITION SIZE
Variables (t+ 10) (t+ 10) (t+10)

DEMOCRACY(t+10) 1.000
PARTICIPATION COMPETITION (t+10) 0.935 1.000
COALITIONSIZE (t+O10) 0.857 0.830 1.000

Note: N = 4,369.

Each dependentvariableis made up of the scorefortherelevantinstitutional

variabletenyearslaterthantheyearof observation foreach of theindependent
variablesand thelaggeddependent variable.A ten-yearlead providesampletime
forinstitutional Table 1 reportsthebivari-
effectsto takehold and be observed.26
ate correlations
amongthesedependentvariables.

Variables
Independent

Observations arecountry-years. A variableMILYNis constructed thatis coded as 1

foreach countryfortheyearthatit experiencedan intervention and foreach of
thenexttenyears.For instance,therewas a U.S. military intervention in Grenada
in 1983. MILYN forGrenadais 0 before1983, 1 each yearfrom1983-92, and 0
foryearsafter1992 (assumingthattherehas notbeen anothermilitary interven-
tionin theinterim), whilethedependent variableforGrenadain 1983 is thevalue
oftherelevantinstitutional variablein 1993 orthechangein theinstitutionalvari-
able between1993 and 1983,in 1984 thedependent variableis therelevantinsti-
tutionalvariable for 1994 or its change between 1984 and 1994, and so on
throughout thetimeseries.27In cases withoutan intervention and in thosecases
thatexperiencedan intervention, forthe yearsbeforethe intervention, or more
thanten yearsafterthe intervention, MILYNis coded as 0. Thus MILYNrecords
thata country has had an interventionwithintenyearsof theyearof theobserva-
tion.This allows us to evaluatethelong-term consequencesfora country's polit-
ical institutionsthatfollow its experienceof an outside,third-party military
intervention fromthetenthyearthrough thetwentiethyearaftertheintervention.

26. Wereplicate
ouranalyses lead.To conserve
as wellwitha five-year space,thefive-year
results
are notreportedhere,butare availableat (http://www.nyu.edu/gsas/dept/politics/faculty/bdm/
Theyaresubstantively
Intervention_and_Democracy/). equivalenttotheten-yearresults
wereport in
thebodyofthetext.
27. Forthereplication
basedon five-yearperiods, areadjusted
all variables accordinglytoreflect
theappropriate
timelinesuchthatthedependent variable fiveyearslaterthanthe
is alwaysobserved
variables.
independent

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
 Intervention
and Democracy 639

We createthreedummyvariables,UN, USA,and DEMOCRAT, referring respec-

tivelyto theUN, theUnitedStates,and any additionaldemocracy(forexample,
France,UK, Belgium,and India) thatparticipated as a military intervener in a
givencountryin a givenyear.We assigna 1 to thisvariable,as withMILYN,for
each yearup to andincludingtenyearsaftertheintervention, scoringit as 0 in all
othercircumstances.
The distinctionbetweentheUnitedStatesand otherdemocratic interveners,as
discussedin moredetaillater,is made primarily forinterpretive purposesrather
thanfortheoretical reasons.We anticipate thatdemocratic interveners facingcom-
parablecircumstances respondcomparablyin alteringthetarget'sinstitutions to
createmalleable,policy-compliant regimesthathavelittlemorethanthesymbolic
trappings of democracy. Because theUnitedStatesis themostfrequent and most
powerfuldemocratic intervener andbecauseitis theintervener thatreadersofthis
journalhavethebestprospectof influencing, we highlight U.S. patterns ofaction.
The statisticaleffectofautocraticinterveners on thedependent variablesis esti-
matedby thecoefficient forMILYN.The effectof otherinterveners is thesumof
thecoefficients forMILYNplus thedummyvariableor variablesrepresenting the
specificinterveners.In thiswaywe can directly assesstherelationship betweeneach
typeofintervener andsubsequent movement inthetarget statetowardorawayfrom
democracy, a centralconcernof ourhypotheses. For cases thatdid notundergoa
militaryintervention,thevalueofeachofthesedummyvariablesis,ofcourse,zero.
Our concernis withthedemocratization aftermath of foreigninterventions and
nottheircauses or ex antestatedobjectives.To gaugetheimpactofinterventions,
however,we mustcontrolforthecounterfactual expectedchangein institutions in
theabsenceof an intervention. This is a trickyproblembecause it is likelythat
thereare selectioneffectsinfluencing whenand whereinterventions occur28We
attempt to correctforthesegeneralselectioneffectsby includingcountry-specific
fixedeffectsto correctforcharacteristics of each country thatdo notchangeover
time.We correctforanyseculartrendsin democratization by controlling forthe
yearof theobservation. The controlelementsformthebaselineagainstwhichwe
estimatethemarginalimpacton democratization createdby different interveners.
Those marginaleffectsrepresent thetestsof ourthreehypotheses.
The basic structure of themodelswe use to testourargument is as follows:

Institutional
indicatort+10

= a + b1 (dependent
variableto)
+ b2 (MILYN = 0, 1) + b4 (USA = 0, 1)

+ b5 (UN= 0, 1) + b6(DEMOCRAT= 0, 1) + b7 YEAR

+ b8 ... bn (Country-specific
fixedeffects)+ e

28. See Gilliganand Stedman2003, 6, 13; and Fortna,2004a, 14.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
640 International
Organization

The designofthetestsensuresthatwe can see theeffectof specificinterveners

and comparethateffectto thecounterfactual expecteddegreeof democratization
in the absence of an intervention. The effect of democraticinterventions, for
instance,is simplythesumoftheregression coefficientsfor MILYN + DEMOCRAT.
Addingthe coefficient forthe UnitedStatesto thesetwo coefficients gives the
effectof U.S. interventions on democratization. Positivecoefficients are associ-
ated withan improvement in thedegreeof democracy over the time horizon we
assess whilenegativecoefficients of
indicatean erosion democracy. The coeffi-
cientshavea naturalsubstantive interpretationas well.Because each ofthedepen-
dentvariableshas an upperboundof 1,thecoefficients fromourregression analyses
forthedummyvariablesforintervention can be interpreted approximately as the
of
percentageincreaseor decreasein thedegree democracypredicted fora coun-
tryas a resultofinterventionbythenationororganization identified the
by dummy
independent variable.Thussmallercoefficientsareassociatedwithnominalchanges
in thedegreeof democracy(up or downdependingon thesignof thecoefficient)
while largercoefficients indicatemore substantialand meaningful institutional
change.Withthemodelspecification we use, we can assess themarginaleffects
of a UN intervention, Americanaction,an intervention
a unilateral by anydemoc-
racyoranyautocracy, combinations include,forinstance
as wellas multilateral that
a joint UN-U.S. or joint UN-Democracyintervention. Whenwe summarizeour
findings we will drawattention to thesemarginaleffects.

Analysisof Intervenersand InstitutionalChange

Table 2 summarizesthe country-specific fixedeffectsanalysesthatallow us to

assess keyfeatures ofinstitutionalchangesthatoccuraftera military intervention.
Recall thatwe hypothesizedthatfollowingforeign intervention the degree of
democracyin thetargetstateis likelyto be unchanged or to have declined when
theUN intervenes as comparedto thecounterfactual expectation for democratic
changeif no intervention had occurred.When autocratsintervene we expectto
observean initialshockbutrelatively littleifanymeaningful long-term impacton
thetargetstate'spoliticalinstitutions.Aftertenyearswe expectto see thetarget
statepursuingdemocracyat a pace thatis comparableto or somewhatslowerthan
is observedforequivalentstatesthatdid notexperienceintervention.
Whendemocrats intervene,theprediction is morecomplicated becauseitdepends
on thedegreeto whichthestatehad democratized at thetimeof intervention. If
thetargetwas relativelydemocraticat theoutsetof theintervention, say above
0.7 or 0.8 on our 0 to 1 normalizedPolityDemocracy-Autocracy scale, thenwe
anticipatea decline(or at leastno increase)in thetarget'sdegreeof democracy.29

29. These values are equivalentto scores of 4 to 6 on the nonnormalized,

+10 to -10 Polity
scale.A valueof0.50onournormalized
Democracy-Autocracy toa valueof0 on
scaleis equivalent

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
 and Democracy 641
Intervention

TABLE2. Interveners tento

changeinpoliticalinstitutions
and institutional
yearslater,country-specific
twenty model
fixed-effects

PARTICIPATION
DEMOCRACY(t+10) w: COALITION SIZE COMPETITION(t+10)

Dependentvariable(tO) 0.143 0.131 0.204

(0.017) (0.013) (0.017)
0.000 0.000 0.000
MILYN -0.075 0.000 -0.010
(0.024) (0.019) (0.016)
0.002 0.998 0.537
DEMOCRAT
INTERVENER -0.045 -0.066 -0.042
(0.033) (0.028) (0.022)
0.171 0.019 0.056
USA 0.176 0.090 0.094
(0.039) (0.031) (0.026)
0.000 0.004 0.000
UN -0.064 -0.062 -0.026
(0.052) (0.036) (0.034)
0.221 0.079 0.455
YEAR 0.007 0.002 0.005
(0.000) (0.000) (0.000)
0.000 0.000 0.000
CONSTANT - 14.24 -4.254 -8.721
(0.517) (0.358) (0.341)
0.000 0.000 0.000
N 4,487 7,111 4,487
Fixed effects 146 191 146
F 162.09 50.92 169.08
R2 0.253 0.235 0.410

Note:Cellentries
listfirst
theregression error
thenitsstandard
coefficient, inparentheses, bythestatistical
followed
ofthecoefficient.
significance

This is so because thedemocratic intervener will wantto avoid theelevatedrisk

thatthetargetmightquicklybecomea full-fledged democracythatwouldthenbe
unableto deliverthepolicyconcessionssoughtby thedemocraticintervener. If,
the
however, target is farfrom for
beingdemocratic, example, below a value of
0.5, then a democratic will
intervener leave theinstitutional
arrangements largely
unchangedbecausetheyare alreadyconsistent witha regimethatwillprovidethe
cooperation itrequires,butitmaywell add symbolictrappings to satisfytheweak

thestandard-10 to +10 scale. Valuesabove 0.50 on thenormalizedscale,thenarecountriesthat,on

balance,are moredemocraticthanautocratic.Values below 0.50 indicatecountrieswithmoreauto-
craticthandemocraticgovernance.Full-fledged democraciesscore +10 on thePolityindex,or 1 on
typicallytreatscountrieswithscoresabove 6; thatis, 0.80 when
ournormalizedscale. The literature
normalized,as beingdemocratic.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
642 International
Organization

TABLE3. Percentagechangein democracytenyearsaftera military

an intervention
comparedto cases without
intervention

PARTICIPATION
Intervener DEMOCRACY
(t+10) SIZE (t+10)
COALITION COMPETITION
(t+10)

No intervention +7.45% +2.42% +4.68%

Autocrat -0.01** +2.42# +3.69#
UnitedNations -6.39** -3.82* +1.12#
Democrat -4.55*** -4.14** -0.50**
UnitedStates +13.01** +4.82# +8.90**
UN+Democrat - 10.93*** - 10.38*** -3.07**
UN+ UnitedStates +6.63# - 1.41# +6.33#

Note:Allpredicted
percent include
changes theten-year duetothesecular
gainsexpected
positive indemoc-
trend
ratization bythevariable
captured YEAR. arerounded
Percentages tenth
tothenearest ofa percent.
Statistical
signif-
oftheintervention
icancethattheeffect from
changeis different
on democratic theeffect on
ofno intervention
democratic
change:*p < .10;**p< .05;***p< .01;#nostatistical
difference.

preferences of its constituency fordemocratization abroad.Such trappings, asso-

ciatedwithan increasein thedemocracyscore,fallat a level thatexcludesinsti-
tutionalchangesthatraise theriskof thetargetquicklybecominga full-fledged
democracyofthesortthatwouldbe counterproductive to theintervener'sprimary
policygoals. Thus we expect that when such symbolicchanges are made, they
leave thetargetwitha democracyscoreof less than0.80.30
To summarize, thelowerthedemocracyscoreat theoutset,themorelikelythat
a democraticintervener will raise it modestlyand symbolically relativeto com-
parable states
that do notexperience intervention;the higherthedegree ofdemoc-
racy before an intervention,the more likelythat theintervener will unravel sub-
stantivelyimportant of
aspects democracy, leavingsymbolic trappings butno more
thanthat.
Table 2 presents resultswhileTable3 showstheexpectedpercent-
thestatistical
in
age change democracy over the ten-year periodfollowingintervention based on
theresultsreportedin Table 2.

30. One mightobjectthatthedemocratictrappings installedby democratic interveners are a form

of cheaptalkand thereforeare notcredibleand so reapno domesticbenefits fortheintervener. How-
ever,sucha viewassumesthatintervener'sconstituentscan distinguishex antebetweengenuinereforms
thatputthecountryon thepathto fulldemocratization (forexample,as in post-WorldWar II Ger-
many)frominstitutional changesthatprove,expost,to be meresymbolicgestures(forexample,prom-
ised reform in Kuwait).Who,forinstance,can say withcertainty whether recentIraqi electionsare a
harbingerof greaterdemocraticchangein thefuture or thattheywillproveto be an emptyexercisein
democratic symbolism? Evenin thosecases whereitquicklybecomesapparent thatinstitutional
changes
areonlysymbolic,votersin theintervening statehaveto untanglean informational agencyproblemof
determining whetherthefauxinstitutionwas meaninglessfromtheoutsetor corrupted by thetarget
statepoliticianswho tookover.In such cases, theintervening state'sleaderprobablywill scapegoat
thetargetstatepoliticians.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
 Intervention
and Democracy 643

Withregardto thefirst hypothesis, thatis, theeffectswhentheUN is theinter-

vener,one-tailedtestsof significance are appropriate because we predictedthe
directionof change.Those generallyshowreductions in democratic reform rela-
tiveto thecounterfactual base case of no intervention. For changein democracy
as measuredby Polity,theprobability thatthe relationship is due to chance is
0.007; forcoalitionsize, it is 0.050; and forparticipation competition, 0.17. As
expected,targetsof intervention by theUN fareno betterand generallydo worse
thanwouldhavebeenexpectedhadtheynotsuffered an intervention. Table3 indi-
catesthatUN interventions results,on average,in a 6.40 percentdeclineoverten
yearsin democracy(as measuredby Polity)relativeto comparablestatesthatdid
notexperienceintervention.31 Coalitionsize apparently shrinks by about3.82 per-
centoverthesame tenyears,whileparticipation competition does notmeaning-
fullychange.We find,then,thatthefirst hypothesis is supported.
The thirdhypothesis, thatautocratic interveners eitherdo notalterthecounter-
factualexpectedchangein democratization or slow democratization downcom-
paredto thecounterfactual expectation, is also supported by theevidence.In the
case of coalitionsize and participant competition thereis no difference between
thecounterfactual changein democracyand theexperienceof thosetargetedfor
intervention byautocrats. In thecase oftheestimatebased on Polity'sdemocracy-
autocracyindex,thereis a statistically significant declinein democracy relativeto
thecounterfactual. The smallcoefficient in Table 2 mightmisleadone intothink-
ing thatthereis no effect,but thereal difference betweenthoseundergoing an
autocraticintervention andthosenotexperiencing intervention, as seenin Table 3,
is thatthelatterenjoya 7.45 percentseculartrendimprovement in theirextentof
democracy, whilethetargets ofautocrats experienceno changewhatsoever (-0.01
percent);theyjust stayas theywere.Thus autocratseitherdo notalterthecoun-
terfactualexpectedpace ofdemocratic reform basedon seculartrendsortheycoun-
teracttheseculartrend,leadingto no improvement whatsoever in theinstitutional
arrangements of theirtargets.
The secondhypothesis is perhapsthemostimportant and mostcomplex.It pre-
dictsthatdemocraticinterveners retardtherateof democraticreform relativeto
theexpectedseculartrendin theabsenceofan intervention, whentargets arefairly
democratic, butmayimprovethe symbolismof democracywhenintervening in
especiallynondemocratic countries. The hypothesis makesno distinctions among
democraticinterveners, althoughit does distinguish expectations as a function of
thepriorinstitutional circumstances of theirtargets.
At firstglance Tables 2 and 3 mightlead one to inferthatthishypothesisis
supportedwithregardto mostdemocraticinterveners butthatit is refutedwith

31. Thisand comparabletenyearcalculationsaremadeas follows.The predicted changein democ-

racyis equal to 10(YearCoefficient) + UN Coefficient
+ MilYN Coefficient = 10(0.00744)-0.07456 -
0.06380 = 0.06396basedon thestatistical at (http://www.nyu.edu/gsas/dept/politics/
outputreported
faculty/bdm/Intervention_and_Democracy/) and roundedto threeplaces in Table 2.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
644 International
Organization

TABLE4. Selectioneffects Do targetsof U.S.

amongdemocraticinterveners:
intervention
end up moredemocraticthantargetsofotherdemocratic
interveners
?

COALITION COALITION PARTICIPATION PARTICIPATION

DEMOCRACY DEMOCRACY SIZE SIZE COMPETITION COMPETITION
Variables (tO) (t+ 10) (tO) (t+ 10) (tO) (t+ 10)

U.S. as intervener
Mean 0.463 0.538 0.474 0.520 0.570 0.607
Standarddeviation 0.295 0.300 0.269 0.263 0.168 0.161
N 292 205 338 237 292 205
Otherdemocracy
as intervener
Mean 0.526 0.554 0.525 0.549 0.596 0.616
Standarddeviation 0.389 0.377 0.310 0.297 0.246 0.240
N 200 164 223 169 200 164

respectto theUnitedStates.The coefficients associatedwiththeUnitedStatesare

invariablypositive,relativelylarge comparedto othercoefficients, and statisti-
callysignificant, whilethecoefficientsforotherdemocratic interveners are gener-
allysignificantly negative.Thereis,however,a selectioneffectoperating thatneeds
to be considered.
Recall thatthesecondhypothesis statestheexpectation thatdemocraticinter-
venersare motivatedto createregimesin thetargetstatethatpossess the trap-
pingsbut not the substanceof liberaldemocracy.Whatthismeans in a given
instancedependson thetarget'sstarting point.This raisesthepossibility thatthe
difference betweenthecoefficients fortheUnitedStatesand forotherdemocratic
interveners is theconsequenceof selectionregarding thechoiceof statesin which
to intervene. If theUnitedStatesintervenes in statesthatare initiallymoreauto-
craticthanthosein whichotherdemocraciesintervene, thenit would have less
occasionto reducethelevel of democracythatpresently exists;thiscouldexplain
thedifference intheaveragelevelofdemocracy eventually achievedbytheirrespec-
tivetargetstates.
To learnwhetherthiswas thecase we createTable 4, whichshowsthemean
and standarddeviationforthe threelagged dependentvariables(DEMOCRACY-
AUTOCRACY, COALITION SIZE, and PARTICIPATION COMPETITION) split by whether
the intervenerwas the United States or was some otherdemocracy.It also
showsthesamesummary statistics
forthedependent variables,measuredtenyears
afterthelaggeddependentvariable,again withthesamplessplitby whetherthe
UnitedStatesor some otherdemocracywas theintervener.
Table 4 confirms thata selectionbias exists.The UnitedStates,on average,
targetsless democraticstatesforintervention thando other(mostlyEuropean)

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
 Intervention
and Democracy 645

democracies.32 The meansotherthanthatof to forparticipation competition are

significantly different. This suggeststhattheUnitedStateswill have less incen-
tiveon averageto reducedemocracyin thetargetstatethanotherinterveners and
comparatively moreincentiveto increasethe symbolictrappingsof democracy
thatare preferred (if weakly)by its winningcoalition.Note thatthemediannor-
malizedPolityscore,forinstance,is only0.35-tilted well to theautocraticside
of thescale-when theUnitedStatesintervenes, butit is 0.70-tilted strongly on
thedemocraticside-when otherdemocratsintervene.
It is also worthnotingthattenyearsafteran intervention, thedegreeofdemoc-
racyin statesthatwereU.S. intervention targetsandin statesthatwerethetargets
ofintervention bysomeotherdemocracy (ordemocracies) arestatisticallythesame.
The relevantt-statistics and probabilitiesare as follows:forthedifference in the
meansforU.S. targetsand targetsof intervention by a different democracybased
on the Polityindexfordemocracytenyearsafteran intervention, t = -0.568,
p < 0.60; forcoalitionsize tenyearsaftertheintervention, t = - 1.009,p < 0.40;
and forparticipation competition,t = -0.429, p < 0.70. Thatis, all democratic
interveners, as expected,tendto restrict the extentof democraticfreedomsfor
theirtargets to comparablelevelsof symbolism without democratic substance.The
absolutescoresof democratictargetstatesare also revealing.The standardcut-
pointin theliterature fordefiningdemocraciesbased on thePolityindexis equiv-
alentto a scoreof 0.80. Contrastthiswiththemeandemocracyscorestenyears
aftera democratic interventionwhichareonly0.538 (U.S.) and0.554 (otherdemoc-
racies),well below the standardthresholdand roughlyequivalentto between0
and +1 on thePolity-10 to +10 scale; thatis, thescoresforsuchcountriesas
Zambia and Liberiain thelate 1990s. The same is truefortheothertwo depen-
dentvariables,highlighting thefactthatneithertheUnitedStatesnorotherdem-
ocraticinterveners createdemocratic polities.
How does the UnitedStatesbehaverelativeto otherdemocracieswhentheir
intervention targetsare comparable?Figure1 providesa way to see thatcompar-
ison. Because we wantto examineand comparecomparableinterventions by the
UnitedStatesalone and by otherdemocratsalone,we redefine thevariableDEM-
OCRAT thatappearsin Table 2 so thatit onlyincludesdemocratic interveners other
thanthe UnitedStates.By doing so we createa coefficient forUSAthatwith
MILYN-ratherthanalso summedwiththecoefficient forDEMOCRAT-measures
theimpactof U.S. interventions on subsequentdemocraticreform. Additionally,
in thisanalysiswe excludethecases of UN intervention as we are onlyinterested
in the comparisonbetweenthe UnitedStatesand otherdemocracieswhenthey
instigateinterventions withouteach other.

32. The sourceof thisselectioneffectis beyondthescope of thisstudy.It maybe a consequenceof

differencesin the militarycapabilities of the United States compared to other democratic interveners,
or to the more extensive ties to formercolonies among other democratic interveners,or to some other
factors that we leave for futureresearch.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
646 International
Organization

. 1

. .6 .8-

o-
USA
DEMOCRAT

.4
.2
o0

0 -.2-

0 .05 .10.15.20.25.30.35 .40.45.50 .55.60 .65.70.75.80.85 .90.95 1.0

butnotthesubstanceof
fosterthesymbolism
FIGURE1. Democraticinterveners
in
democracy targetstates

To correctfortheobservedbias in selectingtargets, we splitour data set into

threeseparategroupsforanalysis:one in whichthedemocracyscoresforthetar-
getsatthetimeofintervention wereat orbelow0.35; at orabove0.70; andbetween
0.35 and 0.70. The 0.35 threshold, recall,is themedianscoreof a U.S. targetof
intervention and 0.70 is themedianscorefora targetof a democraticintervener
otherthantheUnitedStates.Naturally, thereare cases ofAmericanand of other
democraticinterventions in each of the threegroupings.We save thepredicted
democracyvalueswhentheUSAis theintervener in each setand separately when
a democracyotherthantheUSAis theintervener. These predicted values are the
basis of thecomparisons in Figure1.
The fixed-effects regression analysisfromwhichFigure1 is constructed evalu-
ates thechangein democracyoverrollingten-year periods, focused now on the
ten-yearfirst-difference betweenthedemocracyindicator and its value a decade
later.So, ratherthanthedependentvariablebeingthedemocracyscoretenyears
lateras in Table 2, it is now thechangein thatscore.This is mucheasierto inter-
pretgraphically thanthestructure used in Table 2.
The independent variablesare MILYN,USA,and DEMOCRAT, plus a controlfor
yearas beforeandcountry fixedeffects.Because we are now lookingat theactual
changein democracy, thereis no reasonto include a laggeddependentvariable.
We also excludecases in whichtheUN was a participant as thatinevitably means
participation by some democraticinterveners and so muddies the separationwe
seek to achievehere.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
 and Democracy 647
Intervention

In orderforoursecondhypothesis to be supported, we shouldobservecompa-

rablepatternsof democratic changeas a function of thetarget'sinitialdemocratic
conditionsregardlessof whichdemocraticintervener was involvedin the case.
The verticalaxis in Figure1 plotsthemeanchangeovertenyearsin thenormal-
ized Polityscore,whilethehorizontal axis locatesthatmeanchangeaccordingto
each initialdemocracyvalue amongthe set of targets.As is evidentfromthe
figure,33whencomparabletargets areassessed,thereis no difference in thenation-
buildingimpactwhether theintervener is theUnitedStatesor someotherdemoc-
racy.Whatis more,as thefigure also shows,evenwhenthetrappings ofdemocracy
are improved, theachievedlevel of democracydoes notreachor exceedthecom-
monlyused and relativelyweak threshold fordefining democracy, let alone the
upperboundthatsignifiesa genuine,full-fledged democratic polity.

Conclusions

While leadersof intervening statesfrequently assertthatthedemocratization of

thetargetstateis one of theirmaingoals,thisarticlehas presented evidencesug-
gestingthatthisgoal is rarelyachieved.More oftenthannotleadersof interven-
ingstatesarefacedwiththechoiceof trying to satisfythepolicyinterests of their
constituents to remainin officeor promoting the democraticaspirationsof the
citizensin targeted states.Because thecitizensin a transformed democratic target
stateare likelyto have different policypriorities than those of the intervener's
winningcoalitionbackhome,theintervening statewilltendto pass up theoppor-
tunity to establish a liberaldemocracy in favor of establishingan autocraticor
rigged election politythatit expects will be more willingto agreeto implement
thepoliciesit desires.As a result,intervention does littleto promotedemocracy
and oftenleads to its erosionand thesubstitution of largelysymbolicreforms, a
factthatis clearfordemocratic interveners fromFigure1.
This outcomeis notparticularly surprising in thecase of autocratic interveners,
butin thecase of democraticinterveners manyare likelyto findit to be disap-
pointing. Whileitis tempting to blamethefailureofthequalityof leadershipthat
existsin connection withdemocratic interveners, we believethatit is moreindic-
ative of the limitations of democraticinstitutions themselvesand the natureof
democraticrepresentation. Democraticleadersare constitutionally chargedwith
beingagents of theirdomestic constituencies and their voters'policypriorities are
rarely identical with thoseof citizens in the target state.The fault,if there is one,
lies less in themotivesof thedemocratic executivethanin thepolicypriorities of
democratic votersand theincentivescreatedby democratic institutions.

33. Confirmedby the regression analysis at (http://www.nyu.edu/gsas/dept/politics/faculty/bdm/

Intervention_and_Democracy/).

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
648 InternationalOrganization

Appendix

We estimatethesize of W as a compositeindexbased on thevariablesREGTYPE, XRCOMP

XROPEN,and PARCOMP.WhenREGTYPEis notmissingdataand is notequal to codes 2 or 3
in thePolitydataset,so thattheregimetypewas nota military or military/civilianregime,
we awardone pointto W. WhenXRCOMP, thatis, thecompetitiveness of executiverecruit-
ment,is largerthanor equal to code 2 thenanotherpointis assignedto W. An XRCOMP
code of 1 meansthatthechiefexecutivewas selectedby heredity or in rigged,unopposed
elections,suggestingdependenceon few people. Code values of 2 and 3 referto greater
degreesof responsiveness to supporters,indicating a largerwinningcoalition.XROPEN, the
openness of executive recruitment, contributes an additional pointto W if theexecutive is
recruitedin a moreopensetting thanheredity (thatis, thevariable'svalueis greater than2).
Finally,one morepointcan be contributed to theindexof W if PARCOMP, competitive-
nessofparticipation, is coded as a 5, meaningthat"therearerelatively stableand enduring
politicalgroupswhichregularly competeforpoliticalinfluence at thenationallevel"34We
divideby the maximumvalue of 4. The normalizedminimumvalue, then,is 0 and the
maximum is 1. It is evidentthattheprogression from0 to 1 to 2, up to 4 is notlinear,as the
indicatorsuggestsand so theindexis bestthought of as a logarithmic scale thatestimates
theorderof magnitude of thewinningcoalition.
DEMOCRACY, PARTICIPATION COMPETITION, and W are all normalizedso thattheyfall
between0 and 1. Highervalues reflectrespectively moredemocracy, a morecompetitive
partysystem,a largerselectorate, and a largerwinningcoalition.

References

Alesina,
Alberto, AidtoWhomandWhy?Journal
andDavidDollar.2000.WhoGivesForeign of
EconomicGrowth5 (1):33-63.
Bueno de Mesquita,Bruce,AlastairSmith,RandolphM. Siverson,and JamesD. Morrow.2003. The
Logic ofPoliticalSurvival.Cambridge,Mass.: MIT Press.
Enterline,
Andrew, ofImposed
andJ.MichaelGreig.2003.BeaconsofHope?TheImpact Democracy
onRegionalPeace,Democracy andProsperity. attheAnnual
Paperpresented ofthePeace
Meeting
ScienceSociety(International),
November,AnnArbor, Mich.
War,andCovertAction.JournalofPeace Research29 (4):385-95.
David P. 1992.Democracy,
Forsythe,
Fortna, Page.2004a.DoesPeacekeeping
Virginia andtheDura-
Intervention
KeepPeace?International
tionof Peace AfterCivil War.International 48 (2):269-92.
StudiesQuarterly
--. 2004b.Interstate CausalMechanisms
Peacekeeping: Effects.
andEmpirical Politics56
World
(4):481-519.
Gilligan,Michael,and StephenJohnStedman.2003. WhereDo thePeacekeepersGo? International
StudiesReview5 (4):37-54.
Gleditsch, LeneSiljeholm
NilsPetter, andHhvard
Christiansen, Hegre.2004.Democratic Mil-
Jihad?
itaryIntervention
andDemocracy. at theAnnualMeetingof theInternational
Paperpresented
StudiesAssociation,March,Montreal.
Law. New York:Oxford
JackL., and Eric A. Posner.2005. The Limitsof International
Goldsmith,
Press.
University

34. Gurr1997,18.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions
 Intervention
and Democracy 649

Gurr,Ted Robert.1997.PolityII: PoliticalStructures andRegimeChange.1800-1986.[computer file]

(studyno. 9263). ICPSR, AnnArbor,Mich.: Inter-University ConsortiumforPoliticaland Social
Research.(www.icpsr.umich.edu).
Herman,Edward,and FrankBroadhead.1984. Demonstration Elections:U.S. StagedElectionsin the
DominicanRepublic,Vietnam, and El Salvador.Boston:SouthEnd Press.
Hermann,MargaretG., and CharlesW. Kegley.1998. The U.S. Use of MilitaryIntervention to Pro-
moteDemocracy:EvaluatingtheRecord.International Interactions24 (2): 91-114.
Huntington, Samuel P. 1991. The ThirdWave:Democratization in theLate Twentieth Century.Nor-
man:University of OklahomaPress.
Karl,Terry.1986. ImposingConsent:ElectoralismVersusDemocratization in El Salvador.In Elec-
tionsand Democratization in LatinAmerica,1980-1985,editedby Paul Drakeand EduardoSilva,
9-36. San Diego: University of California.
. 1990. Dilemmasof Democratization in LatinAmerica.Comparative Politics23 (1):1-21.
Kegley,CharlesW., Jr.,and MargaretG. Hermann.1997. PuttingMilitaryIntervention intotheDem-
ocraticPeace: A ResearchNote. Comparative PoliticalStudies30 (1):78-107.
Lowenthal,AbrahamF 1991. Exporting Democracy:The UnitedStatesand LatinAmerica-Themes
and Issues. Baltimore:JohnsHopkinsUniversity Press.
Meernik,James.1996. UnitedStatesMilitaryIntervention and thePromotion of Democracy.Journal
ofPeace Research33 (4):391-402.
Muravchik, Joshua.1991.Exporting Democracy:Fulfilling America'sDestiny.Washington, D.C.: Amer-
ican Enterprise Institute
Press.
O'Donnel,Guillermo, PhilippeSchmitter,andLaurenceWhitehead, eds. 1986. Transitions
fromAuthor-
itarianRule: Comparative Perspectives.Baltimore:JohnsHopkinsUniversity Press.
Regan,Patrick.2000. CivilWarsand ForeignPowers:OutsideIntervention inIntrastateConflict. Ann
Arbor:University of MichiganPress.
Rueschmeyer, Dietrich,EvelynHuberStephens,andJohnStephens.1992. CapitalistDevelopment and
Democracy.Chicago:University of ChicagoPress.
Voeten,Erik. 2001. OutsideOptionsand the Logic of SecurityCouncilAction.AmericanPolitical
ScienceReview95 (4):845-58.
. 2004. ResistingtheLonelySuperpower:Responsesof Statesin theUnitedNationsto U.S.
Dominance.JournalofPolitics66 (3):729-54.
. 2005. The PoliticalOriginsof the UN SecurityCouncil'sAbilityto Legitimizethe Use of
Force.International Organization59 (3):527-57.
Wantchekon, Leonard,and Zvika Neeman.2002. A Theoryof Post-CivilWarDemocratization. Jour-
nal of TheoreticalPolitics14 (4):439-64.
Whitehead, Laurence.1991.The Imposition ofDemocracy.In Exporting Democracy:TheUnitedStates
and LatinAmerica,editedby AbrahamLowenthal,234-60. Baltimore:JohnsHopkinsUniversity
Press.

This content downloaded from 185.44.78.129 on Mon, 16 Jun 2014 22:59:37 PM

All use subject to JSTOR Terms and Conditions