Review: The Missing Variable: Institutions and the Study of Regime Change

Author(s): Richard Snyder and James Mahoney

Reviewed work(s):
Red Sunset: The Failure of the Soviet Union by Philip G. Roeder
Democratic Experiments in Africa: Regime Transitions in Comparative Perspective by
Michael Bratton ; Nicholas van de Walle
Source: Comparative Politics, Vol. 32, No. 1 (Oct., 1999), pp. 103-122
Published by: Ph.D. Program in Political Science of the City University of New York
Stable URL: http://www.jstor.org/stable/422435
Accessed: 06/04/2010 12:10

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=phd.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

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.

Ph.D. Program in Political Science of the City University of New York is collaborating with JSTOR to digitize,
preserve and extend access to Comparative Politics.

http://www.jstor.org
 Review Article
The Missing Variable

Institutionsand the Study of Regime Change

Richard Snyder and James Mahoney

Philip G. Roeder,Red Sunset: The Failure of the Soviet Union, Princeton,Princeton

UniversityPress, 1993.

Michael Bratton and Nicholas van de Walle, Democratic Experiments in Africa:

Regime Transitionsin ComparativePerspective, New York, CambridgeUniversity
Press, 1997.

Institutionshave takencenterstage in comparativepolitics. Scholarsstudyingsuch var-

ied topics as nationalism,public policy, and economic performancehave increasingly
cast institutionalfactorsin a leadingexplanatoryrole.' This surgeof interestin institu-
tionalanalysishas been especiallynotablein the comparativestudyof politicalregimes.
In their efforts to explain the varied dynamics and performanceof new democratic
regimes resulting from the "thirdwave" of democratization,scholars have focused
extensively on political institutions,such as electoral laws, constitutionalrules, and
partysystems.2
Althoughinstitutionalfactorscommandgreatinterestamongthose studyingthe per-
formanceof new democracies,they have curiouslyplayeda far less significantrole in
explainingthese regimes'origins.3The studyof transitionsto democracyduringthe last
decade has been dominatedby voluntaristanalysesthat focus on contingentleadership
choice. Such analysesview transitionsas open-endedprocesses of strategicinteraction
and tend to overlookhow political institutionsshape these processes.4A similardeem-
phasis of institutionscharacterizedearlierwork on the originsof authoritarianregimes.
Those analyses focused mainly on socioeconomic structuralvariables,either ignoring
institutionsaltogetheror treatingthem as epiphenomenalmanifestationsof macrostruc-
tural forces.5 Institutions have thus been a missing variable in theories of regime
change.
This omission is puzzling because regime change fundamentallyinvolves institu-
tional transformation.Regimes are the formal and informalinstitutionsthat structure
politicalinteraction,and a changeof regimeoccurswhen actorsreconfigurethese insti-
tutions.We shouldexpectregimeinstitutionsto havean importantimpacton the capaci-
ties and behaviorof incumbentswho seek to defendthem. Similarly,we shouldexpect

103
ComparativePolitics October 1999

regime institutionsto influence the strategiesof challengers who seek to transform

them. Hence a focus on institutionalfactorsmight offer valuableinsightin understand-
ing, notjust how regimesperform,butalso why they change.
This article assesses how institutionalvariables could enrich theories of regime
change. It examines two recent studies of regime transformationthat self-consciously
focus on institutionalfactors,Philip G. Roeder'sRed Sunset and Michael Brattonand
Nicolas van de Walle'sDemocraticExperimentsin Africa,and comparesthem to JuanJ.
Linz andAlfred Stepan'sProblemsof DemocraticTransitionand Consolidation.6These
works form an appropriatecomparisonnot only because they sharea focus on institu-
tional variables,but also because they exemplify distinctperspectiveson how institu-
tions can be incorporatedinto the studyof regimechange.
These three studies show that institutionalanalysis offers significant leverage in
addressinga core problemof regime change. Why do incumbents,who have vested
interestsin a regime'ssurvival,fail to preserveit? The issue of incumbentfailurehas
not been satisfactorilyanalyzed in previous studies, which often explain incumbent
behaviorin terms of skill or luck or as a consequenceof underlyingstructuralforces.
By contrast,the books underreviewhighlighthow incumbentsact in variedinstitutional
contextsthat shape and constraintheirbehaviorin characteristicways. Analysis of the
contexts of incumbent decision making helps explain why incumbents fail to save
"their"regimeand in some instancesactuallytake stepsthatundermineit.
A focus on institutionalvariablesalso helps accountfor the success of challengersin
transformingregimes.These books illustratehow the strengthsand strategiesof opposi-
tion groupsare shapedin crucialways by the very regimeinstitutionsthey seek to trans-
form. Furthermore,they show how the effects of old regime institutionson opposition
behaviorcan carry over to the posttransitionperiod,therebyinfluencingprospectsof
consolidatingnew regimes.
The shift to institutionalvariablesmarksan importantdeparturefrom the volun-
taristand structuralapproachesthat have long dominatedthe study of regime change.
However,noninstitutionalvariablesassociated with earlierapproachesshould not be
abandoned.Rather,a focus on institutionsoffers new opportunitiesto integratevolun-
taristand structuralperspectives.

The Rules of Regime Change

Red Sunsetillustrateshow institutionshelp explain incumbents'failureto defend their

grip on power.Roederseeks to accountfor the demise of the Soviet Union by showing
how informalpoliticalrules-what he calls the "constitutionof Bolshevism"-inhibited
policy innovationand institutionaladaptation.Althoughthese rules initiallyhelped sta-
bilize the Soviet regime, they later contributedto its breakdownby preventingleaders
fromimplementingreformsin the face of mountingpressuresfor change.

104
 RichardSnyderand James Mahoney

Roeder makes a valuablecontributionby illuminatinghow old regime institutions

constrainreformistincumbents.His analysis focuses on a common strategicdilemma
faced by reformers.They are often bound by the very rules they seek to change, and
these rules are enforcedby antireformincumbentsprivilegedby the statusquo. To suc-
ceed, reformersmust find a way to use the rules of the statusquo to subvertit, which
involvesthe difficulttask of "inducingpoliticalactorsempoweredby an existingconsti-
tutionalorderto changethe rulesof the game"(p. 233).
Incumbentcleavagesoverthe questionof reformhavebeen a familiartheme in stud-
ies of democratization.7 However,the voluntaristunderpinningsof most previousanaly-
ses resultedin a tendencyto treatincumbentcleavagesas riftsthatemergedand unfold-
ed in a fluid,institution-freecontextof extremecontingency.Roeder'sstudybreaksnew
groundby situatingsuch cleavagesandthe strategicinteractionsthatresultfromthem in
the contextof constraintsimposedby old regimerules. He shows how analysisof these
constraintsoffers fresh insights into the varied forms taken by intraregimecleavages
and helps explain why they so often can not be resolved in ways that avert regime
breakdown.Hence Red Sunsetmakes an importantcontributionby showingthat a cru-
cial stage of regime change, the failureof reformers,is not necessarilythe fortuitous
resultof contingentchoices.This pivotaleventmay insteadbe a predictableoutcomeof
rule-governedbehaviorinducedby old regimeinstitutions.

Institutional Foundations of the Old Regime To explain the Soviet leadership's

inabilityto achievereform,Roederfocuses on informalrules that structuredleadership
accountabilityand elite recruitment.Threecore institutions-reciprocalaccountability,
job slot rules of elite recruitment,and balanced leadership-contributed to regime
breakdownby posing barriersto reform.8Rules of reciprocalaccountabilitygoverned
the relationshipbetweenthe Soviet leadershipand what Roedercalls the "selectorate"
(state bureaucratsand party cadres responsible for selecting leaders); each had the
powerto appointand removethe other.Roeder(pp. 42-58) tracesthe rootsof reciprocal
accountabilityto the Bolsheviks'policy of "colonizing"key bureaucraticagencies by
cooptingtheircadresinto the centralcommitteeof the party.Reciprocalaccountability
became increasinglyinstitutionalizedafterStalin'sdeathin 1953 and formedthe basis
of a two tier politicalstructurethatpromotedstabilityby providingchecks againstarbi-
trary,discretionaryleadership.To survivein office, leadersin the first tier had to build
andmaintainsupportcoalitionsamongbureaucratsin the secondtier.
Althoughthey helpedstabilizethe SovietregimeafterStalin'sdeath,checks on lead-
ership discretionimposed by the rules of reciprocalaccountabilityalso createdrigid
barriersto institutionalreform.In orderto achieve theirpolicy goals reformistleaders
hadto buildsupportcoalitionsin the selectorate.However,the compositionof the selec-
torate,which was determinedmainly by a job slot recruitmentrule guaranteeingposi-
tions to occupants of specific bureaucraticoffices, made it extremely difficult for
reform-oriented leadersto find allies (p. 56).

105
ComparativePolitics October 1999

Thejob slot norm of recruitmentprivilegedbureaucratswith especiallystrongpref-

erencesfor defendingthe statusquo: employeesof so-called"irontriangle"agencies of
early industrializationand state building (the party,economic ministries,and coercive
apparatus).Althoughthe leadershipexerted some influence over appointmentsto the
selectorate,the pool of candidatesfromwhich they could choose was restrictedto these
iron triangleagencies. Hence the institutionof reciprocalaccountabilitycementedthe
leadership'sdependenceon antireformbureaucrats.Accordingto Roeder,"thecontinu-
ing constraintof the irontrianglelimitedthe leader'sopportunitiesto shiftpolicy priori-
ties" (p. 236).
The institutionof balanced leadershipfurtherconstrainedattemptsto reform the
politicalsystem.After Stalin'sdeathnew rules were establishedto solve the problemof
cycling between individualand collective leadershipby reinforcingthe latter.These
rules included clear demarcationand specification of the general secretary'spowers,
mutual checking to limit the power of each member of the leadership,especially the
generalsecretary,and "proceduralism" to reinforcecollectiveresponsibilitythroughsta-
ble norms for interactionamong leaders and provisionof "earlywarning signals" of
threatsto collective governance(p. 98). The rules of balancedleadershippreventeda
general secretaryfrom staffing the politburowith a loyal team of followers.Like the
verticalchecks againstleadershipdiscretionimposed by reciprocalaccountability,the
horizontalchecks of balancedleadershipconstrainedopportunitiesfor policy innovation
and politicalreform.Takentogether,these informalrules of the Soviet regimeresulted
in "institutionalized
stagnation"(pp. 119-43).

Regime Institutions and Failure of Reform A focus on regimeinstitutionshelps pre-

dict the tasks confrontingreformers.Roedershows how reformersin contextsdefined
by reciprocalaccountabilityface two characteristicchallenges: expandingthe selec-
torateand transformingreciprocalinto hierarchicalaccountability.Hence he explains
thatreformslaunchedby MikhailGorbachevin the late 1980s focused on wideningthe
selectorateto undercutthe monopolycontrolof iron trianglebureaucrats.Roeder also
explainswhy Gorbachevsoughtto strengthenthe expandedselectorate'sunilateralcon-
trol overantireformistleaders:to changereciprocalinto hierarchicalaccountability.
Old regime institutionsalso help account for the failureof incumbentsto achieve
reform.Roeder shows how effortsto enlargethe selectorateand restructureleadership
accountabilityrequiredreformersto build a coalition that simultaneouslymaintained
supportamong old guardelites in the selectorateand expandedthe politicalprocess to
bringin new groups.Sustainingsuch a coalition"placeda premiumon the generalsec-
retary[Gorbachev]holdingthe centerpositionin the coalitionspace"(pp. 228-29). Yet,
as the coalition grew to include newly enfranchised groups, this center position
changed.The institutionalsettingin which Gorbachevcarriedout his reforminitiatives
thus defineda difficultstrategiccontext:successfulreformdependedon holdingtogeth-
er a growing coalitionthat, because of its shiftingcenterand heterogeneouscomposi-

106
 RichardSnyderand James Mahoney

tion, faced the constantthreatof defectionof one wing or both.The constraintsimposed

on Gorbachevby the rules of balancedleadershipfurtherhinderedhis abilityto manage
the dilemmaof coalitionmaintenance.The coalitionultimatelyfell apartaftera "double
defection"by both the iron trianglerightand reformistleft (pp. 239-44). Accordingto
Roeder,this failureof reformsealedthe old regime'sfate.

Limitations of Intraregime Analysis While Roeder'sanalysis shows how old regime

institutionspreventedthe Soviet leadershipfrom achievingreform,it is less successful
in explainingwhy leaders had to implementreforms in the first place. It sheds little
light on the social transformations(such as the rise of a new diversifiedintelligentsia)
thatcreatedpressuresfor reform.Yet,accordingto Roeder,these pressures,in the guise
of whathe vaguelycalls "newsocial forces"(p. 211), play a key role in his explanation,
becausethey ultimatelyforcedSovietleadersto face the choice of constitutionalchange
or regimecollapse.Althoughthe failureof incumbentsto achievereformwas a decisive
event in the fall of the Sovietregime,this failurewasjust one episode in a largerprocess
involvingotheractorsand dynamicsnot analyzedby Roeder.
This limitationin the analysisstems from the kinds of institutionson which Roeder
focuses. He explicitlyrestrictedhis studyto "institutionaldynamicswithin the authori-
tarianstate"and excludedinstitutionsthat mediaterelationsbetween state and society
(p. 251). Roeder'sstudyshows how analysisof rulesthatgovernleadershipaccountabil-
ity and elite recruitmenttells a lot aboutthe behaviorof regime incumbentsand their
capacitiesto implementreforms.However,a focus on the institutionalconstraintsunder
which regime incumbents labor may not help explain the behavior of extraregime
actors,such as oppositiongroups,becausethey are not necessarilyboundby those con-
straints.Admittedly,oppositiongroups played an insignificantrole in the Soviet case.
Yet,because Roeder(pp. 8-13, 250-53) boldly claims to offer a "new institutionalism
of authoritarianism" that could be employed"aroundthe globe,"this idiosyncraticfea-
ture of the Soviet case does not justify his neglect of how regime institutionsaffect
extraregimeactors.
Institutionalanalysis offers students of regime change powerfultools in explain-
ing the strengths and strategies of extraregimeactors. However, it is necessary to
look beyond internalregime rules to institutionsthat connect incumbentsto social
actors.

A New Focus on Regime Type

A distinguishing feature of comparative regime analysis has been its taxonomic

sophistication. Starting with efforts during the 1950s to conceptualize totalitarian
systems, scholars have elaborateda complex set of categories for classifying regime
types and subtypes.9 Because of the dominance of structural and voluntarist

107
ComparativePolitics October 1999

approaches,however,work on regime change has tended to neglect these categories

as explanatory variables. Consequently, regime type has played a minor role in
explainingregime change.
As studentsof regime transformationincreasinglyfocus on the institutionalcon-
texts in which incumbentsand challengers interact,they encounternew opportuni-
ties to harness regime categories as an explanatory variable. Democratic
Experiments in Africa illustrates the analytic leverage regime typologies offer in
explainingcross-nationalvariationsin transitionsto democracy.10Brattonand van de
Walle highlight three importantways that the type of nondemocraticregime shapes
democratizationprocesses. First, broad differences in old regime institutionsdistin-
guish the modal path of regime change in Africa from paths in other regions.
Second, regime institutionshelp define the roles challengers play in initiating and
pushing forward transitions. Finally, old regime institutions critically influence
incumbents'strategiesof crisis management.

Neopatrimonialism and Africa's Modal Path of Regime Change Bratton and

van de Walle argue that differences in old regime type lead to distinct stages and
processes in democratization.Specifically, core institutions of neopatrimonialism,
the hallmarkancien regime of the more than forty sub-SaharanAfrican countries
analyzed in their book, led to a different modal path of democratization than in
regions dominated by highly bureaucratizedregimes, such as Latin America and
southernEurope.
Neopatrimonialregimes are characterizedby concentrationof power in the hands
of an individualruler who maintains control mainly by distributingpatronageto a
networkof clients."lAlthough formal institutionsand the rule of impersonallaw are
notoriously weak in neopatrimonialregimes, Bratton and van de Walle argue that
"informal, partially hidden, and extralegal institutions"structurepolitical life and
thus create predictablepatternsof transition(p. 274). These institutionalattributesof
neopatrimonialismexplain the distinctivemodal path of regime change that, accord-
ing to Brattonand van de Walle, differentiatesAfrica from most developing coun-
tries in otherregions.
The paradigmatic voluntarist account portrays democratization as a process
launchedby splits among regime incumbents.'2Fromthis perspectiveregime transi-
tion is understood as a strategic game between regime softliners and hardliners.
Political liberalizationoccurs after softlinersgain the upperhand,and elite pacts and
compromises pave the way to democratization.Bratton and van de Walle make a
valuablecontributionby showing that, althoughthis paradigmaticaccountmay accu-
rately describe transitions from bureaucratized, authoritarian regimes in Latin
America and southernEurope,it does not fit the neopatrimonialcases of Africa.'3
First, transitions from neopatrimonialism were typically initiated by popular
protest, not by splits among regime incumbentsor liberalizingreformsby softliners.

108
 RichardSnyderand James Mahoney

Neopatrimonialdictatorsrarely undertookpolitical liberalizationin the absence of

popularprotest.In contrastto leadersof more bureaucratizedregimes, neopatrimoni-
al rulers typically lacked institutionalizedties to social groups and thus could not
accuratelyassess popularsentimentsor judge the need for preemptivepolitical liber-
alization (p. 84). In the absence of institutionalizedchannels of interest representa-
tion, popularprotest was often the only way citizens could express grievances and
voice demands.Moreover,the exclusionarypatronagesystem and economic stagna-
tion associated with neopatrimonialregimes tended to foster such grievances. In
short, "personalrulers were unlikely to initiate political liberalizationor relinquish
powerwithout a struggle;they had to be forced out" (p. 84).
Second, in contrast to incumbents in bureaucratized,authoritarianregimes in
LatinAmerica and southernEurope,who usually split over the question of political
liberalization,neopatrimonialelites divided over access to patronage (pp. 85-86).
Neopatrimonialrulers rotatedofficeholders frequently,and elites forced from office
lost their access to spoils. In an effort to regain control of the spoils of office, disen-
franchisedelites often joined or organizedopposition movements. Hence, instead of
a division between hardlinersand softliners over the issue of political liberalization,
the African modal path was characterizedby elite divisions rooted in a struggle for
state-controlledpatronage.
Finally, the low levels of political institutionalizationin neopatrimonialregimes
reduced the possibility that enduring elite compromises and pacts could be forged
during regime transitions. In contrast to their counterpartsin Latin America and
Europe, nondemocratic elites in Africa's neopatrimonial regimes often did not
belong to cohesive, well-organized groups. Because these African elites typically
"representedno more than a tiny coterie of clients," they had difficulty in forging
consensus aroundintraeliteagreements(p. 87).

Regime Type and Challenger Strategies In contrast to Roeder, who limits his
study to institutional dynamics inside the authoritarianstate, Bratton and van de
Walle analyze how institutionsmediate the state'srelationshipwith society. By con-
structing a typology of neopatrimonialregimes that highlights varied patterns of
state-society relations, they show how institutionalfactors condition the likelihood
and level of popularprotest. Because Brattonand van de Walle combine their focus
on challengers with a focus on incumbents,they can explain more components of
the process of regime change than Roeder, who, as noted earlier, focuses only on
incumbentsand their capacitiesto achieve reform.
Building on the work of Robert Dahl, Brattonand van de Walle distinguish four
types of neopatrimonialregimes according to varied levels of "participation"(the
extent to which the state permits social mobilization) and "competition"(the extent
to which the state tolerates autonomouspolitical associations): military oligarchies
(low participation,low competition),multipartysystems (medium to high participa-

109
ComparativePolitics October 1999

tion, medium to high competition),plebiscitaryone party systems (medium partici-

pation, low competition),and competitive one party systems (medium participation,
medium competition).'4Although scholars often use these same four categories to
describe political regimes quite distinct from neopatrimonialism,Brattonand van de
Walle understandthem as "variantson a neopatrimonialtheme" that captureimpor-
tant differencesamongAfrican cases that all "featured'big man' rule"(p. 77).
The authorslink these four types of neopatrimonialregimes to the strengthsand
strategiesof oppositiongroups.They arguethat"politicalprotestwas least likely in mil-
itaryregimes;it was somewhatmore likely in multipartysystems;but it was most likely
in [plebiscitaryand competitive]one-partysystems"(p. 144). To explainthese patterns
they explorehow the distinctiveinstitutionalfeaturesof each type of regime shapethe
costs andbenefits for social groupsin organizingprotests.
Althoughmilitaryregimes gave citizens powerfulincentivesfor protestby denying
them channels to influence public policy, their capacity to suppress opposition also
raised the costs of such activity. The extremely low levels of political competition
allowedby militaryregimesdeniedoppositiongroupsopportunitiesto accumulateorga-
nizationalexperience,furtherincreasingthe costs of collective action. Hence "political
protestrarelyoccurredin Africa'smilitaryregimes because the state elite's controlof
the apparatusof coercionand the absenceof availablechannelsfor politicalexpression
made it too risky" (p. 144).
The costs of protestwere lowerin multipartyregimes.These regimestolerateda sig-
nificantdegreeof politicalopposition,and competingpoliticalpartiesprovidedinstitu-
tional channelsto voice discontent.Hence visible oppositionoccurredmore frequently
in multipartythanin militaryregimes.
One partyregimeswere the most likely to experienceprotest.On the one hand,like
multipartysystems they had representativeinstitutionsand low levels of repression.On
the otherhand,they also providedmanyof the same incentivesfor protestfoundin mili-
tary regimes. The monopoly control over interestmediation held by officially sanc-
tioned organizationsstifled autonomouscitizen participationand blockedexpressionof
grievances.The hegemonyof state-controlledinterestassociationsgave mobilized citi-
zens little choice but to experimentwith extralegalmodes of participationoutsideoffi-
cial partyor legislativechannels.
In sum, Brattonand van de Walle explain challengers'behaviorin two steps. They
first establishcorrelationsbetweendistincttypes of neopatrimonialregimes and likeli-
hood of protest.They then explainthese correlationsby showing how the institutional
characteristicsof each regimetype shapedthe instrumentalcalculationsof social actors
in ways thateitherencouragedor inhibitedprotest.

Regime Type and Incumbent Strategies Whethertriggeredby protest"frombelow"

or elite divisions "from above," changes of regime almost always involve political
crises. A central issue in the analysis of regime change thus concerns the strategies

110
 RichardSnyderand James Mahoney

incumbentschoose to defend their interestsin contextsof crisis. Like Roeder,Bratton

and van de Walle show how the constraintsthat old regime institutions impose on
incumbentshelp explaintheirstrategies
Because regimetransitionsin Africatypicallybegan with oppositionprotest,incum-
bents (neopatrimonialdictatorsand their cronies) generally faced a mobilized group
that sought to overthrowthem.15In response to these challenges incumbentsacross
Africa pursued three crisis management strategies: managed transition, transition
througha nationalconference,and transitionthroughrapidelections (pp. 169-77). In
managedtransitionsincumbentsinitiateda tightlycontrolledreformprocess linkedto a
well-definedprojectto oversee and directthe transition.In nationalconferencetransi-
tions incumbentsbroughtnationalelites togetherto write new constitutionalrules. In
rapidelectiontransitions,by contrast,incumbentsfirstheld electionsand laterconvened
a nationalassemblyto drafta new constitution.
Leaders of military regimes usually pursued managed transitions. In military
regimes the armed forces deeply penetratedpolity and society, and military officers
generallyled the government.Consequently,militaryincumbentswere well-positioned
to controlthe transition.The armedforces'virtualmonopolyoverthe meansof coercion
reinforced the capability of military incumbents to dominate the transition.
Furthermore,a managedtransition"flatteredthe military'sidealizedview of itself as a
rational,orderly,and organizedforce tryingto impose orderon an inherentlydisorderly
civilian political arena"(p. 171). Finally, a managed transitionafforded the armed
forces attractiveopportunitiesto protectthe military'sinstitutionalintegrityand prevent
reprisalsundera successorregime.
Incumbentsin one partyplebiscitaryregimes preferrednationalconferences.They
were accustomedto plebiscitaryprotocolsand thus were predisposedto hold national
conferencesto vindicatethe governmentand restorepolitical stability.They believed
they could manipulatesuch conferencesto their advantage,as they had often done in
the past.Thus, "thenationalconferencewas a logical extensionof the institutionalcon-
figuration of the plebiscitary regime" (p. 175).
In competitiveone partyregimes,by contrast,incumbentsfavoredelections without
holding nationalassemblies beforehand.They calculatedthat they enjoyed sufficient
legitimacyto win a competitiveelection, especially if the oppositionhad little time to
prepare.In addition,theircontrolover public funds,electoralmachinery,and the press
furtherbolsteredtheirconfidence aboutthe prospectsfor victory.Elites in competitive
one party regimes thus viewed rapid elections as mechanismsthat would help them
keep power.

Limitations of Typological Analysis Because Brattonand van de Walle link old

regime type to the strategiesof both incumbentsand challengers, they offer a more
robust explanatory framework than Roeder, who focuses only on incumbents.
However, their approach to regime analysis, which relies heavily on broad cross-

Ill
ComparativePolitics October 1999

nationalcorrelationsbetween old regime types and actor strategies,tends to obscure

the micro-level rules and incentivesthat define contexts of incumbentand challenger
decision making. In contrastto Roeder, who carefully specifies the Soviet institu-
tions that constrainedreformers,Brattonand van de Walle often fail to pinpointspe-
cific rules and incentives that explain correlationsbetween types of African regimes
and actor strategies.
Their limited success in identifying the causal mechanisms that explain cross-
national patternsestablishedthroughcorrelationalanalysis stems in part from their
reliance on highly aggregated regime categories. Bratton and van de Walle define
types of African regimes in terms of the broad,Dahlian dimensions of political com-
petition and participation(p. 68). Yet these two dimensions do not captureimportant
institutional factors the authors themselves argue shaped the strategic options of
incumbentsand challengers.
For example, the distinctive dynamics of transitions from military regimes,
against which opposition groups rarely launched political protests, can not be ade-
quatelyexplainedin terms of these regimes' characteristicallylow degrees of partici-
pation and competition.Civilian regimes with these same attributesmight be expect-
ed to experience significant levels of protest, especially since the authorsargue that
political exclusion can create strong incentives for extralegalmodes of participation
(p. 145). Indeed,as Brattonand van de Walle acknowledge,the military'smonopoly
over coercive resources, not low degrees of participationand competition (pp. 144,
171), inhibited protest. Focusing on aggregate levels of participationand competi-
tion thus obscures importantvariationsin who regulates and enforces these levels:
military governmentswith guns or civilian governmentsthat lack direct control over
coercive resources.
Furthermore,as the authorssuggest, variationsin the structureof patronage,such
as the degree to which the ruler'sclientelist networkpenetratesstate and social orga-
nizations and the differentkinds of benefits distributedthroughsuch networks,also
have an importantimpact on the strategiesof opposition and incumbentgroups (pp.
85-86). The dimensions of participationand competition that anchor Bratton and
van de Walle's regime typology, however, do not discriminate among cases of
neopatrimonialismin terms of their variedpatronageinstitutions.16A more nuanced
typological frameworkthat exposed such core micro-institutionsof neopatrimonial
regimes would have strengthenedthe analysis.

The Constitutive Effects of Institutions

The ability of institutionalvariablesto help explain regime change in contexts as dif-

ferent as the Soviet Union and sub-SaharanAfrica suggests that political institutions
play importantroles in processes of regime transformationacross a very broadrange

112
 RichardSnyderand James Mahoney

of cases. Juan J. Linz and Alfred Stepan'sProblems of Democratic Transitionand

Consolidationconfirms this expectation.This landmarkstudy shows that old regime
institutionshelp account for regime change across a remarkablyheterogeneousset of
countrieswith variedtypes of nondemocraticsystems. Linz and Stepananalyzedemo-
cratic transitionand consolidationin fourteencountriesfrom three regions: southern
Europe, the southern cone of South America, and postcommunistEurope. The old
regimes in these cases span virtuallythe entire spectrumof modern nondemocratic
systems: authoritarianism,totalitarianism,posttotalitarianism,and sultanism.l7Linz
and Stepan'swork thus providesa strongbasis for inferringthat institutionscan help
explainregimechangeacrossthe full set of"thirdwave"democracies.
Like Brattonand van de Walle, Linz and Stepan correlateregime type with the
strategies of incumbents and challengers. However, they rely on different assump-
tions about how institutions affect actors' behavior.For Bratton and van de Walle,
institutionsshape the strategiccalculationsof actors assumedto have relativelyfixed
interests.Fromthis perspectiveregime type correlateswith actors' behaviorbecause
political institutionsdefine strategiccontexts that constrainthe self-interestedbehav-
ior of incumbents and challengers. Linz and Stepan acknowledge that old regime
institutions can influence behavior in this way. However, they also focus on how
institutionsdefine identities and interests,showing that regime institutionscan mold
the self-images, goals, and preferences of incumbents and challengers. Linz and
Stepan's study thus exemplifies the constitutive approach to regime change. This
approach explains political behavior during regime transitions by exploring how
institutionsconstitutethe interestsand identities of actors. By endogenizing interest
and identity formation,it moves beyond perspectivesthat treat institutionsas incen-
tive structuresthat merely define strategic opportunities for actors understood to
have stablepreferencesand orientations.18

The Institutional Origins of Incumbents and Challengers Voluntaristanalyses

conceptualize regime transitionsas games of strategic interactionbetween various
incumbentand opposition factions. These factions are usually defined by their pref-
erences about liberalizationand democratization.Thus, many scholarsregardregime
hardliners and softliners and opposition maximalists and moderates as the core
actors in transitionsto democracy.19Such distinctions have proven useful, and it is
now widely accepted that the relative strengths of these incumbentand opposition
factions help explain variedtransitionpaths, or "modes of transition."20
Studentsof regime change,however,have had less success in explainingthe origins
of these factions.Many studies simply assume that hardliners,softliners,maximalists,
and moderatesexist, withoutspecifyingthe conditionsunderwhich such groupsorga-
nize or fail to organize.As illustratedby Linz and Stepan'sanalysisof how old regime
institutionsshapedthe formationof incumbentand oppositiongroups,focusing on the
constitutivepropertiesof institutionscan help overcomethis limitation.

113
ComparativePolitics October 1999

Linz and Stepan show that the paradigmaticmodel of a four player transition
game, based on the democratizationof Spain'sauthoritarianregime in the late 1970s,
does not fit transitionsfrom othertypes of nondemocraticregimes. Old regime insti-
tutions can narrowthe cast of actors. For example, the characteristicallyhigh degree
of penetrationinto state and society by the patronagenetworksof sultanisticregimes
and the mobilizational parties of totalitarian regimes minimizes possibilities for
groups with reformist or moderate identities to organize. Softline incumbents and
moderatechallengersare typically absent in such regimes, resulting either in stable
one player equilibria (just hardliners) or volatile two player games (hardliners
opposed by maximalist, usually armed oppositions, as often occurs with sultanistic
regimes). Linz and Stepanhighlight the constitutiveeffects of regime institutionsby
describingthe violent overthrowofNicolae Ceausescu'shybridsultanisticltotalitarian
regime in Romaniaas a case of "missing players for a 'pacted transition."'(p. 356).
Their analysis shows how the absence of softline incumbents and moderate chal-
lengers foreclosedthe possibility of a nonviolentnegotiatedtransitionin Romania.
Old regime institutions also affect the capabilities of incumbent and opposition
factions. For example, Linz and Stepanshow how the distinctive featuresof Poland's
"authoritariancommunist"regime help account for the far greaterstrengthof maxi-
malist opposition groups in Poland than in East Europeancountries that were ruled
by full-blown totalitarianregimes. The party-statein Poland did not penetratecivil
society as fully as party-stateselsewhere in easternEurope (pp. 255-69). This limit-
ed penetration provided a comparatively favorable context for forging opposition
identities; social groups had the space to experimentwith noncommunistmodes of
self-identification. Once these identities were forged, the relatively circumscribed
reach of the Polish state facilitated an extraordinarydegree of "self-organization"
and anticommunistmobilizationby civil society (p. 262). Hence the distinctiveinsti-
tutions of the old regime contributedto the emergence of a powerful opposition,
which in turn helps explain why Poland experienced a pacted transition in which
incumbentsnegotiatedthe terms of their extricationwith the opposition.

Opposition Identities and Protest Repertoires Incumbentand opposition groups

often have broaderidentities. These group identities warrantour attention because
they can lead to importantcleavages among actors with similar preferences about
democratization.For example, ethnic differences can cause divisions within moder-
ate or maximalistcamps, despite sharedpreferencesabout democracythat may help
unite each group.
Linz and Stepan show how a focus on institutional factors provides valuable
leverage in understandingthe formationand reproductionof opposition identities.
Their study highlights how engaging and seeking to modify political institutionscan
alter actors' self-understandingsand goals. 21 Hence the "practice"of opposing a
regime may redefine a group'sidentityand conception of appropriatebehavior.

114
 RichardSnyderand James Mahoney

Linz and Stepan address the linkage between opposition practice and identity in
their analysis of the East European and post-Soviet cases, where the "flattened"
political landscapes left by totalitarianregimes provide an excellent opportunityto
explore issues of group formation (p. 269). Poland offers a vivid example. The
Polish opposition forged an identity of "antipolitics"in the process of challenging
the old regime.As noted above, Linz and Stepan show how authoritarianCommunist
institutionsprovideda favorablecontext for craftingopposition identities.They also
analyze the complex process through which these identities were constructed.
Repeated accumulated interactions between opposition groups and the incumbent
regime yielded a disdain by the former for routinized institutionsand compromise.
Between 1976 and 1989 the opposition developed a language, self-definition, and
set of tactics orientedtowardspontaneityand informality.Takentogether,these ele-
ments formed a coherent "ethics of oppositional behavior"that was in many ways
the mirrorimage of the bureaucratizedregime institutionsthe opposition sought to
transform.According to Linz and Stepan, the opposition was "so eager to avoid
becoming capturedin the routines and symbols of the party-statethat they elevated
the situationalethics of oppositionalbehaviorinto a generalprincipleof the 'politics
of anti-politics"'(p. 271).
Although its "antipolitical"identity helps explain the Polish opposition'sremark-
able ability to sustainitself for nearlytwo decades, this quasi-anarchistconception of
opposition politics as the antithesis of old regime practices created importantbarri-
ers to the subsequent consolidation of democracy.Not surprisingly,after years of
dominationby a single party,political parties became a dirty word for the opposi-
tion. Hence the old regime left a legacy of strongambivalenceabout core democratic
political institutions such as political parties and hindered the consolidation of
democracy.
Opposition identities forged during the transitionto democracy in Brazil had a
similarconstrainingeffect on democraticconsolidation.In Brazil a powerfulmilitary
organizationled the authoritarianregime and tightly controlled the pace of political
liberalization. Consequently, Brazil witnessed a "long, constrained transition" in
which the opposition struggled against the regime for more than a decade (p. 166).
Like their Polish counterparts,the Brazilian opposition developed an "antipolitical"
identity during this struggle. In both Brazil and Poland the long, slow struggle
against a bureaucratic,highly routinizedregime "engenderedvalues and patternsof
action in the arenaof civil society that impededthe constructionof a democratically
effective political society" (pp. 232-33). Since completing the democratictransition,
Brazilianshave been strikinglyambivalentabout the value of democracy,especially
when comparedto citizens of other SouthAmericandemocracies.22
These similaritiesbetween Poland and Brazil raise intriguingquestions for future
researchon how constitutivepropertiesof old regime institutionsinfluence prospects
for democratic consolidation. Do slow transitions from highly bureaucratized

115
ComparativePolitics October 1999

regimes systematically yield the kinds of problems for democratic consolidation

observed in Poland and Brazil? How do "antipolitical"opposition identities forged
in challenging nondemocraticregimes interactwith other mechanisms, such as elite
pacts, that scholarshave also arguedhinderconsolidation of democracy?23What are
the constitutiveeffects of regime institutionsin cases of rapidtransitionfrom highly
bureaucratizedregimes where opposition groups may have little time to develop
antipoliticalidentities?

Old Regime Legacies and Democratic Consolidation In contrastto most recent

work on democratictransitions,studies of democraticconsolidationhave relied con-
siderablyon structuralvariables,often analyzing factors that temporallyprecede the
old regime's breakdown.24Despite these importantefforts to link antecedent struc-
tural factors to subsequentconsolidation dynamics, however,old regime institutions
have not played a centralexplanatoryrole in the study of democraticconsolidation.
Linz and Stepan make a major contributionby showing how old regime institu-
tions help explain the tasks of democraticconsolidation faced by incumbentsin new
democracies.They specify these tasks by constructinga new typology that classifies
nondemocratic regimes according to their varied institutional components. This
typology identifies "five interrelatedarenas"of a consolidated democracy:a robust
civil society, a free political society, the rule of law, a state bureaucracybased on
rational-legalnorms, and an economy with marketautonomyand ownershipdiversi-
ty (pp. 7-15). Linz and Stepanconverteach of these arenasinto a typological dimen-
sion and then score (high, medium, low) differentkinds of nondemocraticregimes
across the five dimensions. Variedtypes of nondemocraticregimes are thus arrayed
in a typological space according to their "distance"from consolidated democracy.
Sultanismand totalitarianismare the most distant,and authoritarianismthe least dis-
tant (p. 56). Scores on each of the five dimensions indicatethe difficulty of the tasks
democratizersface in consolidatingdemocracy.
This typology helps the authorsdeduce the prospectsof democratizationfrom old
regime institutions.For example, in the case of sultanisticregimes, which score low
across all five dimensions, their "typology direct[s] attention to the fact that the
immediate implications of a sultanistic regime for democracy-crafters(as in Haiti)
are that they will have to begin the constructionof civil society, constitutionalism
and a rule of law, professional norms of the bureaucracy,economic society, and
political institutionsfrom a very low base" (p. 56). When comparedto authoritarian
regimes, which score higher on most dimensions, sultanistic regimes pose much
more formidabletasks of democratization.Indeed,as with totalitarianregimes, "the
failureof democraticconsolidationis almost overdetermined"(p. 233).
Linz and Stepan thus show how old regime institutions affect the dynamics of
consolidation by defining the specific tasks facing incumbents of new democratic
regimes. Futureresearchmight take the furtherstep of exploring the impact of old

116
 RichardSnyderand James Mahoney

regime institutionson nondemocraticopposition groups. Their effects on democratic

incumbents and nondemocratic challengers could yield powerful insights about
democraticconsolidation.

Limitations of Constitutive Analysis Linz and Stepan's efforts to make actors'

interestsand identities endogenous to institutionalanalysis have several limitations.
Because they treat institutionsas both consequences and causes of group identities,
their argumentlacks the parsimonyof analyses that exogenize identity formationand
do not explore constitutiveeffects of institutions.Furthermore,in their case studies
they do not always specify clearly the causal links between old regime institutions,
actors' identities and behavior,and regime transformation.The reader is thus often
given the difficult task of disentanglingthe complex, interactivecausal relationships
among these variables.
Endogenizinginterest and identity formationalso tends to restrictthe generality
of Linz and Stepan'sarguments.Indeed, several of their findings apply to just one
country. For example, they highlight quite different institutionalfactors to explain
the formationof group identities especially favorableto democracyin Chile, Spain,
and Uruguay.In Chile a nationwidereferendumfostered a prodemocraticcivic iden-
tity. In Spain rules governing the sequencing of national and subnationalelections
duringthe transitioncontributedto group identities that favored democracy.Finally,
in Uruguay institutionsallowing controlledparticipationby traditionalpolitical par-
ties weakened the military's antidemocraticattitudes. While intriguing, these idio-
syncratic findings about the impact of electoral rules and other regime institutions
on group preferences towarddemocracy do not add up to a generalizableexplana-
tion. Rather,they suggest that quite differentinstitutionalfactors account for prode-
mocraticidentitiesacross the three cases.
Linking political institutionsto the constitutionof actors' identities and interests
represents an important, promising innovation in the study of regime change.
However,steps should be takento manage the problemsof lost parsimonyand weak
generalizability.

Directions for Future Research

Althoughmany subfields of comparativepolitics have recentlyturnedto institutions,

most work on regime change pays little attentionto institutionalvariables.The domi-
nance of structural and voluntarist approaches has focused attention instead on
socioeconomic structures and contingent elite choices. Consequently, students of
regime change started to incorporateinstitutionalvariables significantly later than
their colleagues in other areas of comparativepolitics.
This delay is unfortunatebecause institutionalanalysis offers significant leverage

117
ComparativePolitics October1999

in explainingregime change. Roeder shows how the formal and informalrules under
which incumbents strive to maintain power can explain the failure of reformersto
solve problems of regime maintenance.Brattonand van de Walle demonstratehow
broad cross-nationalvariationsin regime institutionsdefine the strategic options of
both incumbentswho defend the old regime and challengerswho undermineit. Linz
and Stepan, in additionto underscoringthe impact of institutionalfactors on actors'
strategies,highlight how the constitutivepropertiesof institutionsshape group iden-
tities, interests,and goals.
In their sharedrecognitionthat regime change can be a rule-governedprocess, all
three analyses unequivocallyretreatfrom the extreme voluntarismthat characterized
most previous studies inspiredby the thirdwave of democratization.However,none
of these works seeks to revive the old structuralism,which privileged macro-level
socioeconomic or cultural variables to the neglect of human agency. By focusing
instead on the institutionalcontexts of choice and decision making, they achieve a
high degree of sensitivity to agency and political action. Through this linkage of
institutional constraints to the shaping of actor's choice, these studies stake out a
long-awaited middle ground between the voluntarist and structuralextremes that
dominatedearlierwork on regime change.25
Future research should take several directions to fortify this emerging middle
ground. First, importantconceptualwork needs to be done. In the past regime cate-
gories served mainly to describe and classify different political systems, entering
into explanationprimarilyas dependentvariables.As regime categories are increas-
ingly employed as explanatoryvariables,however,scholars need to modify existing
conceptual frameworks.For example, Linz and Stepan (pp. 39-40) show that the
standardtripartitedistinction among democracy,authoritarianism,and totalitarian-
ism is inadequate in explaining regime transitions because it fails to discriminate
among the more thanninety percentof modernnondemocraticregimes that sharethe
same typological space, authoritarianism.To achieve a conceptual frameworkwith
greaterdiscriminatingpower Linz and Stepanexpand the existing tripartitetypology
by adding two new categories, posttotalitarianismand sultanism.26Likewise, the
need for more finely grained typologies is evident in the limitations of Brattonand
van de Walle's highly aggregated categories of neopatrimonialregimes. An impor-
tant priority in futureresearchthus involves crafting nuanced concepts that expose
the institutionallogics of nondemocraticregimes by pinpointingthe core rules that
constrainincumbentsand social actors.
Second, scholars who wish to explore how institutionsconstitute actors' interests
and identities should strive for greatermethodologicalrigor and sophistication.They
should avoid the temptation to conflate independent and dependent variables and
should carefullyspecify when institutionalfactors are causes of actors' identities and
interests and when actors are causers of institutional outcomes.27 In addition to
adhering closely to standardmethodological guidelines for causal assessment, stu-

118
 RichardSnyderand James Mahoney

dents of regime change who study constitutive properties of political institutions

need to specify better the causal mechanisms that connect institutions to actors'
identities and preferences. A stronger understandingof these mechanisms should
help solve the difficult problems of reciprocal causation created by endogenizing
interestand identity formation.
Finally,studentsof regime change should guardagainst overestimatingthe power
of institutional variables. While important,institutions do not explain everything.
Roeder's book illustrates the limits of institutionalanalysis. Although its focus on
intraregime institutions of elite recruitment and leadership accountability helps
explain a crucial link in the causal chain connecting the Soviet and post-Soviet
regimes-the failure of reformers-, this link did not by itself add up to regime
change. As Roeder admits, socioeconomic transformationsthat createdpressuresfor
reformwere also a necessary condition for the Soviet regime'sdemise.
Recent comparativehistorical work on long-term patterns of regime evolution
also alerts us to the risks of ignoring the noninstitutionaldimensions of institutional
change. It shows how major instances of institutional transformationoften result
from extrainstitutionalfactors, such as changes in the global economy and demo-
graphic shifts, that reconfigure actors' interests and power.28Acknowledgmentthat
institutionshave been a missing variablein the study of regime change thus should
not imply that they are the only variablethat warrantsanalysis. Indeed,althoughwe
have focused on the analytic primacy of institutional factors, Bratton and van de
Walle and Linz and Stepan also devote considerable attention to noninstitutional
variables,such as socioeconomic and culturalstructures.
We thereforecaution against substitutinginstitutionaldeterminismfor the struc-
tural determinism and voluntarist indeterminism of earlier work. The study of
regime change has been handicappedby its tendency to advance "one variableat a
time," from fixation with macro structureto fixation with contingent elite choice.29
Rather than shift fully to institutional analysis, students of regime change should
instead strive to develop integrativeresearchstrategiesthat combine human agency,
political institutions,and macro structures.30Such multivariateintegrativethinking
will provide a firm basis for bringing institutionalfactors into the study of regime
change without denying the potentialimportanceof noninstitutionalvariables.

NOTES

The authorsthank GerardoMunck and Ezra N. Suleiman for helpful comments on an earlier draft.
Richard Snyder's work on this article was supported by a postdoctoral fellowship from the Harvard
Academy for InternationalandArea Studies.
1. PeterA. Hall and RosemaryC. R. Taylor,"PoliticalScience and the Three New Institutionalisms,"
Political Studies, 44 (1996), 936-57; Karen L. Remmer, "Theoretical Decay and Theoretical
Development: The Resurgence of Institutional Analysis," WorldPolitics, 50 (October 1997), 34-61;

119
ComparativePolitics October 1999

Rogers Brubaker,NationalismReframed.:Nationhoodand the National Questionin the New Europe(New

York: Cambridge University Press, 1996); Peter Evans, Embedded Autonomy. States and Industrial
Transformation(Princeton:PrincetonUniversity Press, 1995); Robert H. Bates, Open-EconomyPolitics:
The Political Economy of the WorldCoffee Trade(Princeton: Princeton University Press, 1997); Paul
Pierson, Dismantling the WelfareState? Reagan, Thatcher,and the Politics of Retrenchment(New York:
CambridgeUniversityPress, 1994).
2. Matthew S. Shugartand John M. Carey, Presidents and Assemblies.:ConstitutionalDesign and
Electoral Dynamics (New York:CambridgeUniversity Press, 1992); Juan J. Linz and ArturoValenzuela,
eds., The Failure of Presidential Democracy (Baltimore: The Johns Hopkins University Press, 1994);
Arend Lijphartand Carlos H. Waisman,eds., InstitutionalDesign in New Democracies: Eastern Europe
and LatinAmerica (Boulder: Westview, 1996); Scott Mainwaringand Timothy R. Scully, eds., Building
Democratic Institutions.:Party Systems in Latin America (Stanford: Stanford University Press, 1995);
Scott Mainwaringand MattthewSoberg Shugart,eds., Presidentialismand Democracy in LatinAmerica
(New York:CambridgeUniversityPress, 1997); Samuel P. Huntington,The ThirdWave.:Democratization
in the Late TwentiethCentury(Norman:Universityof OklahomaPress, 1991).
3. Two notable exceptions are Karen L. Remmer,Military Rule in Latin America (Boston: Unwin
Hyman, 1989), esp. ch. 7; and Huntington.
4. Guillermo O'Donnell, Philippe Schmitter, and Lawrence Whitehead, eds.. Transitionsfrom
AuthoritarianRule: Prospectsfor Democracy (Baltimore: The Johns Hopkins University Press, 1986),
esp. TentativeConclusions about Uncertain Democracies; James M. Malloy and Mitchell A. Seligson,
eds., Authoritarians and Democrats.:Regime Transition in Latin America (Pittsburgh: University of
Pittsburgh Press, 1987); Enrique Baloyra, ed., Comparing New Democracies: Transition and
Consolidationin MediterraneanEuropeand the SouthernCone (Boulder:Westview, 1987); Giuseppe Di
Palma, To Craft Democracies: An Essay on Democratic Transitions(Berkeley: University of California
Press, 1990). Althoughthey focused on the collapse of democraticregimes and emphasizedmainly politi-
cal leadership,the contributorsto Juan J. Linz and Alfred Stepan, eds., The Breakdownof Democratic
Regimes (Baltimore:The Johns Hopkins University Press, 1978), esp. ArturoValenzuela,also considered
structuraland institutionalvariables.
5. See, for example, Guillermo O'Donnell, Modernization and Bureaucratic-Authoritarianism:
Studies in SouthAmericanPolitics (Berkeley: Instituteof InternationalStudies, University of California,
Berkeley, 1973); Philippe C. Schmitter,"Still the Centuryof Corporatism?,"The Review of Politics, 36
(1974), 85-131; BarringtonMoore, Jr.,Social Origins of Dictatorshipand Democracy:Lord and Peasant
in the Making of the Modern World(Boston: Beacon Press, 1966); Gregory M. Luebbert,Liberalism,
Fascism, or Social Democracy? Social Classes and the Political Origins of Regimes in InterwarEurope
(New York:Oxford UniversityPress, 1991).
6. Philip G. Roeder,Red Sunset: TheFailure of Soviet Politics (Princeton:PrincetonUniversityPress,
1993); Michael Brattonand Nicolas van de Walle, DemocraticExperimentsin Africa: Regime Transitions
in Comparative Perspective (New York: Cambridge University Press, 1997); Juan J. Linz and Alfred
Stepan, Problems of Democratic Transitionand Consolidation: Southern Europe, South America, and
Post-CommunistEurope(Baltimore:The Johns Hopkins UniversityPress, 1996).
7. Forexample, O'Donnell and Schmitter,TentativeConclusion,focus on the split between hardliners
who oppose reformand softlinerswho supportit.
8. Roeder,p. 11, writes of "informaland unwrittenrules hidden by formal, but less significant win-
dow dressing."
9. For example, Carl J. Friedrich and Zbigniew K. Brzezinski, Totalitarian Dictatorship and
Autocracy (Cambridge, Mass.: Harvard University Press, 1956); Juan J. Linz, "Totalitarian and
AuthoritarianRegimes," in Fred Greenstein and Nelson Polsby, eds., Handbook of Political Science, 3
(Reading: Addison-Wesley, 1975); David Collier and Steven Levitsky, "Democracy with Adjectives:
ConceptualInnovationin ComparativeResearch,"WorldPolitics, 49 (April 1997), 430-51.

120
 RichardSnyderand James Mahoney

10. Both Remmer,MilitaryRule, and Huntingtonhave arguedthat the outcomes of political transitions
tend to covary with old regime type, beyond the highly aggregated categories of authoritarianismand
democracy.
11. See Richard Snyder, "ExplainingTransitions from NeopatrimonialDictatorships,"Comparative
Politics, 24 (July 1992), 379-99.
12. O'Donnell and Schmitter,TentativeConclusions.
13. Ruth Berins Collier and James Mahoney, "Adding Collective Actors to Collective Outcomes:
Labor and Recent Democratizationin South America and Southern Europe,"ComparativePolitics, 29
(April 1997), 285-303, have questioned whether the paradigmatic voluntarist account accurately
describesdemocratictransitionsin LatinAmericaand southernEurope.
14. Robert A. Dahl, Polyarchv.:Participation and Opposition (New Haven: Yale University Press,
1971).
15. As noted above, the likelihood of protest varied across regime type. However, the majority of
countries(twenty-nineout of forty-seven)analyzed by Brattonand van de Walle were one party systems,
the regime type that correlates most strongly with opposition protest. By contrast,only eleven countries
were military oligarchies,the regime type that correlatesmost weakly with protest. Hence protestcharac-
terized most of the cases.
16. Snyder,"ExplainingTransitionsfrom NeopatrimonialDictatorships."
17. These regime types are defined along the dimensions of pluralism, ideology, mobilization, and
leadership.Linz, "Totalitarianand AuthoritarianRegimes."
18. Such perspectivescharacterizemost rationalchoice analyses. On the differences between rational
choice and constitutive approaches to institutional analysis, see Kathleen Thelen and Sven Steinmo,
"Historical Institutionalism in Comparative Politics," in Sven Steinmo, Kathleen Thelen, and Frank
Longstreth,eds., StructuringPolitics: Historical Institutionalismin ComparativeAnalysis (Cambridge:
CambridgeUniversityPress, 1992); Hall andTaylor.
19. For example, Guillermo O'Donnell, "Transitionsto Democracy:Some NavigationalInstruments,"
in RobertA. Pastor,ed., Democracyin the Americas(New York:Holmes & Meier, 1989).
20. For example, Terry Lynn Karl, "Dilemmas of Democratizationin Latin America," Comparative
Politics, 23 (October 1990), 1-21; Terry Lynn Karl and Philippe C. Schmitter,"Modes of Transitionin
Latin America, Southern and Eastern Europe,"InternationalSocial Science Journal, 128 (May 1991),
269-84; GerardoL. Munck and Carol Skalnik Leff, "Modes of Transitionand Democratization:South
America and Eastern Europe in Comparative Perspective," Comparative Politics, 29 (April 1997),
343-62.
21. As Linz and Stepan, p. 366, argue, "political identities are less primordialand fixed than contin-
gent and changing.They are amenableto being constructedor erodedby political institutions."
22. This ambivalenceabout democracy in Brazil was probablyexacerbatedby two enduring institu-
tional factors that preceded the authoritarianregime: weak political parties and a strong federal design
that tended to fragmentthe nationalpolitical arena. Indeed,accordingto Linz and Stepan, p. 166, Brazil
has had the most difficulty in consolidatinga democraticregime among the SouthAmericanand southern
Europeancases they analyze.
23. See Karl,"Dilemmasof Democratization."
24. Brazil has had an especially importantinfluence in redirectingattentionto structuralfactors. See
Frances Hagopian, TraditionalPolitics and Regime Change in Brazil (New York:CambridgeUniversity
Press, 1996); Guillermo O'Donnell, "Challengesto Democratizationin Brazil,"WorldPolicy Journal, 5
(Spring 1988), 281-300; Scott Mainwaring,Guillermo O'Donnell, and J. Samuel Valenzuela,eds., Issues
in DemocraticConsolidation:The New SouthAmericanDemocracies in ComparativePerspective(South
Bend: Universityof Notre Dame Press, 1992). Adam Przeworskiand FernandoLimongi, "Modernization:
Theories and Facts,"WorldPolitics, 49 (January1997), 155-83, arguethat, althoughtransitionsto democ-
racy should be viewed as contingent events, economic constraintshave a major impacton the stability of

121
ComparativePolitics October 1999

democraticregimes once they are established.

25. Since the late 1980s numerousscholars have called for researchstrategiesthat integratestructural
factors and human agency in explaining regime change. See Daniel H. Levine, "Paradigm Lost:
Dependence to Democracy," WorldPolitics, 40 (1988), 177-94; Nancy Bermeo, "Rethinking Regime
Change," Comparative Politics, 22 (1990), 273-92; Karl, "Dilemmas of Democratization";Karen L.
Remmer, "New Wine or Old Bottlenecks? The Study of Latin American Democracy," Comparative
Politics, 23 (1991), 479-93; HerbertKitschelt, "PoliticalRegime Change: Structureand Process-Driven
Explanations?,"American Political Science Review, 86 (December 1992), 1028-34; GerardoL. Munck,
"DemocraticTransitionsin ComparativePerspective,"ComparativePolitics, 26 (1994), 355-75.
26. See H. E. Chehabi and Juan J. Linz, eds., Sultanistic Regimes (Baltimore: The Johns Hopkins
UniversityPress, 1998).
27. See Huntington,p. 107.
28. See, for example, Ruth Berins Collier and David Collier, Shaping the Political Arena: Critical
Junctures,the LaborMovement,and RegimeDynamics in LatinAmerica (Princeton:PrincetonUniversity
Press, 1991); Dietrich Rueschemeyer, John D. Stephens, and Evelyne Huber Stephens, Capitalist
Developmentand Democracy(Chicago: Universityof Chicago Press, 1992).
29. See Bermeo; Remmer,"New Wine or Old Bottlenecks."
30. See Richard Snyder, "Paths Out of Sultanistic Regimes: Combining Structuraland Voluntarist
Perspectives,"in H. E. Chehabiand Linz, eds.; James Mahoneyand RichardSnyder,"RethinkingAgency
and Structurein the Study of Regime Change,"Studies in ComparativeInternationalDevelopment(forth-
coming).

122