Good Fences Make Good Neighbours? A Comparison of Conflict-Regulation Strategies in Postwar Bosnia

http://jpr.sagepub.com/Journal of Peace Research

http://jpr.sagepub.com/content/41/5/569The online version of this article can be found at:

DOI: 10.1177/0022343304045973

2004 41: 569Journal of Peace ResearchNina Caspersen

Strategies in Postwar BosniaGood Fences Make Good Neighbours? A Comparison of Conflict-Regulation

Published by:

http://www.sagepublications.com

On behalf of:

Peace Research Institute Oslo

Journal of Peace Research Replication Data

can be found at:Journal of Peace ResearchAdditional services and information for

http://jpr.sagepub.com/cgi/alertsEmail Alerts:

http://jpr.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://jpr.sagepub.com/content/41/5/569.refs.htmlCitations:

What is This?

- Aug 27, 2004Version of Record >>

at Scientific library of Moscow State University on February 14, 2014jpr.sagepub.comDownloaded from at Scientific library of Moscow State University on February 14, 2014jpr.sagepub.comDownloaded from

http://jpr.sagepub.com/


http://jpr.sagepub.com/content/41/5/569

http://jpr.sagepub.com/content/41/5/569



http://www.prio.no/Data/

http://www.prio.no/Data/

http://jpr.sagepub.com/cgi/alerts

http://jpr.sagepub.com/cgi/alerts

http://jpr.sagepub.com/subscriptions

http://jpr.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://www.sagepub.com/journalsPermissions.nav

http://jpr.sagepub.com/content/41/5/569.refs.html

http://jpr.sagepub.com/content/41/5/569.refs.html

http://jpr.sagepub.com/content/41/5/569.full.pdf

http://jpr.sagepub.com/content/41/5/569.full.pdf

http://online.sagepub.com/site/sphelp/vorhelp.xhtml

http://online.sagepub.com/site/sphelp/vorhelp.xhtml





569

Introduction

Which institutions are most effective in fos-tering stability following an ethnic war? Thisquestion increasingly has to be answered byinternational policymakers, since inter-national interventions in ethnic wars have

become more commonplace.1 Internationalpresence will often limit the options on theagenda to inclusive solutions, that is, demo-cratic rule within the existing state that doesnot serve one group at the expense of another(Sisk, 2001: 26). However, within thiscategory the options differ significantly, andthe right choice of approach can make thedifference between continued warfare andgradual development of peace and stability.

In the literature on democratic regu-lation of ethnic conflicts, Arend Lijphart’sconsociational model, as presented in, for

© 2004 Journal of Peace Research,vol. 41, no. 5, 2004, pp. 569–588Sage Publications (London, Thousand Oaks, CA and New Delhi) www.sagepublications.comDOI 10.1177/0022343304045973 ISSN 0022-3433

Good Fences Make Good Neighbours? AComparison of Conflict-Regulation Strategies inPostwar Bosnia*

NINA CASPERSEN

Department of Government, London School of Economics and PoliticalScience

This article compares the effectiveness of the consociational and the integrative approach in fosteringstability in postwar Bosnia. Whereas the ethnic groups in Lijphart’s consociational model constitute thebasic units on which the political structure is built, the political structure in Horowitz’s integrativemodel transcends ethnic divisions. The Dayton Agreement that ended the war in Bosnia containselements of both approaches, and the balance between them has been changing in the course of itsimplementation. Bosnia constitutes a very suitable case for a comparison of the effect on stability ofthe two approaches: elements of each approach can be isolated and their effects compared, and theinterplay between the approaches and the effect of international involvement can also be analysed. Aninternational dimension and the phase of (de)escalation of the conflict are variables missing from bothapproaches, but it is important to include them when analysing a postwar situation. Owing to the deepdivisions in the population, the numerical balance between the groups and the maximalist objectivesof the dominant parties, the consociational model has been more effective in fostering stability inBosnia. Currently, a change to an integrative structure seems premature, but a mix of the approacheshas been demonstrated to be able to foster moderation.

* An earlier version of this article was presented at aEuropean Peace Research Association (EuPRA) Con-ference, Schlaining, Austria, 13–17 July 2002, and pub-lished as a Working Paper by the Copenhagen PeaceResearch Institute and by Columbia International AffairsOnline. Thanks to Håkan Wiberg, Peter Viggo Jakobsen,Nils Petter Gleditsch and the anonymous referees for theirhelpful comments. The data used in this article can befound at http://personal.lse.ac.uk/caspersen. E-mail:[email protected].

1 ‘Ethnic’ does not in my usage signify anything inherentor permanent. What is decisive are the labels used and theway in which the conflict is legitimized. In the Bosniancontext, religion is regarded as an ethnic marker.

03 caspersen (ds) 3/8/04 9:35 am Page 569

at Scientific library of Moscow State University on February 14, 2014jpr.sagepub.comDownloaded from

www.sagepublications.com



example, Democracy in Plural Societies(1977), and Donald Horowitz’s integrativemodel, as presented in, for example, EthnicGroups in Conflict (1985), have become thefocal point of both empirical and theoreti-cal debate. The models are both based onthe assumption that a lasting settlementmust be built on inclusive rule. But,whereas the ethnic groups in Lijphart’sapproach constitute the basic units onwhich the political structure is built, thepolitical structure in Horowitz’s approachtranscends ethnic divisions. The approachesthus differ significantly, and the choicebetween them will have consequences forthe ability to foster stability.

The debate is highly relevant in the caseof Bosnia and Herzegovina (henceforthreferred to as BiH or Bosnia), where it isdebated whether to change the politicalstructure in a more integrative direction. TheDayton Agreement that ended the warcontains elements of both Lijphart’s andHorowitz’s approaches, and the balancebetween them has changed in the course ofimplementation. Therefore, the case issuitable for a comparison of the effectivenessof the approaches in promoting stability.With only one case, the analysis is of courselimited. However, by isolating elements fromeach approach, the case makes possible adirect comparison of the effect of theapproaches on stability. Both the consocia-tional and the integrative approach havebeen criticized for failing to specify theempirical conditions under which they areeffective in promoting stability, and it is theaim of this article to throw some light on thisby comparing the effect of the approaches inpostwar Bosnia.

Stability in the consociational and inte-grative approaches hinges on inter-ethnicelite cooperation and moderation, that is,willingness on the part of the political leadersto compromise and to rule inclusively ratherthan exclusively. The comparison of the two

approaches in the Bosnian case shows thatthe consociational model has been moreeffective in fostering stability. This has beendue to the still deep divisions in the popu-lation, the maximalist objectives of thedominant parties and the numerical balancebetween the groups. Currently, the integra-tive model will not be effective in promotingstability, but a mix of the approaches hasbeen demonstrated to be able to foster mod-eration. Given time and the internationalpresence, it is possible to gradually changethe balance of the mix, and the twoapproaches, while they are conceptual poles,are therefore not incompatible in theBosnian case.

Lijphart’s and Horowitz’sApproaches

Lijphart’s consociational approach andHorowitz’s integrative approach havebecome the two most advocated models ofconflict regulation (Kettley, Sullivan & Fyfe,2001: 8). Both approaches advocateinclusive solutions to ethnic conflict, that is,solutions based on inter-ethnic accommo-dation, bargaining and reciprocity. They are,however, conceptual poles, since they deeplydisagree on whether the institutional struc-ture should be built on the ethnic groups ortranscend them (Sisk, 1996: x; Harris &Reilly, 1998: 141). This is based on differingunderlying assumptions on the importanceof identity and its fluidity and on the moti-vation of elites in ethnic conflicts.

Lijphart’s Theory of ConsociationalDemocracyThe theory of consociational democracy isbased on the assumption that successfulaccommodation of ethnic differences ispossible only through inter-ethnic elitecooperation in institutions that explicitlyrecognize the ethnic divisions and make themthe basis of the rules for decisionmaking,

j ournal o f PE AC E RE S E A RC H volume 41 / number 5 / september 2004570





territorial division of power and publicpolicies. It guarantees the protection of grouprights and recognizes the legitimacy of claimsto national self-determination within theexisting state. Consociational democracy ischaracterized by four institutional devices: (1)a power-sharing government, a so-calledgrand coalition with representatives from allprimary groups; (2) minority veto on issuesthat can infringe on national interests; (3)proportionality in the electoral system and inthe civil service; and (4) ethnic autonomy(Lijphart, 1977: 25–44).

Elite cooperation is argued most often tobe based on a ‘self-negating prophecy’, theidea that the ethnic leaders can realize thegrave dangers posed by ethnic differencesand therefore choose to cooperate (Lijphart,1994: 228).

Horowitz: Cooperation BetweenElectorally Motivated PoliticiansHorowitz strongly criticizes the consocia-tional model for failing to specify how theinstitutional arrangements affect the incen-tives for cooperation faced by the elites.Elites, Horowitz asserts, cannot bepresumed to always want accommodation,and owing to intra-ethnic competition, theelites will furthermore face outbidding andlack the freedom of action to compromise(Horowitz, 1985: 574–579). Moreover,Horowitz argues that consociational democ-racy underestimates the fluidity of identity:identities are not rigid, and non-ethniccleavages can be made salient (Horowitz,2002: 24–25).

The institutional mechanisms aredesigned to create incentives for moderationand multi-ethnicity. Horowitz, therefore,prescribes a preferential electoral system inwhich a candidate’s election depends onattracting votes from outside his/her ethnicgroup. This will give the ethnic parties incen-tives to moderate their position and engagein cross-ethnic appeal (Horowitz, 1991:

141).2 Similarly, federalism should prefer-ably be based on ethnically heterogeneouspolitical units, which is argued to foster inte-grative dynamics, more moderate attitudesand fluid identities (Horowitz, 1985:617–621). Finally, Horowitz favours ‘ethni-cally blind’ public policies that can helpreduce the divide rather than augmenting it(Harris & Reilly, 1998: 141).

Horowitz’s approach is a minimalistapproach that is focused on the dynamics ofinteraction that can foster accommodativeattitudes, but the specific content of thisaccommodation is not crucial (Horowitz,2000: 8–9). Group protection is thereforenot a priori guaranteed in the approach. Theaim of the mechanisms prescribed is to abatethe ethnic divide, to foster inter-ethniccooperation and thereby integrate thegroups.3

Brief Analysis of the ApproachesWhile the approaches share some funda-mental assumptions, they differ significantlyon others. Lijphart stresses the importance ofidentity and its relative permanence, whileHorowitz emphasizes its potential forfluidity. Lijphart relies on the goodwill ofelites and on their dominance vis-à-vis theelectorate. Horowitz, on the other hand,notes the self-interested nature of elites andthe constraints posed by followers.

Both approaches can be criticized for anunder-specification of the necessary con-ditions on which they are built. In addition,each approach has been criticized for specificshortcomings. The consociational structureaddresses many of the demands of the partiesto an ethnic war: it guarantees them a sharein power, it recognizes the legitimacy ofethnic demands, and it accommodates claims

Nina Casper s en CO N F L I C T RE G U L AT I O N I N BO S N I A 571

2 The basic idea is that voters may be swayed to cast lowerpreferences across the ethnic divide.3 The use of the term ‘integrative model’, which may havepositive connotations, does not imply that it is regarded asnormatively superior to the consociational model.





to self-determination. However, this meansthat the political structure becomes squarelybased on ethnicity, and Lijphart’s approachhas been criticized for freezing and entrench-ing the ethnic divisions and for being anineffective system of governance (Brass,1991: 10; Reynolds, 2000: 168–169).Horowitz’s approach has been criticizedmainly for being an essentially majoritariansystem that has inadequate protection ofminority rights, and this is argued to make itinappropriate in cases of deeply dividedsocieties (Reynolds, 2000: 159–160;Lijphart, 2002: 47; Sisk, 2001: 30). In myview, the theory lacks an explanation of localacceptance: Why would leaders of a minoritygroup accept a system which gives them noassurances?4 Why would nationalist partiesaccept a system designed to undermine theirbases of power?

Different expectations follow from theapproaches. A consociational structure islikely to be more acceptable to the localleaders, especially if no group constitutes amajority and if the conflict has been domi-nated by self-determination claims. Consoci-ational institutions will, however, have apropensity for deadlock and immobility,while integrative institutions are expected tobe more effective once established. In theconsociational approach, homogeneousunits are expected to lead to more moderateattitudes and the election of political leadersmore willing to agree to inter-ethnic cooper-ation. In the integrative model, hetero-geneous units are ascribed with this quality.

One important variable is absent fromboth approaches: the international dimen-sion. In case a settlement is guaranteed by anexternal power, this will affect the incentivesof the parties involved and affect both the

likelihood of acceptance of the integrativemodel and the risk of breakdown of theconsociational structure. A coercive strategymay backfire, however, and an imposedstructure is accompanied by the question ofwhat will happen once the internationalguarantors leave? Time may also be ofessence: it can generally be assumed thatidentities are more fluid over a longer periodof time, and even Lijphart does not insist onthe permanency of identities. Both the inter-national dimension and the phase of theconflict should therefore be included in thecomparison. The variables hypothesized toaffect the effectiveness of the approaches arethe demographic balance between thegroups, maximalist goals of the leaders, atti-tudes in the population and internationalpresence.

Comparison of the Two Approachesin Bosnia

The aim of this article is to compare theeffectiveness of the consociational and inte-grative model in fostering stability inpostwar Bosnia. The Dayton Agreement thatended the war contains elements of bothapproaches, and during the course ofimplementation, the balance between con-sociational and integrative elements haschanged. The Bosnian case provides thepossibility for a direct, albeit limited, comparison of the consociational and theintegrative approaches. Internationalorganizations play an important role in theimplementation of the agreement and affectthe workings of the consociational and inte-grative elements. The effect of an inter-national dimension, which is absent fromboth approaches as presented by theirauthors, can therefore be included in theanalysis. Civilian implementation of theDayton Agreement is the responsibility ofthe international High Representative(henceforth referred to as HR, and the


4 Horowitz (2002: 20) argues that his approach will bemore readily accepted since it favours majority groups.However, following an ethnic war, the sentiments of notonly the largest group need to be considered, especially ifthe settlement is based on the existence of a ‘mutuallyhurting stalemate’.





institution as OHR), while elections wereconducted under the authority of the OSCEuntil 2002, when election administrationwas transferred to local authorities.

Consociational ElementsThe most visible elements of the DaytonAgreement are consociational, and initiallythe consociational elements were clearly thestrongest. First and foremost are the jointinstitutions, which are composed of thethree-person Presidency, the Council ofMinisters and the two-chamber Parliamentwith its House of Representatives and Houseof Peoples. One finds in these institutions agrand coalition, minority veto provisionsand parity of representation.

Ethnic autonomy is another importantconsociational feature. Bosnia is composedof two entities with considerable auton-omous powers. Republika Srpska (RS) ispredominantly Serb, while the Federation(FBIH) is bi-national with both Bosniacsand Croats.5 Within the Federation, acomplex system of power-sharing exists withminority veto, parity of representation andconsiderable autonomy for homogeneouscantons and municipalities. Owing to theveto provisions at the central level and thegreat degree of decentralization, power tendsto gravitate to more majoritarian, homo-geneous institutions in the two entities, andethnic autonomy is, to a large extent, thedefining feature of the structure.6

Integrative ElementsEven though the consociational elements aremost visible, one should not overlook theintegrative elements in the Dayton Agree-ment. The ethnic autonomy is based on thecongruence of ethnicity and territory and noton ethnicity itself. This is important since

returns and people’s right to vote in their1991 residence have the potential for creatinggreater residential and electoral heterogeneityand thus undermine the ethnic autonomy ofthe consociational structure.

The second integrative element in theagreement is found in the existence of somemore integrative institutions – that is, insti-tutions that, while they are based on ethnicparity, have no veto provisions and decide bysimple majority. These institutions are theConstitutional Court, the Central Bank, theJoint Interim Commission, the HumanRights Chamber and Commission, theCommission for Displaced Persons andRefugees, the Commission to PreserveNational Monuments and the Commissionon Public Cooperations. Although they arenot legislative institutions, they are insti-tutions of considerable power and they makedecisions that are important from a nationalpoint of view.

Despite these integrative elements, thepolitical structure in Bosnia was initiallypredominantly consociational, but duringthe course of its implementation this hasbegun to change (Sharp, 1997/98: 15).Especially since the expansion of the HR’spowers in December 1997, the integrativeelements within the Dayton Agreement areincreasingly emphasized, and new measureshave been introduced. In addition, a morecoercive strategy has been used to underminethe support of the nationalist parties. One ofthe most important examples of a new inte-grative measure was a draft election law tocreate incentives for cross-ethnic appeals andvoting. The draft laws proposed the intro-duction of a preferential system for theelection of the Presidency and would allowall representatives to vote for the delegates ofthe House of Peoples, which had until thenbeen elected by ‘their own’ ethnic represen-tatives. However, these measures and threedifferent draft laws were rejected before anelection law was finally passed in 2001.


5 I use the term ‘Bosniacs’ as synonymous with the groupthat before 1993 referred to itself as Bosnian Muslims.6 This has become less pronounced with the recent changesto the entity constitutions (see below).





Another possibility for a more integrativestructure was the 2000 ruling by the BiHConstitutional Court that Serbs, Croats andBosniacs should have equal status in bothentities. This decision provided the oppor-tunity for a revision of ethnic autonomy, andthe implementation was strongly pushed bythe HR.

Owing to the mix of the two approachesand the changed balance between theelements, it is possible to analyse both theresponses to a more integrative turn and theeffect of the already existing integrative andconsociational elements. The elements formpart of a complex political system, and eventhough it is possible to separate consocia-tional and integrative elements, as outlinedabove, the functioning of these elements willbe affected by incentives produced by the restof the complex system. This will diminishthe extreme effects of each of the approaches:the consociational elements will be affectedby the greater fluidity emanating from theintegrative approach, while the integrativeelements are embedded in a system of conso-ciational and international guarantees thatmakes acceptance by minorities more likely.

Analysis

In the two approaches, stability is defined bythe existence of political leaders who arewilling and able to engage in inter-ethniccooperation. This will, therefore, be themeasure used in the analysis when com-paring the effect of consociational and inte-grative elements. In addition, responses bylocal leaders to the above-mentioned inte-grative turn will be analysed.

The first of the elements that can becompared directly are heterogeneous andhomogeneous municipalities. According tothe integrative model, heterogeneous unitswill foster moderation and inter-ethniccooperation. On the other hand, the conso-ciational model argues that they will foster

greater extremism. The municipalities will becompared in terms of voting patterns: do theextreme nationalist parties have more supportin ethnically homogeneous or heterogeneousmunicipalities? Stability in the two models isbased on the behaviour of political elites, andvoting patterns will not be a direct measureof stability. However, in Bosnia the wartimenationalist parties have been the least willingto accept inter-ethnic accommodation andcooperation, and support for these partieswill make stability less likely. In hetero-geneous municipalities, inter-ethnic cooper-ation is an important indicator of stability;are the elected politicians willing and able tocooperate as foreseen in the integrativemodel? In these municipalities, mayor andspeaker positions must be allocated to indi-viduals from different ethnic groups, and thisrequired minority representation gives incentives for the parties to become multi-ethnic. Consociational and integrativeelements are thereby combined in hetero-geneous municipalities.

Second, the workings of consociationaland more integrative institutions will becompared. According to the consociationalmodel, integrative institutions will not beaccepted and their establishment will there-fore be fraught with difficulty. On the basis ofthe integrative model, on the other hand, onecan argue that the consociational institutionswill be marred by deadlock and ineffective-ness. The analysis will compare the establish-ment and functioning of the joint,consociational institutions and the integrativeinstitutions. It is important to note that theintegrative institutions also consist of inter-nationally appointed representatives who holdthe chairmanship and often the deciding vote.This international presence provides a guaran-tee that is usually lacking from integrativeinstitutions, but the representatives of thethree constituent peoples still run the risk ofbeing outvoted, and such a risk is not presentin the consociational institutions.






In addition to these integrative insti-tutions, responses to refugee returns will beanalysed, as will responses to the introduc-tion of new integrative measures. This willenable a further analysis of the variablesaffecting acceptance or rejection of integra-tive elements and thereby the possibility forturning the Bosnian structure in a more inte-grative direction. Such a change wouldaddress the problems of immobility andfreezing of identities associated with theconsociational structure, but the risk is thatsuch a change may foster negative reactionsand hence promote instability.

The analysis will be divided into twophases, which will enable an analysis of the effect of phases of the conflict and of theinternational dimension: (1) 1996–97, thefirst two years after Dayton, when the HRperformed the role of assistant to the localimplementation of the agreement; and (2)1998–2002. In December 1997, the HR’spowers were strengthened with the so-called Bonn Powers, and the strategy for

implementation changed markedly. This hasbeen described by some observers as acreeping protectorate or a trusteeship (e.g.Cox, 2001: 12; Cousens & Cater, 2001:129).

Developments in Postwar Bosnia

The political scene in postwar Bosnia has beendominated by the wartime nationalist parties:the Serb SDS, the Croat HDZ and theBosniac SDA. The only significant party withcross-ethnic appeal is the reformed Com-munist Party, the Social Democratic Party(SDP), but even this party predominantlyattracts Bosniac voters (UNDP, 2000: 38).

If we look at election results from 1996 to2002 in Figures 1 and 2, the support fornationalist parties has declined, which indi-cates a moderation in public attitudes.7 The


0

10

20

30

40

50

60

70

80

90

SDS SRS SDA Total

1996 1998 2000 2002

Source: Official election results from http://www.oscebih.org and from http://www.izbori.ba.

Figure 1. Nationalist Party Support in RS

7 The drop in nationalist support in RS in 2000 may bedue partly to the banning of the SRS. However, in 2002,the party again ran in elections without this resulting in areversal of the trend.





nationalist parties made some comeback atthe latest elections in 2002, but this need notbe based on a reversal of moderation, since itcan be explained, at least in part, by a verylow turnout and widespread dissatisfactionwith the performance of the unstable SDP-led alliance government.

Party pluralism is still almost entirelyabsent in the Croat community, where theHDZ continues to enjoy overwhelmingsupport (Bose, 2002: 8; Manning, 2001: 35).The Croats are clearly the smallest group, andthey do not have the institutional protectionoffered by their own entity. This, combinedwith actions by the OSCE and OHR per-ceived to undercut the position of Croats, hasled to a perception of their status being underthreat (e.g. Bose, 2002: 28, 258; Manning,2001: 35–36; HDZ, 2000). Changes in theposition of the political parties are discerniblein all groups, at least on the rhetorical level.As Bose argues (2002: 11): ‘Along with thegrowth of party pluralism . . . some major

parties have modified their strategies andrhetoric in a “moderate” direction . . . theprime example being the SDS. There arereasons to believe that a latent reformisttendency exists within even the HDZ-BiH.’8

The more coercive strategy by the inter-national implementers has not led to moreextreme positions, and there has actually beenremarkably little opposition to imposition oflegislation, removal of obstructionist officials,dismantling of parallel power structures andarrest of indicted war criminals (Cox, 2001:13; ESI, 2000: 33).

An overall change in attitudes is indi-cated, both among the electorate and amongthe political elites, even though the change isnot uniform among the three groups. But


Figure 2. Nationalist Party Support in the Federation

0

10

20

30

40

50

60

70

80

90

HDZ SDA Total

1996 1998 2000 2002

Source: Official election results from http://www.oscebih.org and from http://www.izbori.ba.

8 At its most recent conference, the SDS described itself asa ‘national/people’s party of the democratic centre’, readyto cooperate with all democratic parties in Bosnia and opento all citizens (SDS, 2001). The SDA has likewise pro-claimed itself to be ‘a national/people’s party of the politicalcentre open for all citizens of Bosnia and Herzegovina’(SDA, 2001).





what effect have the consociational and inte-grative elements had on stability in post-Dayton Bosnia? Have they fared equally well(or badly) in fostering stability?

Comparison of Heterogeneous andHomogeneous Municipalities

The municipalities were categorized bytaking advantage of the fact that voting inBosnia is almost exclusively along ethniclines.9 Almost all parties are national parties,and by calculating the vote shares, the ethniccomposition of the municipality can be esti-mated. A heterogeneous municipality wasdefined as a municipality in which the largestgroup constitutes less than 66%.10

Support for Nationalist Parties

1997 Elections In the 1997 municipalelections, the homogeneous and hetero-geneous municipalities differ significantly interms of nationalist support coded as supportfor SDA, the HDZ and the SDS, as well asthe extreme nationalist Serb Radical Party(SRS). Almost all parties are national if notnationalist parties, but these four parties are

the wartime nationalist parties that, whilethey have reformed, remain the least willingto compromise on inter-ethnic issues.11 Asshown in Table I, for the electorally hetero-geneous municipalities, the average national-ist vote share was 85%, while it was 73% forthe homogeneous municipalities. For the sixresidentially heterogeneous municipalities,the difference is even more marked, with anaverage of 92% compared to a 75% averagein the residentially homogeneous municipal-ities.

2000 Elections The results of the 2000elections illustrate the overall development asthe average nationalist vote share fell to 52%.Despite these changes, the same generalpattern as in the first elections can beidentified. In the electorally heterogeneousmunicipalities, the average nationalist voteshare is 62% compared with 51% for thehomogeneous ones. For the residentiallyheterogeneous municipalities, the average is60% compared with 51%. This is a signifi-cant decrease since 1997, and it indicatesthat while residential heterogeneity stillcauses greater extremism, moderation maydevelop faster.

A possible explanation for the greaterextremism in heterogeneous municipalitiescan be found in the defence of group status:in case of heterogeneity, the voters do not feelthat pluralism can be afforded and therefore


9 Elections in deeply divided societies often amount to acensus (Horowitz, 1985: 326).10 For the 1997 elections, I identified 31 municipalities outof 135 as being electorally heterogeneous based on ‘inmunicipality’ and ‘out of municipality’ votes. Residentialheterogeneity was found in 6 municipalities. In the 2000elections, 12 of 146 municipalities were electorally hetero-geneous, and of these 5 were heterogeneous. See Appendix2 for further details.

Table I. Nationalist Party Support as a Function of Heterogeneity (%)

Heterogeneous Homogenousmunicipalities municipalities Difference Violent Other Difference

1997 85 (31) 73 (104) 12* 79 (23) 75 (112) 42000 62 (12) 51 (134) 11** 52 (26) 51 (120) 1

** p < 0.01, * p < 0.05. Number of municipalities in parentheses.

11 Unlike the three other parties, the SRS was not in powerduring the war, but in the Serb-controlled areas it func-tioned as a more extreme opposition to the SDS.





close ranks behind the nationalist forces. Theanalysis gives some support for this expla-nation. The biggest difference in nationalistsupport is found with the Bosniac voters: in1997, the SDA’s average share of the Bosniacvote was 93% in heterogeneous municipalitiesand 72% in homogeneous Bosniac-majoritymunicipalities. This could indicate thatpluralism can best be afforded when thepower-base is secure. For the Croats, the per-ception that their status is under threat maylead to an attitude that pluralism is generallytoo costly, in homogeneous as well as hetero-geneous municipalities, and therefore the voteshare of the HDZ does not differ as markedly.In 1997, the HDZ’s average share of the Croatvotes was 89% in homogeneous municipali-ties and 93% in heterogeneous municipalities.In 2000, the numbers were 71% and 81%.

The even greater extremism associatedwith residential heterogeneity does notnecessarily mean that contact breeds intoler-ance, since people may in fact be living inenclaves rather than inter-mixed.12 Thisoften seems to be the case in postwar Bosnia(Bose, 2002: 36). Actual interaction willtherefore be lacking and, shortly after cessa-tion of hostilities, the spatial proximitybetween the groups can provoke greater fearof domination by the other group thanwould purely electoral heterogeneity.

Fear of domination may be less pro-nounced in larger municipalities, which willtend to be more urbanized and in which thesize of the local government makes forgreater proportionality. Therefore, it is notsurprising that nationalist support is nega-tively correlated with the size of the munici-pality: in 1997, the average support fornationalist parties was 78% in municipalitieswith fewer than 8,000 voters and 71% inmunicipalities with more than 18,000voters. In 2000, the numbers were 54% and

38%. The relationship between heterogene-ity and nationalist support does not dis-appear when this additional variable isincluded, and the two variables ‘relative sizeof largest group’ and ‘number of votes in themunicipality’ together explain 14% of thevariance in 1997 and 15% in 2000.

Ethnic Violence as Hidden Variable? Could the correlation between hetero-geneous municipalities and support fornationalist parties be spurious? Is the differ-ence explained by varying levels of violence?Based on reports from the US State Depart-ment (1993), the UN Commission forHuman Rights (Masowiecki, 1993–95) andHelsinki Watch (1992, 1993), the munici-palities have been coded in terms of theintensity of the war.13 As shown in Table I,in 1997 the average nationalist vote share inthe municipalities that had experienced avery high level of violence was 79%,compared with 75% for the remainingmunicipalities. In 2000, the numbers were52% and 51% respectively.

However, the variable of ethnic violence issignificantly correlated with the two othervariables, and if the data are split in two, aninteresting difference emerges betweenmunicipalities with a very high level ofviolence and the remaining municipalities.As shown in Table II, in municipalities thatexperienced a very high level of violenceduring the war, heterogeneity and the size ofthe municipality together explain 60% of thevariance in nationalist support in 1997 and46% in 2000. In the remaining municipali-ties, the corresponding percentages are only9% and 15%.14


12 Allport (1954: 263, 272) argues that if segregation limitscontacts to ‘casual contacts’, then it tends to cause conflictrather than tolerance.

13 The coding was based on several reports of ‘wilfulkillings’ or ‘summary executions’ or reports of ‘massacres’.The Geneva Convention art. 147 defines ‘wilful killings,torture and inhuman treatment’ as ‘grave breaches’ of theConvention.14 Because of multicollinearity, the interaction effect is notstatistically significant, and the file has, therefore, been splitinstead of including an interaction effect in the regression.





In municipalities that have experienced ahigh degree of ethnic violence, the fear ofdomination is greater, especially in smallercommunities. Therefore, in situations ofdeep divisions following a very intense ethnicwar, heterogeneous municipalities seem tofoster extremism rather than moderation.

Inter-Ethnic Cooperation inHeterogeneous Municipalities If heterogeneity results in the election ofmore extreme political parties, how does thisaffect inter-ethnic cooperation between thelocal representatives? Minority represen-tation is required in municipalities in whicha minority group won more than 20% of theseats, and inter-ethnic cooperation has there-fore been imposed in all heterogeneousmunicipalities. This has, however, not beenwithout tension.

In 1997, the power-sharing requirementswere rejected or ignored in many municipal-ities, and the OSCE imposed awards ofmulti-ethnic representation and appointedmayors (Manning, 2001: 27). In some cases,this led to tensions and in others the require-ments were still rejected. The implemen-tation process went much more smoothly in2000, and no municipalities were deniedcertification based on failure to comply withthe power-sharing requirements (internalOSCE BiH report; Manning, 2001: 27).The implementation fostered inter-ethnicalliances as well as intra-ethnic ones, and

even the SDA, the HDZ and the SDS onseveral occasions offered coalition agree-ments to parties from other ethnic groups.Still, no inter-ethnic alliances betweensignificant parties were created before theelections, and the SDP was the only signifi-cant party that made cross-ethnic appeals.The minority representation provisionsrequire representation of candidates, notparties, and the parties could thereforebenefit from being multi-ethnic, since theythen would not have to share power, but theincentives for multi-ethnic appeals are stilltoo weak compared with the incentives formono-ethnic nationalist appeals. However,pragmatic power considerations force theparties to make alliances, and the minorityrepresentation provisions make it necessaryfor these alliances to cross the ethnic divide.The development can therefore be seen as amix of development of greater trust and elec-toral ambitions. On this basis, however,actual inter-ethnic cooperation may stillprove very difficult.

Heterogeneous and HomogeneousMunicipalities The comparison showed that heterogeneousmunicipalities fostered support for national-ist parties rather than moderation. Thisseemed to be explained by the defence ofgroup status and fear of domination. In caseof heterogeneity, party pluralism can beregarded as being too costly, since it could


Table II. Nationalist Party Support as a Function of Heterogeneity and Size of the Municipality

Municipalities with high level of violence Remaining municipalities

1997 2000 1997 2000

Heterogeneity –0.55 (0.15)** –0.68 (0.16)** –0.30 (0.12)* –0.31 (0.13)*Size –0.05 (0.01)** –0.03 (0.01)** –0.04 (0.02)* –0.08 (0.02)**R2 0.60 0.46 0.09 0.15N 23 26 112 120

** p < 0.01, * p < 0.05. Coefficients (standard deviations) in parentheses.





reduce the political power of the group’s rep-resentatives. Extremism was greatest whenthe groups were residentially inter-mixed.However, the analysis did indicate that resi-dential heterogeneity may in the longer termlead to moderation. The variance explainedby the ethnic composition of the municipal-ity was especially great in municipalities thathad experienced a very high level of violenceduring the war. This supports the assertionthat integrative structures especially fosterinstability following an intense war.

Inter-ethnic cooperation was moreproblem-free and widespread in the secondphase, even though incentives for mono-ethnic appeals were still stronger than incen-tives for multi-ethnicity. The mix ofconsociational and integrative elementscombined with the greater degree of partypluralism thus fostered a greater degree ofstability in the second phase.

Comparison of Consociational andIntegrative Institutions

The joint institutions were created after thefirst elections in 1996 and came intofunction in the beginning of 1997. Initially,they were boycotted by the Serb representa-tives (HR, 1996c), but a short-lived boycottof the state’s main institutions is not surpris-ing considering that the Dayton Agreementwas partly imposed.15 What may be surpris-ing is the fact that the establishment of themore integrative institutions was relativelyeasy. The Joint Interim Commission, theHuman Rights Chamber and Commission,the Commission for Displaced Persons andRefugees, the Commission to PreserveNational Monuments and the Commissionon Public Cooperations were all establishedwithin the first year after Dayton. Thenational members were appointed by the

parties, and the establishment seems to havebeen problem-free (HR, 1996a,b). Theestablishment of the Constitutional Courtand the Central Bank awaited the September1996 elections. The Constitutional Courtwas established without significantproblems, while the Central Bank wassomewhat more contentious, and the HDZand the SDS delayed the passing of thenecessary law (ESI, 2000: 51). Apart fromthe Central Bank, the establishment wasthen remarkably easy, and even the CentralBank was established without boycotts. But,once established, how did the consociationaland integrative institutions function?

Consociational InstitutionsAgreeing on legislation in Parliament turnedout to be very difficult, owing to the con-current majority and veto provisions in theconstitution.16 In the first year of the StateParliament’s existence, only ten laws werepassed, and the problem continued in thesecond phase: in 1998–2000, an average offive laws a year were passed by Parliament.After the alliance government came to powerin February 2001, however, things began tochange. Despite continued obstruction bythe Serb caucus in Parliament, close to 40laws were passed between February 2001 andAugust 2002.17 At least 11 of these were lawsthat had previously been imposed by theHR, and the increasing number could seemto reflect greater pragmatism as regards theinternational authority. In addition,however, some important ‘new’ laws were


15 A discussion of the extent to which the Dayton Agree-ment is an imposed settlement is, unfortunately, beyondthe scope of this article.

16 According to the Constitution, legislation can beblocked by two-thirds of the representatives of one ethnicgroup in the House of Representatives or by a majority ofthe representatives of one ethnic group in the House ofPeoples.17 The counting of the number of laws passed is for1997–2000, based on the Reports of the High Represen-tative for Implementation of the Peace Agreement to theSecretary-General of the UN. After 2000, such infor-mation is no longer included in the reports and the sourcesused instead are ICG (2002: 19) and BiH Media Round-up, 24 May 2001.





passed. In particular, the passing of theElection Law in August 2001 was seen as abreakthrough: not only the moderate partiesbut also the HDZ, the SDA and the SDShad been willing to compromise.

Integrative InstitutionsAlmost from the very beginning, the HRnotes how effectively the institutionsfunction (HR, 1996c), but despite thisoptimism, problems with insufficientcooperation from the political authorities areacknowledged. The human rights insti-tutions, in particular, continually facedproblems with local authorities refusing toimplement their decisions and recommen-dations (HR, 1996c, 1997a), and fundingfor those institutions as well as for the Con-stitutional Court was lacking.

Cooperation with local authoritiesimproved in the second phase. Funding wasforthcoming, and implementation was nolonger the exception; for example, theimplementation of decisions from theHuman Rights Chamber increased from33% in 1999 to 73% in 2001 (HR, 2001).The Central Bank has also succeeded inimplementing a common currency, which isnow used in both entities (Bose, 2002:112). When the Constitutional Court madeits decision, in July 2000, on the con-stituent peoples, there was severe criticismfrom Croat and Serb parties, since the Serband Croat judges had voted against thedecision. However, the decision wasaccepted, and the parties took part in itsimplementation.

Consociational and IntegrativeInstitutionsThe effectiveness of both consociational andintegrative institutions has increased in thesecond phase. This is no coincidence, sincethe effectiveness of most of the integrativeinstitutions depended on the cooperation ofconsociational structures. What does this,

along with the relatively unproblematicestablishment of the integrative institutions,tell us? First, it must be remembered that allthe integrative institutions are run underinternational supervision and their effective-ness may have been dependent on theultimate international control. Second, whiledealing with important issues, the integrativeinstitutions lack the legislative power and thesymbolic importance of the consociationalinstitutions, which may have made theirestablishment easier. However, even whenthese two points are considered, the com-parison of integrative and consociationalinstitutions suggests that integrative insti-tutions may, under some circumstances, beacceptable to the local parties and functioneffectively.

Turning Implementation in a MoreIntegrative Direction

Over the years, the implementation strategyhas changed, and more emphasis is now puton integrative elements. But one of the mainproblems for the integrative approach is theproblem of local acceptance, and by focusingon responses to integrative measures, vari-ables affecting rejection or acceptance can beanalysed.

Minority ReturnsMinority returns are returns to areas where adifferent ethnic group retains militarycontrol and population majority. This is oneof the most important elements if theDayton structure is to be turned in a moreintegrative direction, since it will help createthe necessary heterogeneity. Such returns areheavily dependent on both the willingness ofpolitical leaders to accept increased hetero-geneity and the forcefulness of the inter-national strategy. Significant developmentshave occurred in this field.

In the first year after Dayton, it wasdecided by the Peace Implementation






Council18 to prioritize majority returns,since it was argued to be too early forminority returns, owing to the still volatilesecurity situation. In the first year, therefore,the number of minority returns was offset bymovements of continued ethnic separation(HR, 1996a). As of 31 August 1999,minority returns numbered merely 100,714people, but in the last quarter of 1999 thelong-awaited breakthrough appeared, andminority returns took place even in areas thathad experienced genocidal violence (ICG,2000: 8). In 1999, the number of registeredminority returns was 41,007; in 2000 it was67,445; and in 2001 it was 92,061.19 Thisdoes not mean that minority returns are nowunproblematic, but outright obstruction haswaned, and the security situation hasimproved considerably (HR, 1999, 2000a,b,2002). Despite this progress, minorityreturns are not likely to recreate the prewarethnic mix. The numbers are still too small,and those who return often return tominority enclaves rather than mix with themajority ethnic group (Bose, 2002: 36).

An Integrative Electoral System?On the basis of the Dayton Agreement, theOSCE adopted an electoral system that useda PR system for most elections, but wascombined with ethnically defined lists for theelection of the three-person Presidency.Furthermore, members of the Upper Housesof the Bosnian Parliament and the Feder-ation Parliament are elected by the ethnicrepresentatives in the Lower Houses (i.e.Croats elect Croats, Serbs elect Serbs, etc.).However, after the nationalist victory in the1996 and 1997 elections, debates soon beganover the possibility of changing the electoralsystem in a way that would foster modera-tion and multi-ethnicity. The problem was

that the law should be passed by the BosnianParliament, and the nationalist parties werefar from interested in an electoral systemdesigned to undermine their power-base.Different draft laws were rejected three timesbefore a law that preserved the consocia-tional procedures was adopted in August2001 (see http://www.izbori.ba).20

The issues that proved contentious in thedraft laws were the articles introducing someform of preferential system and wateringdown of the consociational structures. Mostcontroversial have been suggestions to adopta preferential system for the election of thePresidency and the suggestion to allow allrepresentatives to vote for delegates for theHouse of Peoples (BiH Media Round-up, 22August 2001). The HDZ argues that theintroduction of a preferential system wouldabolish the Croat veto, since Croats wouldnot exclusively elect their representatives,and they saw the draft election law as promoting the ‘for Croats . . . absolutelyunacceptable voting principle of one manone vote’ (HDZ, 2000: 50, 60). Since a per-manent election law had not been adoptedfor the 2000 elections, the OSCE Head ofMission, Ambassador Barry, adopted thedraft election law for the election of delegatesfor the House of Peoples, and this caused aprolonged crisis between the HDZ and theOSCE, which the HDZ were able to use toconsolidate their electoral base. Therefore,the attempt to impose an integrative solutionbackfired.

Like the HDZ, the Serb parties haveexpressed opposition to the introduction ofintegrative elements in the electoral system(Bose, 2002: 239). The Bosniac parties, onthe other hand, voted against the draft law inJune 2001, because of the lack of integrative


18 The PIC comprises 55 countries and agencies thatsponsor and direct the peace implementation process.19 Numbers from UNHCR’s website, http://www.unhcr.ba.

20 The law left empty the procedures for the election of thePresidency, the House of Peoples and the Presidents of RSand FBiH, pending the implementation of the Consti-tutional Court Decision. However, apart from ethnicquota requirements in RS, the changes imposed by the HRin 2002 are minor (OHR, 2002a).





measures, and they regard as discriminatorythe consociational provisions (BiH MediaRound-up, 14, 22 June 2001).

The Constitutional Court DecisionA ruling in July 2000 by the BiH Consti-tutional Court forced the issues of consocia-tional and integrative structures on to thepolitical agenda. In its decision on the con-stitutionality of the Entity Constitutions, thecourt found elements of both constitutionsto be unconstitutional, since the BosnianConstitution designates Bosniacs, Serbs andCroats as constituent peoples throughout theterritory of BiH. They must, therefore, havepolitical equality in both entities (Perry,2002: 2).21

All-party talks were held in order to reachagreement on the implementation, and on27 March 2002 the so-called Sarajevo Agree-ment was signed.22 The agreement maintainsthe consociational structure in the Feder-ation, but now also with representation ofSerbs and ‘Others’. In the FederationGovernment, there are quotas for represen-tation of each constituent people, but theconsociational guarantees have been slightlyweakened since the use of the minority vetohas been limited.23 In Republika Srpska, aCouncil of Peoples is established, which hasveto rights when vital national interests areconcerned, and ethnic quotas in governmentare required. The ethnic-quota requirementhas integrative aspects, since decisionmakingis majoritarian, and it gives the parties incen-tives to run on multi-ethnic slates. The

changes disconnect group-based represen-tation from territory, but despite the slightweakening of the consociational guaranteesand the introduction of some integrativeelements in the formation of government,the changes maintain a structure based onethnicity. The effects still remain to be seen,but it is nevertheless interesting to analysethe position of the parties, since there was asharp division along ethnic lines as to theproper mix of consociational and integrativeelements.

The parties with a Croat prefix submitteda joint proposal in February 2002, but theHDZ did not sign the Sarajevo Agreement,owing to the limits put on the minority vetoand the integrative elements in the govern-ment structure. As Ante Jelavic, then HDZPresident, put it: ‘If the Croats get someministerial positions . . . practically, it will bethe Ministers who suit the Serb and Bosniakparliamentary majority’ (quoted in BiHMedia Round-up, 28 March 2002).

The Serb parties argued that they werewilling to change the RS Constitution in acivic direction, based on majoritarian andformally non-ethnic principles.24 They didnot sign the parts of the agreement pertain-ing to the definition of vital national inter-ests, which was considered too broad – thatis, the structure was too consociational. Inaddition, the provisions on ethnic quotas ingovernment were not signed. When the Serbparties of the RS Assembly unanimouslyamended the RS Constitution, theseelements departed from the Sarajevo Agree-ment. The parties declared themselveswilling to accept multi-ethnic represen-tation, but not to the extent suggested.25


21 For example, the court found the wording in thepreamble of the RS Constitution ‘State of the Serb people’to be unconstitutional. 22 On 19 April 2002, the HR decided to impose com-pletely the amendments of the Federation Constitution(OHR, 2002b,c). The amendments of the RS Constitutionmade by the RS Assembly were partially altered (OHR,2002d).23 The use of the ‘vital national interest clause’ has nowbeen limited. Agreement on the Implementation of theConstituent Peoples’ Decision of the Constitutional Courtof Bosnia and Herzegovina, 27 March 2002, art. 4, 5(http://www.ohr.int).

24 For example, Simić from the Alliance of IndependentSocial Democrats (SNSD) was dissatisfied with the agree-ment because it emphasized the ‘concept of the nationrather than that of man and his liberties’ (quoted in BiHMedia Round-up, 3 April 2002).25 The then RS Prime Minister Ivanić announced duringthe negotiations that the government should be made moremulti-ethnic (BiH Media Round-up, 26 December 2001).





The proposal from the FBiH Consti-tutional Commission emphasized the civicnature of the Federation and de-emphasizedthe concept of collective rights (BiH Feder-ation House of Representatives, Consti-tutional Commission, 2002).26 Thus, whereBosniacs are a majority, a more integrativestructure is advocated. In RS, on the otherhand, the Bosniac parties have put moreemphasis on the protection of national inter-ests.27 What matters to the SDA is that thesolutions in the entities are symmetrical, andthe party rejected the multiparty agreement,owing to its asymmetry (BiH Media Round-up, 28 March 2002). One could argue thatthis is because such symmetry would makethe entities redundant and make Bosniacsmost powerful as the largest group overall. Tothis end, the party would accept solutionsthat reduce the consociational elements.28

Responses to Integrative MeasuresThe analysis of the responses to integrativemeasures showed that, on the central level,the Croat and the Serb representatives preferconsociational structures whereas theBosniac representatives prefer a more inte-grative structure. This supports the assertionthat minority groups support a consocia-tional structure, especially parties that haveespoused secessionist ideologies. Althoughthey are the largest group, the Bosniacs donot constitute a majority. However, they maybe approaching a majority, and under thecurrent political structure, their plurality isof no political value at the central level (Bose,2002: 257). In addition, the Bosniac parties’preference for an integrative solution should

be seen in the context of their unitary goal.Thus, the variable numerical balance inter-acts with maximalist goals. On the locallevel, the preferences are different, but alsobased on minority–majority positions. TheSerb parties are opposed to a consociationalstructure in RS, whereas the Bosniac partiessupport such a structure in RS, but a moreintegrative structure in the Federation. TheCroat parties demand a more consociationalstructure, and the HDZ perceives integrativemeasures as Bosniac domination ‘under theguise of civic majority’ (Jelavić quoted inBiH Media Round-up, 20 April 2001).

As regards the integrative elements (returnsand the integrative institutions) contained inthe Dayton Agreement, the passing of timeand the more forceful international strategyhave led to improvements. The introductionof new integrative elements has, however, metwith sharp resistance and seems likely to causeinstability rather than stability. It is unlikelythat significant integrative elements will bevoluntarily agreed to as long as the Croat andSerb nationalist parties hold sufficient votingpower. Even if the HR is prepared to face astandoff with the nationalist parties, theimposition of an unpopular integrative struc-ture may make it even more difficult to createa self-sustaining peace. In the resistance tointegrative elements, an important elementwould seem to be the interest of the politicalelites in preserving their hold on power. Theyare, therefore, far removed from Lijphart’selites that choose to cooperate in the face ofthreatening mass antagonisms and are muchcloser to Horowitz’s electorally motivatedpoliticians. But it is precisely motivations suchas these that make them resist a change to amore integrative structure.

Conclusion: Good FencesSometimes Make Good Neighbours

The debate between the supporters of theconsociational model and supporters of the


26 The FBiH Commission also had Croat, Serb and ‘Other’members. However, Bosniacs dominated it, owing toHDZ’s boycott, and the Croat members chose to submittheir own proposal.27 The SDA criticized the RS constitutional amendmentsfor being discriminatory and ensuring Serb domination(BiH Media Round-up, 15 April 2002).28 The SDA suggested that they would accept non-legis-lative bodies in both entities as mechanisms for the pro-tection of vital national interests (Bosnia Daily, 2002).





integrative model has at times been harsh,and the choice between the models is animportant one: should structures designed topromote stability be built on ethnic groupsor should they transcend them. Bothapproaches suffer from an under-specifi-cation of the empirical conditions that affecttheir effectiveness, and in this article, I havesought to identify possible variables by com-paring the approaches in postwar Bosnia.

The comparison of the approaches in thetwo phases showed that they had quitedifferent effects on stability. Immediatelyfollowing the cessation of hostilities, theintegrative approach fostered instabilityrather than stability. In later phases of de-escalation, however, this effect seemed towane, and in the longer term the integrativeapproach may foster more moderate atti-tudes and create incentives for inter-ethniccooperation. The effect of the phases of theconflict and the local experience of violencepointed to an important variable: the inten-sity of the conflict and the resultingdivisions. But even in the later phase, theincentives for multi-ethnic appeals were notsufficiently strong. Integrative institutionsworked more smoothly than consociationalinstitutions in both phases, and their estab-lishment was remarkably problem-free, butthis seemed to depend on the ultimate inter-national control.

While the integrative elements came towork better in the second phase, the intro-duction of new integrative elements has beenrejected by the Croat and Serb parties. Thepositions on consociational and integrativesolutions are sharply divided along ethniclines and follow local majority–minorityconstellations. The numerical balancematters, along with the maximalist objectivesof the dominant parties.

Therefore, in the Bosnian case, the con-sociational model has been more effective inpromoting stability, despite the internationalpresence that makes the need for local

acceptance less pressing. The greater effec-tiveness of the consociational model has beendue to the deep divisions in the population,the dominance of self-determination claimsin the conflict and the absence of a majoritygroup. Attitudes in the general populationand the maximalist objectives of thedominant forces are not unrelated: the abilityof leaders to sustain radical demands isaffected by sentiments in the mass popu-lation, but these are strongly influenced bythe goals and rhetoric prevalent among thepolitical leaders. Furthermore, both of thesevariables can be influenced by the twoapproaches, but they are predominantlyexogenous. The last variable, numericalbalance between the groups, is whollyexogenous to the approaches.

In Bosnia, a change to an integrativestructure clearly seems premature, but a mixof the approaches has been demonstrated tobe able to foster moderation. Given time andthe international presence, it is possible togradually change the balance of the mix ofconsociational and integrative elements, andby including these two variables in theanalysis, the two approaches can be regardedas compatible. Such combinations mayprove more effective in fostering stability –more effective in building fences low enoughfor good neighbourliness to develop.

Appendices

Appendix 1. Acronyms andAbbreviationsFBiH: Federation of Bosnia and Herze-govina HDZ (BiH): Croatian Democratic Com-munityHR: High RepresentativeOHR: Office of the High RepresentativeOSCE: Organization for Security and Co-operation in EuropeRS: Republika Srpska SBiH: Party for Bosnia and Herzegovina






SDA: Party of Democratic Action (Bosniac) SDP: Social Democratic PartySDS: Serb Democratic Party SRS: Serb Radical Party

Appendix 2. Comparing NationalistSupport in Homogeneous andHeterogeneous Municipalities

Identification of Municipalities as Hom-ogeneous or Heterogeneous Coding wasbased on the voting patterns in the munici-palities. The SDP is the only significant partywith some multi-ethnic appeal, but its appealbeyond Bosniac voters is assessed to belimited, and it was therefore coded asBosniac. Ideally, census results would havebeen used, but the last census is from 1991and does not reflect the postwar ethnic com-position.

Nationalist Parties Nationalist partysupport was coded as support for SDA,HDZ, SRS (only 1997) and SDS. The SDAran in coalition with the more moderateSBiH in some municipalities, but since SDAwas the dominant party, this was coded asnationalist support. Other extreme national-ist parties ran in some municipalities (e.g. theCroat Party of Rights), but their support isnegligible and has not been included in thenationalist vote share.

Parties and Independent CandidatesIncluded In 2000, 68 parties and 18 inde-pendent candidates took part in the elec-tions. Due to this vast number, only partiesand independent candidates with a voteshare of at least 3% were included.

Coding of Municipalities with Very HighLevel of Violence The list of areasidentified as having experienced a very highlevel of violence looks as follows: Banja Luka,Bratunac, Brčko, Doboj, Foča, Gacko,Goražde, Konjic, Kotor Varoš, Ključ,

Kozarac, Prijedor, Sanski Most, Srebrenica,Travnik, Trebinje, Višegrad, Visoko, Vitez,Vlasnica, Zvornik. These correspond with 26municipalities in 2000, since six of the areasare divided into more than one municipalunit.

References

Allport, Gordon, 1954. The Nature of Prejudice.Reading, MA: Addison Wesley.

BiH Federation House of Representatives, Con-stitutional Commission, 2002. Amendments tothe Constitution of the Federation of Bosnia andHerzegovina, 2 February.

BiH Media Round-ups, daily news reviewsprovided by OHR (http://www.ohr.int).

Bose, Sumantra, 2002. Bosnia After Dayton.London: Hurst.

Bosnia Daily, 2002. ‘Amendments: February 26Deadline’, 20 February.

Brass, Paul, 1991. Ethnicity and Nationalism:Theory and Comparison. London: Sage.

Cousens, Elizabeth & Charles Cater, 2001.Towards Peace in Bosnia: Implementing theDayton Accords. London: Lynne Rienner.

Cox, Marcus, 2001. ‘State Building and Post-Conflict Reconstruction: Lessons from Bosnia’,paper for the project The Rehabilitation ofWar-Torn Societies, Centre for Applied Studiesin International Negotiations, Geneva.

ESI, 2000. Reshaping International Priorities inBosnia and Herzegovina: Part Two, Inter-national Power in Bosnia, 30 March(http://www.esiweb.org).

Harris, Peter & Benjamin Reilly, eds, 1998.Democracy and Deep-Rooted Conflict: Optionsfor Negotiators. Stockholm: InternationalInstitute for Democracy and Electoral Assist-ance.

HDZ, 2000. Response: From the Croat Leadershipon the Process of Deconstituization of the Croatsin BiH. Sarajevo: Presidency of the BiH,Office of the Croat Member.

Helsinki Watch, 1992. War Crimes in Bosnia-Hercegovina. New York: Human Rights Watch.

Helsinki Watch, 1993. War Crimes in Bosnia-Hercegovina. Vol. II. New York: Human RightsWatch.






Horowitz, Donald, 1985. Ethnic Groups inConflict. Berkeley, CA: University of Cali-fornia Press.

Horowitz, Donald, 1991. A Democratic SouthAfrica? Constitutional Engineering in a DividedSociety. Berkeley, CA: University of CaliforniaPress.

Horowitz, Donald, 2000. ‘Some Realism AboutPeacemaking’, paper presented at the con-ference ‘Facing Ethnic Conflicts’, Center forDevelopment Research, Bonn, 14–16December.

Horowitz, Donald, 2002. ‘ConstitutionalDesign: Proposals versus Processes’, in AndrewReynolds, ed., The Architecture of Democracy.Oxford: Oxford University Press (15–36).

HR, 1996a. Report of the High Representativefor Implementation of the Peace Agreement tothe Secretary-General of the United Nations,1st report (http://www.ohr.int).

HR, 1996b. 3rd report to the UN Secretary-General.

HR, 1996c. 4th report to the UN Secretary-General.

HR, 1997a. 5th report to the UN Secretary-General.

HR, 1999. 15th report to the UN Secretary-General.

HR, 2000a. 16th report to the UN Secretary-General.

HR, 2000b. 17th report to the UN Secretary-General.

HR, 2001. 19th report to the UN Secretary-General.

HR, 2002. 21st report to the UN Secretary-General.

ICG, 2000. Bosnia’s Refugee Logjam Breaks: Is theInternational Community Ready? Brussels:International Crisis Group.

ICG, 2002. Bosnia’s Alliance for (Smallish)Change. Sarajevo & Brussels: InternationalCrisis Group.

Kettley, Carmen; James Sullivan & Jessie Fyfe,2001. Self-Determination Disputes andComplex Power Sharing Arrangements: A Back-ground Paper for Debate. Workshop, 9–10February, Pembroke College, Cambridge Uni-versity (http://www.ecmi.de/cps/download/background1.pdf ).

Lijphart, Arend, 1977. Democracy in Plural

Societies. New Haven, CT: Yale UniversityPress.

Lijphart, Arend, 1994. ‘Prospects for Power-Sharing in the New South Africa’, in AndrewReynolds, ed., Election ’94 South Africa: TheCampaigns, Results and Future Prospects.London: James Currey (221–231).

Lijphart, Arend, 2002. ‘The Wave of Power-Sharing Democracy’, in Andrew Reynolds,ed., The Architecture of Democracy. Oxford:Oxford University Press (37–54).

Manning, Carrie, 2001. ‘Warlords into Demo-crats? Elections and Moderation in Post-WarBosnia and Hercegovina’, paper prepared forthe annual meeting of the American PoliticalScience Association, San Francisco, CA, 30August–1 September.

Masowiecki, Tadeusz, 1993–95. Periodic reportson the situation of human rights in the terri-tory of the former Yugoslavia. UN Com-mission on Human Rights.

OHR, 2002a. Decision amending the BiHElection Law in accordance with the newEntity Constitution, 19 April.

OHR, 2002b. Press release: ‘Process of Consti-tutional Change in Bosnia and Herzegovina’sEntities Is Completed’, 19 April.

OHR, 2002c. Decision Amending the Consti-tution of the Republika Srpska, 19 April.

OHR, 2002d. Decision Amending the Consti-tution of the Federation of Bosnia and Herze-govina, 19 April.

Perry, Valery, 2002. ‘Constitutional Reform andthe “Spirit” of Dayton’, ECMI Issue Brief no.7, February.

Reynolds, Andrew, 2000. ‘Majoritarian or Power-Sharing Government’, in Markus M. L.Crepaz, Thomas A. Koeble & David Wilsford,eds, Democracy and Institutions: The Life Workof Arend Lijphart. Ann Arbor, MI: Universityof Michigan Press (155–195).

SDA, 2001. Declaration, Third SDA Congress,Sarajevo, 13 October.

Sharp, Jane, 1997/98. ‘Dayton Report Card’,International Security 22(3): 101–137.

Sisk, Timothy, 1996. Power Sharing and Inter-national Mediation in Ethnic Conflicts.Washington, DC: United States Institute ofPeace Press.

Sisk, Timothy, 2001. ‘Peace Making in Civil






Wars’, Kroc Institute Occasional Paper 20, 2March (http://www.nd.edu/~krocinst/ocpapers/op_20_2.PDF).

SDS, 2001. Programme Principles. Banja Luka, 24December.

UNDP, 2000. Early Warning System, report, June(http://www.ews.undp.ba).

US State Department, 1993. Reports on WarCrimes in the Former Yugoslavia Pursuant toUN Security Council Resolution 771.

NINA CASPERSEN, b. 1976, MSc in Com-parative Politics and in Political Science (LSE,2000; University of Copenhagen, 2002); PhDcandidate in Government, London School ofEconomics and Political Science (2002– ).Research interests: elite behaviour in ethnicconflicts, intra-ethnic competition, formerYugoslavia. Publications in Southeast EuropeanPolitics and Journal on Ethnopolitics andMinority Issues in Europe.

journal o f PE AC E RE S E A RC H volume 41 / number 5 / september 2004588





Documents

Good Fences Make Good Neighbours? A Comparison of Conflict-Regulation Strategies in Postwar Bosnia