Consociational electoral engineering revisited: findings from Bosnia and Herzegovina.

• Author: Rozic, Peter
• Position: POLSCI PAPERS - Report

Introduction

After the three and a half years-long war, hundreds of thousands of casualties, and millions of people displaced, the architects of the 1995 Dayton Agreement engineered a consociational political arrangement for Bosnia and Herzegovina (hereafter "BiH") (3). This arrangement aimed at democratic peace through sectarian power sharing and elections. The policy included territorial self-government and the incorporation of all the warring parties in the electoral system. The plan was based on scholarly recommendations, which presumed that the power of the ballot box was eventually to negate the attraction of ethnic violence, temper the ethno-nationalist parties and allow for an appreciation of voices marginalized by extreme nationalism (Lijphart 1969; Lijphart 1977; Lijphart et al. 1993). However, as the findings of this study show, the Bosnian consociational system has produced mixed results. The salience of non-nationalist parties has increased over time. Yet, consociationalism has tended to encourage and solidify ethnic parties and cleavages at the expense of non-nationalist parties.

Consociational policies have shown success in Northern Ireland and Belgium but have often failed in Lebanon and several African countries. Despite these ambivalent outcomes, only scant replicable research has assessed the voting behavior in consociational societies, particularly over extended periods of time (Bakke and Wibbels 2006; Hartzell et al. 2001; Hartzell and Hoddie 2003). This lacuna is puzzling for several reasons. The Dayton Accords have been widely praised as most ideal-typical (Belloni 2004) and a "classic example of consociational settlement" (Bose 2002, 216). Yet, it remains unclear what explains post-conflict voting in a society bound by consociationalism. Do nationalistic issues that had fueled conflicts still matter in elections? What is the long-term effect of consociational electoral engineering in societies with significant ethnic cleavages?

The objective of this study is to examine nationalist voting in consociational societies, using the case of BiH. After presenting the Bosnian consociational model, this article considers determinants of nationalist voting and provides a theory of negative and indirect electoral effects of consociationalism. The theory is disaggregated into a twofold hypothesis, which is based on the predicted voting behaviour of each of the main ethnic groups, relative to their position in the consociational arrangement. The subsequent typology of post-conflict political parties provides the definitional crux of this research. The original typology identifies BiH's non-nationalist and non-exclusionary parties. The ensuing regression analysis and in-depth study of the 1996-2010 election period across Bosnian cantons and federal entities show a movement toward non-nationalist parties, as well as a divisive differentiation between ethnic groups. This article finds that the rigid consociational structures may help sustain peace but solidify the underlying tensions that led to the conflict.

Revisiting Consociational Electoral Engineering

Desired vs. unanticipated effects

Consociational power sharing coupled with the power of elections is a popular policy prescription for post-conflict societies where segmental cleavages deeply divide them. Since the absence of meaningful opportunities for all segments to express their political interests may lead the unrepresented groups to violent and undemocratic actions (Whitefield 2002), Arend Lijphart suggested the consociational solution. The model assumes that the incorporation of all the parties into a decentralized form of self-government orients the vested interests of different actors towards stability, while the introduction of elections signals a move towards democratic peace (Nagle and Clancy 2010; O'Leary and McGarry 2012; Wolff 2011). The expected emergence of multiple political cleavages is to subdue the divisive ethnic cleavage through a multiparty system (Carothers 2006; Dalton 1988).

Scholars have often found elections in consociational theories controversial. Democratic transitions are precarious, particularly when social cleavages rely on ethnic diversity or border disputes (Horowitz 1985; Lijphart 1977; McFaul 2002). The early stages of democratization may render these issues problematic and solidify the ethnic cleavage for several reasons. Nationalists may exploit the increased freedoms, elections and border disputes to hijack public debate for illiberal ends (Gibler and Tir 2010; Snyder 2000). The stress on elite cooperation may ignore post-conflict struggles over scarce state resources that are often politicized along ethnic lines. Moreover, with many built-in opportunities to veto governmental decisions, consociationalism can encourage inefficiency and rigidity (Cammett and Malesky 2012, 986). In short, consociational electoral engineering may reduce the strength of nationalist politics to a much lesser degree than intended.

Consociationalism in BiH

In Bosnia and Herzegovina, two features of consociationalism particularly contextualize post-conflict voting: administrative divisions and the electoral system. Administrative divisions followed the priority of Bosnian transition from ethnic warfare to multiethnic peace, over the transition from socialism to democratization (Bojkov 2003, 42). To fulfill the consociational expectation of peace by decentralization and territorial self-government (McGarry and O'Leary 2004; Wolff 2009), the Dayton Accords created three levels of political structuring subordinate to the central state government. The first consists of two governing entities, the Federation of Bosnia and Herzegovina and the Republika Srpska (RS). (4) The second consists of ten cantons within the Federation, each having its own government, courts and police force. On the third level are municipalities with no legislative power. Each of these governmental levels is subject to elections. (5) The question of voter franchise adds to the complexity of the electoral system. Dayton provisions divide citizens into three different ethnic "constituent peoples": Bosniak, Serbs and Croats. In order to bring peace, the model allows for the assigning of political power, representation and rights across ethnic lines only. (6)

The vast powers granted to the international actors by Dayton further complicate the system. Between 1996 and 2002, elections were organized by the Organization for Security and Co-operation in Europe (OSCE). Given the task to monitor peace and elections, the OSCE promoted the optimism of the consociational recipe of Dayton. The OSCE first believed that free and fair elections on a continuous basis bring stability and recovery to a country divided by extreme nationalist political leaders (OSCE 1998). Second, the OSCE was given power to vet candidate lists. Among the "hostile" candidates, the OSCE often removed nationalist politicians. The Office of the High Representative (OHR), holding final authority in the country, can also remove elected officials and candidates if they are found to obstruct the peace process. (7) Political parties have acknowledged their dependence on these authorities and started to comply with international non-nationalist plans.

Methodology

Consociational Effects of Ethnicity

The principal hypothesis ([H.sub.1]) is twofold, concerning separately three major BiH's ethnic groups. The overall expectation is that an overt ethnic cleavage represents a significant electoral predictor. [H.sub.1]: In societies bound by consociationalism, members of an ethnic group are more likely to elect officials according to their ethnic lines. The rationale is that members of an ethnic group vote for candidates that belong to the same ethnic group because of the territorial divisions and power sharing that consociationalism introduces to defuse conflict. This even defies the calls of the international community to transcend the ethnic voting, as citizens often have little voting choice. A particular ethnic group votes largely as a bloc supporting either nationalist or non-nationalist parties.

Because the choice of the members of an ethnic group depends on the stakes it has in the system, the principal hypothesis must be disaggregated. Firstly, The Bosniaks, holding the majority in the Federation of BiH, have much at stake with the Dayton Accords. The failure of implementing Dayton could cost the Bosniaks the state of BiH and their control of the state.

The resulting expectation is therefore the following:

[H.sub.1A]: Cantons with higher Bosniak populations are more closely associated with the non-nationalist and non-exclusionary party representation.

Relatedly, [H.sub.1B]: Cantons with higher Croat and Serb population are less likely to elect non-nationalist and non-exclusionary deputies.

This is because consociationalism prevents Croats and Serbs from seeking territorial unity with the countries of Croatia and Serbia, respectively, and provides Bosniaks with the crucial role in BiH. (8) In summation, ethnicity explains party loyalty according to the level of power an ethnic group holds in a territorial entity.

Alternative Theories

Scholars have presented alternative ethnic-based explanations (cf. H2-3). A first competing hypothesis to H1A-B is population change within an ethnic group. BiH experienced significant population shifts since 1995. Internal displacement caused by political or economic reasons may lead to the expression of grievance but impede within-group collective action (Gellner and Breuilly 2008). For example, the immigrants may have helped the Bosniak nationalist SDA to garner a firm nationalist vote in Sarajevo and other cities which counterbalanced the more cosmopolitan natives. (9)

[H.sub.2]: The more immigrants of an ethnic group that move in a region, the less likely the area votes non-nationalist.

Another ethnicity-based alternative explanation is ethnic...