Cross-border clientelism, commitment, and the protraction of irredentist conflicts

Abstract

Irredentist disputes have produced distinct political ethnoterritories under the de jure sovereignty of recognised parent states, but the de facto political authority of external national homelands. This study problematises the relationship between national homeland and claimed ethnoterritory as a nested game in which, in addition to bargaining with each other, they face internal competition, outbidding, and changing costs of conflict, ultimately reducing commitment to external-facing bargains. This study contends that homelands pursuing irredentist conflict can reduce uncertainty and increase commitment from ethnoterritories by building hegemonic cross-border clientelist pyramids that link ethnoterritorial publics’ and elites’ political survival and livelihoods to supporting homelands’ preferences. Further, these structures marginalise alternative elites who may seek to contravene preferences by escalating conflict and increasing costs on homelands or bargaining across ethnic cleavages. Case studies of protracted conflicts in Cyprus, Kosovo, and Croatia support this argument and further find that public-sector distribution linked to the homeland is most effective in reducing competition and uncertainty, thereby increasing long-term commitment to preferences.

Protracted irredentist conflicts have proven the most difficult to resolve owing to shared cross-border identities, complex patronage networks, and social solidarity rooted in conflict.¹ The complexity of irredentist conflicts is found in the incongruence of states’ administrative borders and the imagined boundaries of national communities, shaping cross-border identities and challenging state sovereignty. Irredentist conflicts and disputes centre on the nationalist drive to reunite a national community or culturally significant ethnoterritory – a terra irredenta – that is ‘trapped’ beyond the de jure borders of a national homeland. Though once a predominant mode of state-building in Europe, the growth of international law, human rights protections, and territorial integrity norms after the Second World War saw a decline in the permissibility of irredentism and rising costs imposed upon states pursuing it.² This is not to say irredentist ambitions disappeared, as national groups persisted outside the legally defined borders of their homelands who in turn claimed a right to protect their co-nationals. Accordingly, pursuit of irredentist conflicts, or even disputes short of armed confrontations, in this normative environment resulted in the confounding situation of ethnoterritories under the de jure sovereignty of one state (‘parent state’), but the de facto political authority of another. This study addresses the dyadic relationship between homelands initiating irredentist disputes and the ethnoterritories or terras irredenta that are the subjects of those disputes.

Extant studies of cross-border and proxy conflict tend to equate shared ethnicity with congruent preferences, even drawing on ethnic affinity as a heuristic for shared preferences.³ Conversely, work on internal dynamics of irredentism and separatism highlights the role of intra-communal competition over resources and public support as driving and prolonging conflict, demonstrating assumptions of stable and shared preferences to be insufficient.⁴ This study problematises the relationship between an ethnic homeland and a terra irredenta as a nested game of simultaneous and constant political competition. Decision-making elites in both arenas suffer from a commitment problem in their dyadic relations – that they are uncertain of the other’s commitment to preferences due to internal political competition. Congruent preferences, and commitment to them, are not inherent, but the outcome of a homeland constraining political competition and thereby political autonomy within a disputed ethnoterritory. How then do national homelands gain this compliance via stable commitments in the context of irredentist conflicts?

This study contends that homelands can efficiently attain stable commitments to discrete dispute-related preferences from terras irredenta through cross-border clientelism. Political clientelism is commonly understood as a non-programmatic form of politics in which elite patrons exchange inducements for the political fealty of clients. While this is associated with such individual practices as vote-buying, collectively clientelism can stabilise political relationships, reduce competition within communities or even parties, and constrain the autonomy of political elites to challenge a leader.⁵ In his analysis of post-transition political order in Eurasia, Henry Hale attributes political stability to the creation of hegemonic clientelist pyramids that define who is in the political order, and therefore the recipients of benefits, by virtue of acquiescence to the policies and preferences of the leader at the pyramid’s peak, and who is excluded. In application to irredentist disputes, the initiating homeland’s government functions as the patron at the top of the pyramid, with decision-making autonomy and control over the flow of resources down the pyramid to intermediaries, ethnic elites in the disputed ethnoterritory, and co-national publics. Receipt of inducements is conditional on upward support for, or at least non-opposition to, the homeland’s preferences at risk of exclusion from the pyramid. The outcome is twofold. One is reduced political competition in the ethnoterritory that marginalises elites and constituencies with divergent preferences from the homeland. The other is the protraction of conflict in a state of indeterminacy, whereby resolution is contingent upon support from the homeland and undermined by prolonged links between ethnoterritorial political factions capable of governance and an external homeland.

In explaining this puzzle of simultaneous intra-group competition and commitment to external preferences, and acknowledging the complexity of protracted conflicts, hegemonic cross-border clientelist pyramids constitute a causal mechanism that intervenes to produce the observed outcome of congruent preferences between homeland and ethnoterritory.⁶ Cross-border co-nationals and a claimed ethnoterritory do not necessarily result in protracted irredentist disputes in which the claimed ethnoterritory commits to the preferences of a homeland. Rather such disputes are observed where this mechanism is present. In practical terms, clientelism is a strategic solution for the homeland to ensure commitment from the ethnoterritory, thus granting it greater leeway to pursue its preferences for the dispute with less uncertainty due to competition at the ethnoterritorial level.

This study is certainly not the first to problematise and interrogate dyadic relations between patron states and client groups. Studies of topics such as colonisation or regime change centre the asymmetric relations between powerful states and local client elites. Studies of cross-border party networks, such as Hungarian parties in Romania and Slovakia or Hindutva parties in Nepal, similarly highlight how patron states may use ethnic affinity to seek favourable policies in parent states. The cross-border clientelism explanation differs in two areas. First, it is not a strictly elite-level bargain between a government and client parties or strongmen, but a simultaneous two-stage bargain between elites in both arenas, and ethnoterritorial elites and their publics. Second is the ability of homelands to monitor and constrain the choices of ethnoterritorial elites, both limiting meaningful political choice and allowing them to change preferences and expect compliance from the ethnoterritory. This entails lower autonomy of ethnoterritorial elites to make decisions and illiberal or anti-pluralist tendencies at the local level. While this differs from coercive and elite-led practices of empire-building, it pertains to post–Second World War disputes in an international context non-permissive to imperial and irredentist expansion.⁷ As such, hegemonic cross-border clientelist pyramid-building is an alternative form of claim-making that results in the persistence of disputed ethnoterritories under the de jure sovereignty of one state and the de facto political decision-making authority of another without the formal expansion of state borders. Ultimately recognition of cross-border clientelism in irredentist conflict not only contributes to understandings of modern ethnonationalist conflict, but also contributes practically to understandings and strategies of conflict resolution.

In the following section I outline a theory of cross-border clientelism derived from the problems of commitment in a nested game and building on Brubaker’s concept of a ‘triadic nexus’ in interstate ethnic politics.⁸ After a brief discussion of case selection and research design, I trace three cases of clientelism in irredentist disputes in Cyprus, Kosovo, and Croatia. I conclude by comparing case findings and considering broader implications for the study and practice of peacemaking.

Clientelism and commitment problems in a nested irredentist game

Irredentism centres on the claim that a co-national community is trapped beyond the legally recognised borders of a state, which seeks to rectify this incongruence through territorial expansion to encompass co-national-inhabited territory.⁹ Rogers Brubaker’s conceptualisation of nationalist politics in post-communist Europe models this as a ‘triadic nexus’ between a sovereign parent state, an ethnic minority within its borders, and an external national ‘homeland’ with influence over the minority’s politics (Figure 1).¹⁰ Departing from assumptions of congruent preferences between external homelands and minority communities in terras irredenta, their relationship in their dyad of the triadic nexus reflects a ‘nested game’. Following Tsebelis, nested games entail political actors bargaining simultaneously in external- and distinct internal-facing arenas.¹¹ Decisions taken in one arena are affected by evaluation of costs and outcomes in the other. Optimal decisions are thus constrained by the set of arenas and their conditions in which decision makers simultaneously bargain.

Figure 1. Triadic Nexus.

The commitment problem inherent in nested games, particularly in international politics, is that simultaneous internal-facing competition weakens commitments in external-facing arenas. In Putnam’s conceptualisation of ‘two-level games’ in negotiations, a state’s negotiator is constrained in their commitments to other states by coalitions in the domestic political arena, and failure to satisfy domestic demands results in the failure of negotiations or removal of the negotiator. While a more domestically constrained negotiator may have more leverage, their commitment is less credible to negotiating partners.¹² Domestically, such internal-facing features as leadership turnover, the size of coalition governments, ability to logroll policies, unity of the opposition, and key social groups, all constrain commitments in external-facing bargains.¹³ Similarly, consociationalism theorists contend that higher levels of political engagement and competition within identity groups limit their ability to commit to bargains with other groups.¹⁴

In the irredentist context, the homeland and ethnoterritory bargain with each other over the conduct and preferences for a conflict or dispute, while each has its own distinct internal-facing arena.¹⁵ In the homeland’s internal-facing arena, preferences and commitments are affected by domestic competition and external constraints imposed upon states for pursuing irredentist claims. Instrumentalist studies of irredentism find governments pursue irredentist claims when bidding for support from constituents or key players with cross-border kin or who covet symbolic territory.¹⁶ Similarly, where superordinate groups are near parity with others, or states have strict majoritarian or military regimes, governments may pursue irredentism to solidify support from key factions or reward allies.¹⁷ Irredentist claims are not inherent in states with cross-border kin, nor are they stable or durable, but rather shaped by domestic competition and institutions as a strategy to build and maintain support. Support for irredentism, though, may be curbed or affected by threats of sanctions or proffered rewards to moderate certain positions.¹⁸

On the ethnoterritory’s side of the dyad, commitment is constrained by unique local formal, informal, and social institutions that develop in the circumstances of ethnic segmentation or conflict with the parent state. Crisis, conflict, and breakdown of state institutions can lead ethnic groups to turn inward for protection and services,¹⁹ empowering certain elites able to provide sought-after public goods such as food, medicine, and physical protection. Many subsequently have the incentive to preserve the status quo.²⁰ Numerous studies demonstrate variation in rebel, criminal, and non-state governance in such settings, highlighting the importance of reciprocal relations between publics and elites needed to maintain their positions.²¹ The result is a multiplex of actors and the emergence of multiple decentralised clientelist pyramids headed by elites able to supply sought-after goods, including protection during conflict.²² These reciprocal relationships are neither static nor guaranteed. Ethnic segmentation may lead elites to outbid one another for in-group support with exclusionary positions towards out-groups.²³ Alternatively, group members may shift support to elites best able to provide material rewards or patronage.²⁴ While ethnic segments remain stable in the short term and serve as boundaries for distribution, leadership within segments is subject to competition.²⁵

Multiplicity of actors and incentives coupled with political competition (see Figure 2) in internal-facing arenas limits the credibility of commitments in the external-facing arena between homeland and ethnoterritory. Commitment to pursuing one’s preferences from the other may last only as long as a leader or faction’s tenure in power, and is affected by changing costs of conflict. Long-term commitment to preferences, including preferences that change, is more credible externally for sides with less competitive internal-facing arenas.

Figure 2. Nested Game in the Ethnoterritory.

Competing explanations: Commitment by force or co-option?

Before outlining an explanation of cross-border clientelist pyramid-building, I consider two competing explanations of inducing commitment to an external state’s preferences and the problems they present. These explanations are based upon observations and analyses of military occupation, colonisation, and regime change – practices of state expansion in effect ruled out for most states in the post-1945 system. One explanation is the administration of claimed territory and its population by coercive occupation. The external or claim-initiating state rules by overwhelming coercive force capable of repressing alternative elite-level opposition and public dissent.²⁶ This strategy hinges upon a state’s capacity to deploy repressive force superior to potential local resistance.²⁷ Coercive rule and exclusion of sections of the population and elite cadres from governance, however, increase nationalist support for rebellion and require increasing repressive capacity and more active repression over time.²⁸

The other explanation is the co-option of local elites to administer claimed territory on the claiming state’s behalf. Rather than repressive force, this strategy hinges on availability and political organisation of local elite cadres. This is argued to decrease incentives and opportunities for rebellion or opposition and is most effective when ex ante dominant factions are recruited, meaning less viable alternative elites to challenge them.²⁹

Elite co-option strategies suffer from two problems that limit elites’ commitment to external-facing preferences. First, elites most viable for co-option due to specialised knowledge or social and political standing, are also those most likely to have private information and preferences, and to deviate from the external state’s preferences to pursue their own interests.³⁰ For example, elites suited to secure borders and control peripheral regions are also the most able to escalate conflicts for personal gain, passing costs to state governments which must adapt to a new status quo.³¹ Second, recruited elites are limited in their ability to commit to externally derived preferences by their local selectorates. If the external state’s and local selectorate’s preferences differ, co-opted elites face the prospect of ousting and instability, including loss of foreign patronage, replacement by alternative elites, or armed rebellion from the population.³²

Clientelist pyramids and political order

The explanation posited in this study does not argue against the existence of violent coercion or co-option in disputed ethnoterritories. Both are features of hegemonic clientelist pyramid-building.³³ Rather these strategies alone do not resolve the problem of commitment to external preferences, especially when external preferences conflict with those held by local elites and publics. Construction of hegemonic cross-border clientelist pyramids resolves the issues of rebellion and elite autonomy or ‘agency slack’ by conditioning political viability on support for preferences, thereby co-opting or marginalising potential challengers, and by reducing the incentives for local ethnoterritorial publics to pursue autonomous preferences politically.

Comparative political studies understand clientelism as non-programmatic conditional distribution that exchanges inducements from patrons for political support or ‘fealty’.³⁴ Clientelism is characteristically dyadic, conditional, hierarchical, and non-programmatic, intended to induce fealty or deference within a client population.³⁵ Clientelism is not characterised by specific inducements – which could include payments, employment, promotion, favour with the bureaucracy, or a host of other options – but rather the targeting of specific constituencies either for political support or to encourage abstention from participation.³⁶ By exchanging inducements from higher tiers to lower, political elites can reduce uncertainty.

In his seminal work on clientelism in Southeast Asia, James Scott conceptualises a ‘clientelist pyramid’ linking a single patron at the peak to a broad clientele at the base by distributing inducements downward through tiers of intermediaries or brokers.³⁷ The downward distribution of inducements and the upward flow of political fealty results in an exclusionary system in which only the top of the pyramid has decision-making autonomy.³⁸ The outcome is a stable political order of intermediaries and clients deferential to policy choices made at the highest tier. Intermediaries recruited into the middle tiers of clientelist pyramids may include local community notables with existing networks that can be subsumed or politically popular actors, who attract support and can monitor compliance.³⁹

Firms and employers also serve as key brokers or intermediaries in a clientelist pyramid. Public-sector firms, civil services, or firms with state contracts are particularly prone to co-option into pyramids given existing flows of resources, budgets, and contracts, which can be increased or withheld to induce political fealty.⁴⁰ Continuous exchanges via publicly linked firms and the possibility of withholding rewards at any point, not just during election periods, can reduce monitoring problems inherent in clientelism – that the client receiving a reward will reciprocate with fealty.⁴¹

Pyramidical clientelism as described by Scott and others entails hierarchical exchange between horizontal segments, conditional on the lower tier reciprocating. Each segment provides fealty in return for an inducement from the segment above, such as a party leader providing resources to regional brokers, who provides resources to neighbourhood organisers, who in turn mobilise voters with resources. In both democracies and autocracies clientelist exchange is commonly associated with election-time vote-buying, but can extend to other forms of fealty.⁴² Inducements flowing down the pyramid via tiers of brokers to broader clients include rewards ranging from cash payments to favours with bureaucracy, or club goods such as development projects or ‘pork’ targeting a collective group.⁴³ Alternatively negative inducements may flow downward, such as threats, harassment, or targeted investigations, or the threat of being cut off from existing benefits – a particularly salient mechanism where clientelist linkages are already established that can deter reneging from existing commitments.⁴⁴ In return, political fealty flowing back up the pyramid includes not only votes, but turnout of undecided voters, abstention that depresses turnout for opposition, or rewards for loyalty that maintains key constituencies.⁴⁵ As such, non-opposition, either as positive affirmation of policy or refraining from supporting alternative elites, is a primary outcome.

In ethnically segmented settings characteristic of irredentist disputes, publics may be induced to support co-ethnics and deterred from inter-segment voting or service-seeking by in-group distribution of rewards, dense communal networks that increase information, and the threat of intra-communal punishment for crossing ethnic boundaries.⁴⁶ While political support may be bounded by ‘psychic rewards’ of supporting co-ethnics, the promise of patronage or material benefits from differing co-ethnic elites can shift to political support within segments to elites best able to deliver benefits (see Figure 2).⁴⁷ Pyramidical clientelism within segments reduces instability and the viability of alternative elites who may seek to escalate a conflict by outbidding or to moderate and resolve a conflict.⁴⁸ A secondary effect of clientelism is to build support for a patron by fracturing potentially viable opposition.⁴⁹ Alternative elites outside of the pyramidical structure lack the same resources for mobilising support and may lose supporters who receive superior rewards from those within the pyramid, or their supporters may be less incentivised to turn out than in-pyramid rivals, depressing their support. Hegemonic clientelist structures deprive the public and intermediaries of meaningful exit options from clientelist relationships, as support for alternative, programmatic elites is tantamount to material loss or forfeiting political viability.⁵⁰ Consequently, the political autonomy of both the public and intermediaries is forfeited for political survival or material well-being in the pyramidical structure.

Hegemonic cross-border clientelist pyramids and irredentist conflict

Adaptation of pyramidical clientelism from comparative intra-state politics to the interstate politics of irredentism requires clarification. First, as illustrated in Figure 3, the top of the clientelist pyramid is the government of the homeland, functioning as a patron, from which inducements flow downward across recognised borders. Studies of unrecognised or ‘de facto’ states note the asymmetrical flow of resources from ‘patron states’ upon which unrecognised states depend, including resources with political conditionality attached.⁵¹ It is further recognised that separatist elites may bid for material and political support from patron governments as an ‘audience’ of their actions.⁵²

Figure 3. Cross-Border Pyramidical Clientelism.

Second is consideration of the co-option of intermediaries as brokers in the pyramid with differing status – such as local political leaders, militia leaders, religious figures, or bureaucrats. Such figures may attain in-group status during conflict, and therefore face the choice to retain their autonomy or maintain political survival by entering the pyramid and subjugating themselves to an external patron’s preferences.⁵³ Last is the question of the fealty that flows upward in the pyramid in return for inducements. Rather than votes in the external state’s elections, cross-border inducements produce two types of fealty: (1) publics agree to support local intermediaries as political elites, and (2) local elites, legitimised by public support, agree not to oppose the patron government’s discrete preferences. The reduction of competition, particularly outbidding to secure support from the ethnoterritorial public, resolves both the two-principal problem highlighted by Downes, and the more general commitment problem noted by Putnam, that external commitment is limited by internal competition.

The primary outcome of hegemonic cross-border clientelist pyramids is stable commitment to the homeland’s preferences in the conflict, meaning reduced uncertainty on the homeland side of the dyad. The fracturing of political orders during conflict creates multiple decentralised clientelist pyramids headed by rebel or militia leaders, warlords, or military officers.⁵⁴ Elite-level competition over resources, funding, and local shares of local public support fuels outbidding and instability.⁵⁵ Stable commitment requires aggregating decentralised pyramids in a single order that isolates non-committed elites and makes political or armed challenges unviable and tantamount to exclusion from a beneficial order.⁵⁶ Aggregation of ethnic elites in the cross-border pyramid reduces opportunities to escalate conflict or cooperate across ethnic boundaries with the parent state.

Ethnoterritorial concentration further resolves the monitoring problem of clientelism – that clients will reciprocate inducements with political fealty. Dense intra-segmental networks increase information and persistent fear of monitoring reduces individual client’s beliefs they can obfuscate crossing ethnic boundaries, non-mobilisation or non-abstention, or support for alternative elites. Similarly social commitments and threats of losing access to exclusive in-group benefits deters defection.⁵⁷ Importantly, exchange based on continuous goods provision such as jobs or welfare, rather than one-off election-time vote-buying, reduces the salience of monitoring problems as clients and brokers face constant threats of losing existing benefits and their livelihoods.⁵⁸

A secondary outcome is protraction of conflict or incomplete resolution, leaving disputed ethnoterritories beyond the parent state’s de facto political authority. While clientelist linkages between homelands and ethnoterritories stabilise commitment in that dyad, they destabilise commitment in the other two dyads of the triadic nexus (Figure 1) – between the homeland and the parent state, and the parent state and ethnoterritory. First, they reduce the viability of alternative elites who may be willing to bargain with non-co-ethnics in order to reach a resolution, especially one that improves public welfare.⁵⁹ Second, they warp the context of conflict resolution. Conditions required for a conflict’s ‘ripeness’ for resolution such as mutual interdependence on opposing sides or untenable costs are reduced by clientelist dependence on an external patron.⁶⁰ Lastly, resolution is dependent upon the interests and preference of the patron government. While clientelism’s non-programmatic nature can allow the initiating state to revise preferences and attain commitment to a settlement, the conditions of maintaining such commitment paradoxically undermine the parent state’s de facto sovereignty.

Cases: Evaluating clientelist pyramids in irredentist conflicts

To further develop the argument, and inductively evaluate variation in the mechanism of cross-border clientelism, I turn to three cases of irredentist conflict – Northern Cyprus, Kosovo, and the Republic of Serbian Krajina (RSK). While there are numerous other examples of influential cross-border linkages between states and ethnic groups, such as Hungary’s influence on its neighbours, India’s influence in Nepal, Turkey’s influence in northern Syria, or Morocco’s control over the western Sahara, the three selected cases are suitable for comparison for three reasons. First, as irredentist conflicts, these represent ‘most likely’ cases to observe congruent preferences based on assumptions of interest congruity between cross-border ethnic kin, yet commitments between homeland governments and ethnoterritories differ between and within cases. Second, in-case variation allows for the mechanism to be evaluated more critically. Third, all three cases include internationally led conflict-resolution processes. This allows for the influence of internationally imposed costs and changes to preferences in homeland governments’ coalitions to be evaluated, further contributing an understanding of cross-border clientelism to conflict protraction and resolution.

To evaluate how the alignment of preferences is reached in dyadic bargains, I combine process tracing with structured case comparison. Following prescriptions for mechanistic process tracing, the hypothesised mechanism is specified ex ante, and evidence of its presence, absence, and variation in producing an observed outcome is assessed in each case.⁶¹ The specified mechanism is hegemonic cross-border clientelist pyramids (1) providing access to excludable goods conditional on political fealty, and (2) excluding alternative elites and opposition supporters who oppose discrete preferences.

The value of cross-case comparison comes in assessing the same questions regarding mechanisms and variation across settings, allowing for generalisable implications particularly in assessing how outcomes occur across cases.⁶² Case evidence is drawn from a variety of sources, including secondary scholarship on specific cases, governmental and intergovernmental organisation reports, and local and international news media from the period of evaluation. Specific evidence is cited throughout the narratives, but in establishing holistic understandings of the cases, and, importantly, the context in which key evidence is considered, I evaluated a larger corpus of more than 11,000 primary source documents and reports on the cases.

Northern Cyprus

The Cyprus conflict has persisted since 1963, characterised by intervention in the Turkish Cypriot community by Turkey, continuous mediation led by the United Nations (UN), and the existence of distinct Turkish Cypriot ethnoterritory beyond the authority of the recognised Republic of Cyprus. Turkey’s role as an ethnic homeland in Northern Cyprus has included co-option of elites, provision of coercive means, and construction of a cross-border clientelist pyramid, pursuant to its central objective of denying Cyprus’ unification with Greece. With Cyprus’ independence looming in the 1950s, Ankara feared the annexation of Cyprus with its Greek Cypriot majority to Greece – the Greek irredentist objective of enosis – which it regarded as a unilateral revision to the territorial status quo established between Athens and Ankara in 1923. In turn it feared enosis would both lead to persecution of Cyprus’ Turkish Cypriot minority and give Greece a strategic position 50 miles from Turkey’s southern coast. It instead favoured taksim, ethnic partition of the island, and in trilateral negotiations between Greece, Turkey, and Britain in 1959, it was agreed that Cyprus would become independent with a power-sharing constitution underwritten by the Treaties of Establishment and Guarantee, which gave all three the right to intervene if the constitution was violated.⁶³

Inter-communal violence in 1963–64 saw the Turkish Cypriots withdraw from government and ethnically mixed locales to distinct ethnic enclaves disparately scattered around Cyprus. The elites Ankara co-opted were headed by Rauf Denktaş, former head of the Turkish Cypriot Communal Chamber, who chaired the Provisional Administration, and included other Turkish Cypriot officials who quit Cyprus’ government in 1963, but notably not Cyprus’ former vice president Fazıl Küçük, whom Ankara regarded as too autonomous and willing to cross ethnic boundaries to bargain with the Greek Cypriots.⁶⁴ Denktaş, who represented the Turkish Cypriots in UN-led mediation beginning in 1964, echoed Ankara’s preferences seeking a guarantee against enosis, reaffirmation of the Treaty of Guarantee, and a politically distinct Turkish Cypriot federal entity - de facto taksim.

Co-option into the Provisional Administration was conditional on supporting these preferences and particularly on enforcing ethnic boundaries. Village headmen, mukhtars, were required to pledge loyalty to the Provisional Administration and reject the Republic of Cyprus’ authority, receiving in return aid, municipal projects funded by Ankara such as housing for displaced persons, and protection by the Turkish Resistance Movement (TMT), a paramilitary force trained and equipped by Ankara.⁶⁵ Opposition, such as the Republican Turkish Party (CTP), formed in 1970 and opposed to continued ethnic division and isolation, was excluded from the Provisional Administration and distributional structures.⁶⁶ The TMT had authority to enforce Provisional Administration laws, including regulating movement, and banning resettlement outside of enclaves or sale of property to non-Turks. Despite poor living conditions and reliance on Turkish and UN aid, resettlement outside of enclaves was low and Denktaş and his allies retained control of the Provisional Administration without viable challengers, including supplanting Küçük in his formal position as vice president in 1973.⁶⁷

The Turkish invasions and partition of Cyprus in July–August 1974 transformed the Turkish Cypriot ethnoterritory from disparate enclaves to the territorially contiguous Turkish Federated State of Cyprus (TFSC) – the northern 38 per cent of the island. While the stationing of a permanent Turkish garrison reduced potential threats, it also reduced the importance of the TMT as a source of legitimacy for Turkish Cypriot elites, particularly the Provisional Administration.⁶⁸ Ankara regarded the 1974 invasion as having achieved the objectives of pre-empting enosis and establishing a distinct Turkish Cypriot entity. Its new preference became maintaining the post-1974 entity and legitimising it in a subsequent settlement – including a distinct Turkish Cypriot entity and its garrison on Cyprus.⁶⁹

Attaining these preferences required support from Turkish Cypriot elites, who would both represent these interests in UN-led mediation and ensure them in the TFSC administration. The Provisional Administration reconstituted itself as the National Unity Party (UBP), which controlled the Turkish Cypriot administration until 1994, and Rauf Denktaş remained ‘president’ until 2005, representing the Turkish Cypriots in mediation. The UBP’s hold on power along with other pro-taksim groups was ensured by distribution of inducements Ankara provided in exchange for fealty to and representation of its preferences. After 1974, this included redistribution of property seized by the Turkish military in the 1974 invasions and transferred to the TFSC and then to key intermediaries and constituencies in exchange for political support. This included housing, businesses, large enterprises, and farms. After this initial property redistribution, distribution of inducements for support was moved to a large public sector controlled by the UBP-led administration, which included publicly funded farms, industries, schools, and a large civil service. Ankara funded 55–85 per cent of the TFSC budget annually, more than 50 per cent of which was devoted to public-sector salaries.⁷⁰ By 1982, unemployment dropped from 50 per cent to less than 3 per cent accounted for in the public sector, which employed an estimated 55,000 on a payroll funded by Ankara.⁷¹ Public-sector unions became key political networks that represented between one third and one quarter of the TFSC population, dependent on the UBP for continuous benefits – their livelihoods.⁷²

Despite its dominance, the UBP faced opposition within the TFSC and later the Turkish Republic of Northern Cyprus (TRNC) as it became in 1983, particularly from the CTP and the Communal Liberation Party (TKP). Though neither supported reverting to the pre-1974 status quo, both opposed over-reliance on Ankara and intransigence in mediation, which they regarded as damaging the Turkish Cypriots’ reputation and economic prospects. The UBP used distribution of inducements at key points to keep them out of power, thereby excluding them from the pyramid for opposition to Ankara’s preferences. Before elections in 1981, after losing its outright majority in the TFSC assembly, the UBP government created 4,000 new public-sector posts and subsequently won the elections. In 1990, after losing its coalition partners and governing as a minority supported by independents, the UBP distributed extra paychecks funded by Ankara to the public sector before elections and won an outright majority.⁷³ Despite growing support for moderation in talks and a solution to continued partition, Denktaş had the leeway and stable support of the UBP to continue to pursue Ankara’s preferences for the status quo, rejecting compromise solutions in 1977–79, 1988, and 1992.⁷⁴

The internal variation in the Northern Cyprus case further indicates the importance of public-sector clientelism as a mechanism in reducing internal-facing competition and increasing the credibility of commitments to the homeland’s preferences. Following the end of Kenan Evren’s military rule in 1983, Turkey enacted neoliberal economic reforms which were incorporated into its Cyprus policy in 1986. Aid to the TRNC became conditioned on cutting public-sector spending and employment and on privatisation, dealing a blow to the established pyramidical structure.⁷⁵ Two notable observations followed. One was increased support for moderate pro-reunification parties that opposed over-reliance on Ankara, namely the CTP, which entered the governing coalition for the first time in 1994, also the first year the UBP was in opposition (see Table 1). While in government, the CTP initiated inter-communal dialogue with Greek Cypriot parties and preliminary discussions on joining the European Union (EU), something Ankara adamantly opposed.⁷⁶ The other observation was an increase in coercive repressive actions targeting alternative elites opposed to Ankara’s positions, including bomb attacks on opposition, attacks on critical media, and the banning of inter-communal meetings after 1999.⁷⁷ Hence, weakening of the hegemonic pyramid due to austerity fostered increased political competition in the Turkish Cypriot internal-facing arena and less certainty of commitment to Ankara’s position, necessitating repressive means to target alternative elites.

Table 1. Turkish Cypriot governing coalitions.

*The TFSC parliament had 40 seats (21 needed for a governing majority); the TRNC has 50 seats (26 for a governing majority).

The weakening of the clientelist pyramid from Ankara after 1986 meant less certainty of commitment from Turkish Cypriot elites, subject to internal-facing competition. This was particularly evident in two instances of Ankara changing its preferences for the conflict, which were subsequently opposed by the Turkish Cypriot elites in power. The first was in 2002 when the new Justice and Development (AKP) government in Ankara came to power and supported the UN-proposed Annan Plan for the reunification of Cyprus, regarding the status quo pro-taksim policies as ‘decades of mistakes’ that alienated neighbours and allies. Significantly the EU appended support for the Annan Plan to Turkey’s membership application, which the AKP favoured, unlike its predecessors.⁷⁸ It could not, however, leverage support from Denktaş in negotiations or the UBP in the TRNC assembly whose support was needed for ratification or a referendum on the Annan Plan.⁷⁹ Turkey supported the CTP in ousting the UBP and forming a coalition in January 2004, with the objective of agreeing to the Annan Plan.⁸⁰ However, the delay precluded the TRNC from signing Cyprus’ EU accession protocol in April 2003, and in 2004, despite the Turkish Cypriots approving the Annan Plan by a margin of 65 to 35, the Greek Cypriots rejected it and joined the EU unilaterally.

The second example of Ankara being unable to ensure compliance with its preferences occurred after the failed Crans-Montana negotiations in 2017, following which Ankara withdrew its support for reunification and favoured a formalised two-state solution. However, it could not ensure commitment to this position from the TRNC, which was led by a pro-unification ‘president’, Mustafa Akıncı, who resumed reunification talks with the Greek Cypriots without Ankara’s support.⁸¹ Ankara’s counter support for the UBP and its leader, Ersin Tatar, who supported a two-state solution, was undermined in 2018 by internal-facing competition.⁸² Though the UBP had the largest plurality, four parties led by the CTP formed a coalition to keep it from power.

Ankara reconstructed a cross-border clientelist pyramid in the TRNC that co-opted the opposition and excluded the governing elite, hoping to marginalise them. In 2019 Ankara withheld aid and funding from the CTP-led government, making it unable to distribute services and salaries while opposing a two-state solution, thus undermining its support.⁸³ Ahead of the 2020 ‘presidential’ elections, Turkish officials and UBP activists threatened to withhold funding in urban centres if Akıncı was re-elected, where he and the CTP drew their support. Additionally, Ankara transferred 117 million lira to TRNC banks, which was distributed officially as ‘Covid relief payments’ by activists hired by the Turkish Embassy and the UBP in exchange for photos of ballots showing votes for the UBP.⁸⁴ Despite high pre-election support for Akıncı and reunification,⁸⁵ this alternative pyramid, from which he and the CTP were excluded, both suppressed his support by coercing abstentions and mobilised UBP support via vote-buying. The result was a victory for Tatar, who committed to Ankara’s preference for a two-state solution.

Northern Cyprus illustrates hegemonic cross-border clientelist pyramid-building during the Provisional Administration and TFSC, but also indicates lower commitment once a pyramidical structure was weakened. An effect of the post-1983 neoliberal reforms was to weaken the stability of public-sector clientelism in the TRNC, reducing stability and commitment, evident in increasing pluralism and short-lived coalition governments with an average tenure of 18 months (see Table 1). This meant Ankara was unable to gain commitment to its revised preferences after 2002 and 2018 without a change in government in the TRNC. In 2019–20 it reconstituted a clientelist pyramid in the TRNC, but one that co-opted the opposition, marginalising the status quo leaders who were supported by a sizeable pro-reunification constituency.

The Kosovo Serbs

In the context of Kosovo, Serbia has historically attempted to maintain sovereignty over the territory while simultaneously denying the trappings of sovereignty to its Albanian majority population. This included heavy-handed repression and centralised rule under the administration of Aleksandar Ranković (1945–66), the leading Serb in early socialist Yugoslavia, and later mimicked under Slobodan Milošević’s regime following the revocation of Kosovo’s status as an ‘Autonomous Province’ in 1989. Though the Kosovar Albanians received increased self-rule in the interregnum, Serbian nationalists, notably the Serbian Orthodox Church, opposed this loss of sovereignty over what was regarded as historically and culturally Serbian territory and advocated a ‘return to Kosovo’.⁸⁶ The conclusion of the Kosovo War in June 1999, ending in intervention by the North Atlantic Treaty Organization (NATO), saw the administrative and security institutions of the Serbian state expelled from Kosovo. The Serb minority that remained after the state’s withdrawal faced indiscriminate violence and harassment from the Albanian majority and trusted neither the United Nations Interim Administration for Kosovo (UNMIK) nor its subordinate Albanian-dominated institutions in Prishtina, the Provisional Institutions of Self-Government (PISGs). Instead they withdrew to ethnically distinct and defensible enclaves outside of cities, near religious sites, or north of the River Ibar near the Serbian border. They remained closely aligned with Belgrade and only irregularly participated in the PISGs or Kosovo’s independent institutions (following its unilateral declaration of independence in 2008).⁸⁷ Not until 2013 did Serbs participate in Kosovo’s political system en masse, and even this was highly regulated and constrained by Belgrade.

Belgrade’s post-1999 preference, despite Milošević’s ousting in 2000, remained retention of sovereignty over Kosovo and conversely denial of its own sovereignty and statehood pursued by the Kosovar Albanian majority.⁸⁸ This entailed not only rejection of Kosovo’s independence claims and enshrining its status as a part of Serbia in Serbia’s 2006 constitution, but also opposition to UNMIK and the PISGs, which it regarded as precursors to independent state institutions. From 2003 to 2006, Serbia quit or delayed UN-led mediation over Kosovo’s status three times in protest of UNMIK transferring authority to the PISGs.⁸⁹ After ethnic riots targeting the Serb community in March 2004, Belgrade implemented a boycott policy discouraging political partition by Kosovo Serbs.⁹⁰ This undermined UNMIK’s position, which viewed integration of minorities, and particularly the Serbs, in governance as paramount to its mission.⁹¹ For Serbia, maintaining political authority over the Serbs denied Kosovo full sovereignty over its territory and population.

With its security institutions expelled in June 1999, Belgrade relied upon elite co-option and clientelism to enforce commitment to its preferences regarding participation and denial of sovereignty. The state’s expulsion in 1999 and the change of regime in 2000 left few Serb elite cadres.⁹² Two notable factions emerged in 1999. The Serb National Council of Kosovo and Metohija (SNV-KiM) was based in the central enclave of Gračanica/Graçanicë, which Serbs treated as a ‘capital-in-exile’ from Prishtina, and participated in dialogue with UNMIK, Kosovar Albanians, and later its members joined the PISGs (2001–4). The other, the Serb National Council of Kosovska Mitrovica (SNV-M), based in the northern half of the divided city of Mitrovica, rejected cooperation, declared its autonomy from UNMIK, and expelled members who joined the PISGs.⁹³

Leaders from both factions were co-opted into a pyramidical structure of distribution originating from Belgrade – the so-called ‘parallel structures’. These operated strictly in Serb enclaves, filling the role of a state, providing municipal administration, education, healthcare, public services, welfare distribution, and even policing and judicial services, all funded by relevant line ministries in Belgrade. Parallel healthcare and education were the largest employers in Serb enclaves, and the majority of the Serb population received welfare benefits distributed by parallel administrations, including pensions, family stipends, and unemployment compensation for those previously employed in state-owned enterprises that ceased operations after 1999.⁹⁴ This kept many Serbs’ livelihoods linked to parallel officials as intermediaries in the pyramid. Municipal officials and administrators in parallel services had discretion over distribution, making them prominent local elites linked to a broad co-ethnic public despite ostensibly bureaucratic posts, such as hospital or school administrators with discretion over hiring and budgets. Parallel employees in Kosovo received a 200 per cent salary compared to those employed in the same positions in Serbia proper.⁹⁵

Provision of parallel services began after 1999, but their use as a hegemonic clientelist pyramid was most evident after Belgrade’s 2004 boycott policy. Access to parallel benefits, including employment in parallel institutions, welfare benefits, and accrued pensions, was conditional on boycotting, meaning not voting in PISG elections, not taking UNMIK contracts, and not seeking public services from UNMIK or Prishtina.⁹⁶ Compliance with the boycott was monitored through tiers of local administrators, employers, and brokers, with knowledge of local communities and able to make decisions on withholding benefits.⁹⁷ By 2006, 70 per cent of Serb UNMIK employees quit under threat of losing parallel pay or benefits, while voters were deterred from participation with threats of monitoring by co-ethnics and benefits being withheld ahead of Kosovo’s elections.⁹⁸

While this depressed Serbs’ participation in Kosovo’s political system, it also kept Serb enclaves isolated from Kosovo’s infrastructure, including power, water, and telecommunications, meaning underdevelopment relative to comparable Albanian-majority municipalities. Despite this, moderate elites who favoured closer cooperation with UNMIK and Prishtina were marginalised by their exclusion from the clientelist pyramid and their support undermined by threats to supporters’ livelihoods. They could not effectively pursue moderate, albeit collectively beneficial positions that deviated from Belgrade’s preferences. For example, in 2004, the Return Coalition (KP), which had participated in the PISGs from 2001 to 2004, ran in PISG elections, contravening the boycott. However, they forfeited their mandates after the elections due to low turnout. The party subsequently split, with one faction sacrificing its autonomy for political viability within the parallel structures, and the other rejecting the parallel structures, but losing support and taking non-elected positions as bureaucrats in UNMIK or municipal institutions – ostensibly without political decision-making power.⁹⁹ Similarly, during UN-led status talks in 2006, Belgrade curbed the autonomy of local Serb elites, including punishing those who conferred with the UN envoy independently.¹⁰⁰

Similar to Northern Cyprus, two internal variations in the case demonstrate the importance of the cross-border clientelism mechanism in gaining commitment to preferences. One was the difference between larger Serb enclaves with strong parallel institutions and smaller enclaves in which the parallel structures had less influence. The latter allowed for the emergence of alternative elites, namely the Independent Liberal Party (SLS) and local ‘citizens’ initiatives’ (GIs). The constitution of six Serb-majority municipalities in Gračanica/Graçanicë, Klokot/Kllokot, Novo Brdo/Novobërda, Parteš/Partesh, Ranilug/Ranillug, and Štrpce/Shtërpcë as separate electoral districts in 2009–10 created an opportunity for alternative elites to challenge parallel structures and secure inducements from Prishtina. They ran in Kosovo’s 2009–10 municipal elections and won undisputed control of five municipalities. The newly elected mayor of Štrpce/Shtërpcë, where 2,000 Serbs registered for welfare benefits from Prishtina, forced the parallel institutions to close.¹⁰¹ Though alternative elites ran in Gračanica/Graçanicë, as the Serbs’ ‘capital-in-exile’, it had robust parallel institutions, including 1,000 municipal officials who alternative elites could not force out.¹⁰² The election of alternative elites in Kosovo’s system meant the creation of multiple alternative distributional pyramids, with elected mayors and councils able to distribute funding and civil service posts originating from Prishtina rather than Belgrade, meaning Belgrade lost control of these municipalities, which openly contravened the boycott and legitimised Kosovo’s sovereignty.¹⁰³

The second internal variation, as in Cyprus, was variation in the homeland’s governing coalition, which changed its preferences for the conflict. In 2008 the Democratic Party of Serbia (DSS), responsible for the boycott policy, was ousted by a coalition led by the more liberal Democratic Party (DS). Though it did not support recognising Kosovo’s independence, it ran on a platform of ‘Kosovo AND the EU’ and advocated pragmatic cooperation on Kosovo to advance Serbia’s EU membership bid. The DSS opposed EU accession, fearing that Serbia would be forced to sacrifice Kosovo.¹⁰⁴ The DS-led government cooperated with the EU on the deployment of the EU Rule of Law Mission in Kosovo and agreed to EU-led mediation with Kosovo in 2010. Problematically, the local Kosovo Serb elites remained aligned with the DSS and used their positions to undermine the DS.¹⁰⁵ This was most evident in 2011, when Serbs in northern Kosovo blockaded the border with Serbia to spoil a customs agreement, leading to months of unrest and the EU withholding Serbia’s candidacy for membership.¹⁰⁶ Non-commitment to Belgrade’s preferences incurred tangible costs for Belgrade. Importantly, Belgrade did not condition access to inducements on support for mediated agreements, and instead attempted to replace DSS-aligned elites with its own loyalists. However, this increased intra-communal competition and saw their ousting by the status quo elites and public.¹⁰⁷

When the newly formed Progressive Party (SNS) replaced the DS-led coalition in Serbia after the 2012 elections it continued the ‘Kosovo AND the EU’ policy and reaffirmed support for agreements reached in 2011–12 EU-led talks. In April 2013 the SNS-led government agreed to the Brussels Agreement with Prishtina under an ultimatum from the EU. The Brussels Agreement would exchange Serbia opening accession negotiations for Serb participation in Kosovo’s 2013 elections and the dismantling of parallel security and judicial structures. Problematically, this required support from the unstable Kosovo Serb political arena, characterised by multiple clientelist pyramids after 2009–10 and opposition to Belgrade’s authority after DS rule. The same elites who spoiled the 2011 customs agreement in northern Kosovo opposed the Brussels Agreement, rejected Belgrade’s authority to negotiate on Kosovo, and organised to boycott the 2013 elections.¹⁰⁸

The SNS-led coalition reconstituted a hegemonic cross-border clientelist pyramid within a single party, the Serb List (SL), to run in the 2013 elections and marginalise alternative elites seeking to outbid by boycotting or fully integrating in Prishtina’s institutions as during the 2009–10 elections. Parallel distribution continued to flow down the pyramid from Belgrade and the relevant Serbian line ministries, but the key intermediaries in the pyramid were replaced by SL officials, such as administrators of public services and municipalities. In August–September 2013, the SNS-led government in Belgrade dissolved the municipal governments in northern Kosovo, depriving status quo elites threatening an election boycott of their means of enforcing boycotts for the previous nine years. Three of four appointed interim administrators were elected mayors in 2013.¹⁰⁹

The SL’s success in 2013 was followed by monopolisation of Kosovo Serb politics by 2019 (Figure 4). Strategies previously employed to ensure commitment to the boycott policy and ethnic closure were refocused on maintaining support for the SL and marginalising alternative elites, namely threats of exclusion from public-sector employment or welfare benefits. Alternative elites, their families, and their supporters faced losing jobs, payments, or access to services for challenging the SL. Alternative elites faced a choice to forfeit political autonomy for inclusion in the SL or lose political viability, resulting in the co-option of moderate elites who contravened the boycott in 2009–10 and former DSS officials who had opposed the Brussels Agreement into the SL. The SL officials or factions that contravened Belgrade’s preferences could be expelled and politically marginalised.¹¹⁰ Notably this occurred in 2015 when the SL’s head blocked continued negotiations supported by Serbia and was removed from the party,¹¹¹ and in 2016 when the Socialist Movement faction was expelled from the SL for opposing Belgrade’s position on municipal decentralisation.¹¹²

Figure 4. Percentages of Seats Held in Serb Majority Municipalities.

The outcome was stable commitment to Belgrade’s preferences from the Kosovo Serbs after 2013. In the short term it allowed Belgrade to ensure commitment to its revised preferences of ending the boycott and opening Serbia’s EU accession talks. Longer term, it solidified Belgrade’s political influence in the Kosovo Serb community and undermined Kosovo’s sovereignty and political stability more broadly, protracting the conflict and inducing non-cooperative positions within the Serb community despite formal participation in Kosovo’s institutions. The SL engaged in repeated boycotts of Kosovo’s government in opposition to policies opposed by Belgrade, including policing in northern Kosovo in 2018–20, imposition of tariffs on Serbian goods in 2018–19, and municipal governance in 2023–24.

The Kosovo Serb community presents a case in which the homeland could not deploy coercive force and was faced with a lack of ex ante dominant elites to co-opt. Elite recruitment and clientelist pyramid-building were integrated processes whereby those with intermediary positions in the parallel structures attained local political status with low autonomy from Belgrade. Given Serbs’ reliance on parallel benefits or employment, the threat of exclusion induced fealty to Belgrade’s preferences. Both during the boycott period (post 2004) and under the SL (post 2013), in order to gain access to the clientelist pyramid, local elites were required to forfeit autonomy to Belgrade’s preferences, be they ethnic closure, support for the SL, or opposition to Prishtina’s policies, particularly those aimed at asserting sovereignty over northern Kosovo.

Croatia: Republic of Serbian Krajina

The Republic of Serbian Krajina (RSK), the Serb separatist entity in Croatia from 1990 to 1995, presents a negative case in which the homeland, Serbia, did not construct a hegemonic cross-border clientelist pyramid. The RSK did not persist as an irredentist dispute or an ethnoterritory beyond Croatian sovereignty, and by 1995 it collapsed from within, facilitating its reconquest by Croatia. Elite co-option resulted in intra-ethnic competition characterised by persistent intra-elite conflict and inability to commit to Belgrade’s preferences, especially after they were revised. In initiating separatist conflict in Croatia, Serbia recruited like-minded local notables with irredentist preferences for unification in ‘Greater Serbia’ and supplied the means of coercive violence, including Yugoslav Army (JNA) interventions in 1990–1, support for Serb militias, and 30,000 JNA troops transferred to the RSK’s command in 1992.¹¹³ By mid-1992, though, Belgrade lost influence over the RSK, and its administration fractured between elite factions with different preferences for the conflict that ranged from negotiation on autonomy within Croatia, annexation to Serbia, secession of different regions, and unification with Republika Srpska in Bosnia-Herzegovina (BiH).

Serbia’s initial preference for Serb-inhabited territory in Croatia was annexation or incorporation in ‘Greater Serbia’, which envisioned a unified state of Serb-inhabited regions. Milošević, his allies, and like-minded nationalists fomented counter-secession among Serbs in Croatia and BiH, promoting beliefs that Serbs would be politically marginalised or persecuted if the republics declared independence from Yugoslavia.¹¹⁴ In Croatia, Belgrade responded to looming independence by encouraging separatism and supporting the ‘Log Revolution’ by which Serb-inhabited territory cut itself off from Zagreb and declared its autonomy as SAO Krajina under a provisional government in Knin. This was aided by interventions from the Serb-dominated JNA, commanded from Belgrade, which disarmed Croatia in 1990 and then invaded in 1991 after Croatia declared independence. In 1991 the Croatian Serbs declared themselves independent as the Republic of Serbian Krajina (RSK) with the intention of joining Serbia.¹¹⁵

In pursuit of irridentism and counter-secession in Croatia, Belgrade co-opted local Serb elites into the Serb Democratic Party (SDS) in 1990. The SDS held only 5 of 365 seats in Croatia’s government but was dominant among the Serb population.¹¹⁶ Belgrade recruited Milan Babić, the SDS head of the Knin municipal government, to lead the party ahead of the 1990 Log Revolution for his shared irredentist preferences with Belgrade.¹¹⁷

Belgrade revised its official preference for annexation of the RSK in late 1991 under diplomatic and economic pressure from the EU, the UN, and the United States.¹¹⁸ Seeking to alleviate economic sanctions and avoid further ostracisation, Belgrade agreed to the UN-proposed Vance Plan in 1991, by which the JNA would withdraw from Croatia and UN peacekeepers (UNPROFOR) would deploy to supervise a ceasefire. Belgrade subsequently renounced its objective of annexation and publicly supported a negotiated settlement on the autonomy of the RSK in a Croatian state.¹¹⁹ This did not amount to Belgrade renouncing ethnic nationalism or irredentist ‘Greater Serbia’, but was a revision to a discrete preference for the RSK’s short-term political status.

Belgrade’s revised preference to support the Vance Plan was incongruent with Babić’s preferences, the reason underlying his recruitment to head the SDS. He opposed autonomy and deployment of UNPROFOR. The UN mediators regarded Babić as the primary obstacle to a settlement, while Belgrade regarded him as intransigent and increasing costs for Serbia, and publicly withdrew support for him in late 1991.¹²⁰ Belgrade sought to co-opt alternative elites in the SDS, namely Goran Hadžić in Vukovar and former Knin police chief Milan Martić. It orchestrated Babić’s ousting as RSK ‘president’ in February 1992 at a special session of the RSK assembly in Glina rather than Babić’s political stronghold of Knin.¹²¹

This triggered intra-communal competition and instability in the intra-RSK political arena. Ability to distribute inducements down a clientelist pyramid originating in Belgrade was constrained by international sanctions limiting Belgrade’s available resources,¹²² the outbreak of war in BiH which cut off supply routes from Serbia to the western RSK including Knin,¹²³ and neglected internal institutions in the RSK unable to effectively distribute inducement downwards through brokers. This latter constraint was particularly evident in the underdevelopment of a public sector due to the assumption of future annexation by Serbia, which meant exchanges between intermediaries and the broader public were not institutionalised.¹²⁴ Food, consumer goods, and petrol became scarce after war began in BiH.¹²⁵ The largest publicly owned enterprise in Knin operated at less than 5 per cent capacity after 1991.¹²⁶ By 1994 public employees, including hospital and school staff, and soldiers in the RSK military were striking over unpaid wages.¹²⁷ Unlike the TFSC and Kosovo Serb enclaves, there was a not a functional public sector to distribute inducements in exchange for political fealty to cross-border preferences.

Inability to consolidate a single hegemonic clientelist pyramid, including public-sector mechanisms for distribution, resulted in intra-elite conflict between elites with differing preferences for the outcome of the conflict, reducing commitment to Belgrade’s preferences. Babić, Hadžić, and Martić each controlled their own decentralised support networks with their own means of violence. Babić retained control of Knin and support from the RSK army, which refused orders from Martić, who retained support from the police and his own militia and manipulated elections in 1993 to keep Babić from power.¹²⁸ Hadžić, who controlled his own militias and dominated the black market economy in the eastern RSK, split with Knin in 1993, banned Babić and Martić from areas he controlled, and supported secession from the rest of the RSK.¹²⁹ While Martić was willing to open talks with Croatia on autonomy, congruent with Belgrade’s post-1991 preference, he was ousted in 1994 by Babić, who supported unification with the Republika Srpska in lieu of annexation by Serbia.¹³⁰ Low and unpaid wages, including in the RSK army, incentivised joining predatory paramilitaries and organised crime groups, both of which provided inducements in exchange for supporting individual elites – decentralised clientelist pyramids.¹³¹ Organised criminals smuggled and distributed needed goods at inflated prices, while paramilitaries seized and distributed loot and property from Croat and Serb civilians with impunity.¹³²

Despite elite co-option from the dominant faction and furnishing the means of coercion, Belgrade could not ensure commitment from the RSK after it revised its preferences in 1991. Inability to consolidate a hegemonic clientelist pyramid exchanging inducements for fealty between Belgrade and the RSK and, in turn, RSK elites and the public, resulted in multiple clientelist pyramids linking individual elites to specific paramilitaries or gangs. Intra-elite conflict between those heading decentralised pyramids meant multiple viable alternative elites, each with different preferences that diverged from Belgrade’s post-1991 preference supporting the Vance Plan. By 1995, when Croatia retook control of the RSK in Operations Flash and Storm, the RSK had ceased to function as a unitary political entity, and Serb fighters received conflicting orders to surrender, retreat, or stand and fight based on which elites they were loyal to.¹³³

Case comparison and discussion

While the RSK ceased to exist within four years and its territory was fully reintegrated under Croatian sovereignty by 1998, the Turkish Cypriot and Kosovo Serb ethnoterritories have persisted in an uncertain state beyond the authority of their legal parent states, politically influenced by external homelands since 1964 and 1999, respectively. Comparison across cases allows for evaluation of the mechanism in producing variation in this observed outcome. First, absence of the mechanism results in multiple decentralised pyramids, characteristic of conflict¹³⁴ and of the inability of ethnoterritorial elites to commit to homelands’ preferences. This is most evident in the RSK, where elite co-option and violent coercion were present, but Belgrade could not consolidate a hegemonic cross-border clientelist pyramid. This produced intra-elite competition between viable alternative elite factions in the RSK and the inability of the RSK to commit to Belgrade’s post-1991 preferences. Similarly, in Northern Cyprus the hegemony of the cross-border pyramid was weakened by Turkey’s neoliberal reforms, increasing political pluralism and encouraging alternative elites to challenge the UBP.

Second, the cases highlight the importance of local-level public sectors as mechanisms for distribution within disputed territories, linking publics to local elites serving as intermediaries. Employment-based clientelism is argued to resolve monitoring problems in clientelist exchanges due to continuous provisions of rewards and credible threats of losing access to them for not reciprocating political fealty.¹³⁵ This was readily observable in Kosovo, where parallel structures accounted for most employment and welfare distribution in Serb enclaves, and bureaucrats serving as intermediaries functioned as local elites. Threats of losing access to parallel employment or benefits effectively reduced intra-communal competition and leveraged political fealty to Belgrade’s preferences for maintaining ethnic closure from 2004 to 2013, and support for the SL after 2013.

Variation within cases further highlights the importance of public-sector clientelism. In Northern Cyprus, where post-1986 reforms cut the TRNC’s public sector, the UBP’s hold on power was weakened and intra-communal competition increased in the 1990s, able to challenge Anakara’s preferences for maintaining the status quo. Similarly in Kosovo in 2009–10, smaller Serb enclaves with small public sectors and administrations rejected Belgrade’s boycott policy and supported alternative elites in Kosovo’s elections, gaining access to Kosovo’s institutions and particularly welfare benefits. Further comparison to the RSK demonstrates that the absence of a viable public sector, including dysfunctional publicly owned enterprises and unpaid wages, led to participation in decentralised networks linked to individual alternative elites.

Last is the question of revising preferences in the context of internationally led conflict resolution and imposed costs for violating taboos on irredentism. In all three cases the homeland revised its discrete preferences for the conflict following political changes in their own internal-facing arenas. Where a hegemonic cross-border clientelist pyramid linked the homeland government to local ethnoterritorial elites and to local publics via public-sector institutions, the homeland could expect commitment to its revised preferences by revising the conditions of exchange of inducements. Accordingly, local elites, such as in Kosovo, exchanged political autonomy for inclusion as intermediaries in the hegemonic pyramid, securing political survival. Where these linkages were absent, increased competition between ethnoterritorial elites in the internal-facing arena, increased uncertainty and reduced credibility of commitment to new preferences were observed.

Conclusion

Congruent preferences between national homelands that pursue irredentist disputes and co-national communities in disputed ethnoterritories are not inherent. This study problematises the relationship between the two as a nested game in which elites simultaneously bargain in a dyadic external-facing arena and respective internal-facing arenas subject to competition, outbidding, and the changing costs of conflict. Commitment in an external-facing bargain is limited in its credibility by this competition in internal-facing arenas, thereby limiting the homeland’s ability to pursue certain dispute-related objectives, including efforts to escalate a conflict or pursue resolution to reduce externally imposed costs upon it. While cross-border ethnic and irredentist conflicts are remarkably complex, and this study does not purport to explain them in their entirety, the creation of hegemonic cross-border clientelist pyramids reduces internal-facing competition within disputed ethnoterritories and thereby increases commitments to external homeland’s preferences for a conflict. Case studies of Northern Cyprus, the Kosovo Serbs, and the RSK further indicate that variations in clientelist pyramids, including the strength of local public sectors and meaningful alternatives, can explain variation in commitment to external preferences and specific moments suitable for conflict resolution or points of escalation. While a strong cross-border clientelist pyramid may facilitate either escalation or resolution based upon the homeland’s preferences for a conflict, as Cyprus and Kosovo demonstrate, a weak one may likewise create opportunities for either by allowing autonomous local elites to pursue their own preferences, including inter-ethnic cooperation or unilateral escalation that passes costs to the homeland.

The outcome of hegemonic, cross-border, clientelist pyramid-building is the protracting of conflict and the persistence of states of uncertainty in disputed ethnoterritories. This contributes an explanation to the existence, in some cases for decades, of distinct ethnoterritories that are legally parts of parent states, but under the de facto political authority of others. Ultimately, in an international environment non-permissive to irredentist or imperial state-building, cross-border clientelism constitutes an alternative strategy for expanding state influence. Accordingly, it provides a long-term strategic explanation for the development of distributional infrastructure and institutions in disputed territories – laying the foundation of a cross-border clientelist pyramid that facilitates long-term political influence in claimed territories. Such developments were observable in Turkey’s incursions into northern Syria after 2014, a region it has long sought influence in, which included construction of public institutions and utilities networks, and training of local police inaddition to military operations.¹³⁶ It further contributes to practical understandings of resolving irredentist disputes, by centring the importance of clientelist networks in influencing preferences. On one hand, as in Kosovo, they may present a paradoxical challenge to conflict resolution by which a strong pyramid is needed to enforce a homeland’s preference for reaching a settlement but also solidifies long-term political influence that undermines sovereignty. On the other hand, this points to the need to counter cross-border clientelism and for peace builders to focus efforts on local-level political networks and distribution that links publics to central structures.