Corruption, patronage and illiberal peace: forging political settlement in post-conflict Kyrgyzstan

Abstract This article engages critically with recent literature on political settlements through a case study of inter-ethnic conflict in southern Kyrgyzstan. The case study traces how a new political settlement emerged in the aftermath of conflict, despite a rejection of international proposals on conflict resolution. Instead, local elites constructed an exclusionary form of social order, forged through dispossession and violence, maintained by informal institutions of patronage and clientage. The article explains why this new political settlement appeared remarkably resilient, despite its failure to address traditional liberal concerns regarding transitional justice and minority grievances. The case study highlights two major problems with the political settlements literature. First, it contests a widespread conceptualisation of political settlements as indicating a cessation of conflict, instead pointing to how a political settlement can be initiated and maintained through different forms of violence. Second, it questions notions of inclusivity in political settlements, noting that many political settlements combine logics of both inclusion and exclusion. In many cases, they are marked by exclusionary, authoritarian practices that together constitute a form of ‘illiberal peace’. These findings caution against a simplistic use of political settlements theory to inform policies aimed at resolving internal conflicts.


Introduction
In contrast to the expectations of liberal peacebuilders, most post-conflict political orders in the late twentieth and early twenty-first century have been marked by a pattern of clientelistic order-building, in which the coercive redistribution of property has played a prominent role. 1 These outcomes pose a significant challenge to theories of liberal peacebuilding, which assumed that political and economic liberalisation, combined with 'good governance' , would prove successful in achieving sustainable peace after conflict. Recent research on political settlements and on 'illiberal' or 'authoritarian' modes of conflict management offers new theoretical insights into why post-conflict regimes have often rejected liberal peacebuilding prescriptions and instead relied on hierarchical and exclusionary political and economic structures to produce relatively stable forms of political order. 2 We apply these theoretical approaches in a detailed case study of inter-ethnic conflict in the post-Soviet state of Kyrgyzstan to explain how political settlements may be renegotiated through violence and exclusion.
The article proceeds as follows. First, we engage with the literature on political settlements, particularly the work of Mushtaq Khan. We develop this research in a new direction, by linking it to an emerging literature on authoritarian forms of post-conflict order, notably the model of authoritarian conflict management developed by Lewis,Heathershaw and Megoran. 3 We then test the relevance of these theoretical frameworks through an analysis of conflict and post-conflict settlement in southern Kyrgyzstan, where inter-ethnic violence broke out in the city of Osh in June 2010 between Kyrgyz and Uzbek communities, leading to hundreds of deaths and the destruction of some 2000 homes. In our study we attempt to explain why an existing political settlement broke down, and how a new one was forged through a process of violent conflict and the renegotiation of informal institutions. This post-conflict order was not accompanied by an explicit 'elite bargain' or a formal peace process. There was no attempt to address underlying historical grievances or to engage in a public process of transitional justice. Instead, a new political settlement in southern Kyrgyzstan was achieved primarily through the renegotiation of informal institutions in the political economy, in ways which were ultimately accepted as broadly legitimate -and therefore sustainable -by key elites in different communities, despite their being forged through violence and exclusion.
Studying informal institutions is complex because they are not recorded in written documents and codified procedures. Our research in Kyrgyzstan builds on the tradition of political ethnography explored by Schatz, Wedeen and others as an appropriate method to understand both the functioning and the meaning of informal political institutions. 4 Not only does ethnography provide a way to understand how informal systems work, but it also provides insight into how different communities interpret these practices. Schatz argues that an ethnographic approach represents a certain 'sensibility' , 'an approach that cares to glean the meanings that the people under study attribute to their social and political reality' . 5 In Kyrgyzstan, local meanings of concepts such as peace, conflict, sovereignty and justice often diverge markedly from the categories used by outsiders. 6 Similarly, informal and formal practices in business and trade are not viewed as simply economic transactions based on cost-benefit analysis, but as socially embedded institutions informed by a deeper web of meanings and beliefs derived from historical and social experience. 7 The raw data for our study comprised more than 50 interviews with local businesspeople, political leaders, activists, security officials and representatives of local NGOs and international organisations, gathered through a series of visits to Kyrgyzstan over three years in 2012-2015. These interviews aimed to understand the discourse and practices of local elites who had a role in the post-conflict settlement in Kyrgyzstan. Since the potential for political bias is high in contested political environments, oral interviews were verified by comparing the accounts of representatives of different social groups and communities, and where possible triangulated with reference to public written sources, including academic research, public inquiries, government reports, etc. The credibility of interviews was also assessed longitudinally, since the authors had long periods of residence and research in Osh city at other times during 2000-2015, and contact with interviewees was in some cases maintained over a long period of time. Interviews were largely unstructured and often took place in social or informal settings; in many cases, they comprised multiple interviews, conversations and other social interactions in different settings in Osh, Bishkek and other locations over several visits. Interviews were conducted in Russian, Kyrgyz and English. Since much of the information was sensitive, the interviews were all anonymised to protect the security of interviewees. Where relevant we also used official statements, international reports, media reports and -where possible -official statistics and data, although in many cases these were not accessible or were considered unreliable. In particular, information on the ethnic breakdown of business ownership, the extent of property transfers during the conflict period, and the extent and significance of informal institutions in the economy are all difficult to verify, and largely depend on careful assessment and triangulation of non-official sources.

Political settlements and conflict
An emerging literature concludes that many externally-brokered peace processes have failed because they did not take account of the underlying 'political settlement' , the basic distribution of political and economic power in a society. Political settlements theory argues that 'violent conflict stabilises only when the allocation of benefits, opportunities and resources (such as political positions or business prospects) is consistent with how power is distributed in society' . 8 These findings have been influential in a new policy literature, which argues that excluding powerful elites from any peace process reduces the likelihood of a sustainable peace; consequently, external actors should support more inclusive political settlements. 9 As we demonstrate in our case study, the redistribution of property and business between different communities is central to understanding why violence breaks out and why it eventually abates.
These policy-oriented accounts of political settlement are heavily indebted to the academic work of the economist Mushtaq Khan. 10 Khan defines a political settlement as 'a combination of power and institutions that is mutually compatible and also sustainable in terms of economic and political viability' . 11 If the underlying power distribution is not adequately reflected in institutional structures, particularly those that govern the distribution of economic benefits, powerful groups will try to change the system, if necessary through violence. Conflicts are likely to break out 'when competing organizations do not accept a distribution of benefits as reflecting their true relative power' . According to Khan, 'a political settlement emerges when the distribution of benefits supported by its institutions is consistent with the distribution of power in society, and the economic and political outcomes of these institutions are sustainable over time' . 12 In Khan's model, conflicts act as 'a mechanism for establishing both a shared understanding of relative power and a distribution of benefits that is in line with this distribution of holding power' . 13 In other words, violent conflicts are a grim reckoning of accounts between rivals, a test of what Khan terms 'holding power' , defined as 'the ability of a particular organization to hold out in actual or potential conflicts against other organizations, including the enforcement organizations of the state' . Although 'holding power' has economic and coercive aspects, it also reflects non-material capacities, such as the capacity to mobilise supporters through ideology. 14 Since holding power is difficult to measure objectively and depends also on public perceptions, conflicts act as a mechanism to assess whether such perceptions are correct, to rebalance the distribution of resources and potentially to institute a new political settlement. Either 'institutions … adapt to the distribution of power, or conflicts can continue till a new distribution of power emerges' . 15 Ending conflict therefore requires not only new formal rules of the game, but a renegotiation of informal institutions, which act as the primary mechanism for distributing resources in developing economies.
Our study emphasises three additional points which are either disputed or overlooked in the current political settlements debate. First, one strand in the policy-oriented literature tends to conflate political settlements with inclusive 'elite bargains' , defining a political settlement as a 'conflict-ending agreement among powerful groups' . 16 But political settlements and elite bargains are distinct concepts. As Khan argues, the concept of political settlement is more analytically useful when it references underlying distributions of organisational power that achieve some measure of sustainability. 17 Political settlements may not end all forms of violent conflict. In a study of political settlement in the Niger Delta, Schultze-Kraft demonstrates how organised violence and organised criminal activities are integral to certain aspects of the political settlement. 18 In the Democratic Republic of the Congo (DRC), Perera finds that an 'inclusive' but violent settlement has created a system in which elites have significant incentives to continue to perpetuate violence. 19 A political settlement -as we find in our own study -may be forged through the violent redistribution of property, and imposed and maintained through coercion and exclusion. Political settlements should not be understood as merely a new incarnation of liberal models of inclusive peacebuilding, but represent a much more normatively challenging model, in which political order is often constructed through coercive practices and structural violence.
Second, although rent-sharing mechanisms are an important part of most political settlements, these are not simple economic exchanges to 'buy peace' , but embedded cultural processes, which have meaning for participants beyond simple financial transactions. Khan argues that 'all developing countries have variants of "clientelist" political settlements' , defined as those in which 'significant holding power is based on sources outside the incomes generated by formal institutions' . 20 These informal institutions provide mechanisms to support elite bargains based on rent-sharing agreements, which can stabilise violent conflict by persuading armed factions to participate in a stable political coalition. 21 Such arrangements can lay the basis for what North terms a 'limited access order' , a rent-sharing agreement that achieves some measure of sustainable political order. 22 In most cases, informal institutions go beyond simple forms of rent-sharing; they are also embedded in existing systems of social relations and cultural practices, and are shaped by a set of 'semiotic practices' , in which the meanings ascribed to everyday economic transactions have profound importance for their perceived legitimacy in society. 23 In short, economic models of political order based on rent-seeking behaviour need to be complemented by an interpretive account of how such mechanisms are understood and legitimised in local communities.
Third, the political settlement literature emphasises the importance of inclusivity -the incorporation of diverse elite groups in political processes -but in reality, political settlements are characterised by logics of both inclusion and exclusion. They often result in a hierarchical, authoritarian form of political settlement, achieved through the repression and dispossession of rival elites. Many conflict-affected societies are characterised by what Khan terms 'competitive clientelism' , implying a constant process of negotiation among different groups over mutual 'holding power' , i.e. over the terms and conditions of the political settlement. 24 Since the analysis of holding power is always uncertain, such systems are often marked by frequent attempts to test the power of opponents through small-scale unrest, riots, election violence or other forms of contested politics. To avoid the constant risk of conflict associated with competitive clientelism, post-conflict governments often attempt to impose an authoritarian system, resulting in what recent literature has termed 'authoritarian' or 'illiberal' modes of conflict management, observed in a range of recent conflicts, such as Chechnya, Sri Lanka or Syria. 25 In one model Lewis, Heathershaw and Megoran conceptualise this as 'authoritarian conflict management' , a set of practices ranged across the domains of discourse, space and political economy. Such an authoritarian post-conflict order aims to construct a 'single-patron' rather than a 'multiple-patron' order in the political economy, forming a monolithic pyramid that fuses political and economic power and avoids the conflictual nature of 'competitive clientelism' . 26 The result is a hierarchical, illiberal 'peace' , in which the violent redistribution of economic resources during a conflict contributes to the emergence of a new political settlement marked by the institutionalisation of new power realities within authoritarian political structures. 27

The breakdown of political settlement in Kyrgyzstan
Post-Soviet Kyrgyzstan provides a clear case study of how informal political settlements can be undermined by shifts in holding power between two groups, and how they may be reconstructed in the aftermath of conflict through both informal institutions in the political economy and patterns of violence and coercion. A focus on the south of Kyrgyzstan highlights the importance of sub-national conflict dynamics, seeking to address a concern that research on political settlements primarily focuses on national-level conflict dynamics and settlements. 28 The Osh region of southern Kyrgyzstan was severely affected by the tensions and fractures stemming from post-Soviet economic collapse, the influence of radical Islamist groups and competitive state-building projects in Uzbekistan, Kyrgyzstan and Tajikistan, which created serious border tensions and divided societies in the Fergana valley, where the three states meet. In June 1990, some 300 people died in fighting between the majority Kyrgyz and the minority Uzbek community in the region of Osh. The conflict was sparked by demographic shifts and disputes over land and other economic resources, as ethnic Kyrgyz began to move in greater numbers into Osh and other towns, which had been traditionally dominated by ethnic Uzbeks and Russian-speaking groups. 29 KGB reports after the 1990 violence pointed to multiple causes, including 'a perception among poorer ethnic Kyrgyz that Uzbeks were becoming more prosperous […] and controlling the markets' . 30 The perceived economic prosperity of the Uzbek community was often exaggeratedmany Uzbek residents were just as poor as their Kyrgyz neighbours -but many of the leading businesspeople in the city came from the Uzbek community. 31 Uzbek entrepreneurs dominated some key sectors of the economy, particularly service sector businesses and crafts such as catering, hospitality, retail, transport and cross-border trade, although even before 2010 Kyrgyz entrepreneurs were increasingly active in a growing number of economic sectors.
This economic influence of Uzbek elites was not matched by political power. Ethnic Uzbeks enjoyed some limited, formalistic representation in local and national politics. While the mayor of Osh was always a Kyrgyz, one of two deputy mayors was usually an Uzbek, while the chairman of the city council was also usually an Uzbek. The Uzbek community enjoyed some representation in the national parliament, with between five and seven Uzbek representatives being elected in successive parliaments between 1995 and 2005. They also usually maintained at least one individual in the national government at the level of deputy minister. 32 In the 2007 parliament there were six Uzbek deputies out of 90 (6.6%), but in the 2010 parliament, elected after the June 2010 violence, only three Uzbek deputies (2.5%) out of 120 remained, reflecting the new political realities of the post-conflict period. 33 But all these figures were much lower than the proportion of Uzbeks in the national population. 34 The situation was even worse in the civil service and the security forces. According to official figures, in 2009 out of 17,978 state employees, there were only 181 representatives of ethnic minorities. 35 The local police and procuracy were dominated by ethnic Kyrgyz, and as Soviet-era appointments retired or left, the imbalance was worsening. 36 By 2010 there was only one Uzbek judge out of 110 judges in the 28 courts of southern Kyrgyzstan, and only one Uzbek investigator in the national security agency. 37 The relative imbalance between the 'holding power' of Uzbek and Kyrgyz elites was managed partly through a range of discursive and symbolic mechanisms, but primarily through an institution of 'clientage' , which at its simplest involves the payment of protection money to a patron in exchange for commercial autonomy. 38 In practice, it could involve much more complex exchanges, including rights to trade (licit and illicit), informal credit facilities, property protection or financial investment. Although this system was effectively a protection racket, it was institutionalised and viewed by leaders in the Uzbek community as the most effective way to maintain property rights (in the absence of reliable courts or other formal institutions) and to reduce the potential for conflict. As one local deputy recalls: 'Whenever a new mayor came to power, it was Uzbeks who always came to him first, they brought money and [said] "We won't touch you and you don't touch us"' . 39 In other words, Uzbek leaders gained a certain level of protection from possible violence or problems in their business activities through regular payments to local formal and informal authorities; for many of its participants this system was seen as a useful and legitimate mechanism for maintaining a reproducible political settlement marked by relative inter-ethnic peace.
Under the presidency of Kurmanbek Bakiev, after 2005, these informal clientage institutions came under increasing pressure. The relative political influence of the Uzbek community declined under the new regime, as they faced the rising aspirations of a post-Soviet generation of ethnic Kyrgyz entrepreneurs, mobilised in part by a wave of nationalist feeling. These represented what Michael Kalecki termed 'intermediate class groups' in developing countries, a kind of petit bourgeois proto-middle class which struggled for power in post-colonial regimes. These groups, which typically emerged outside formal institutions and conventional business, played an important role as political entrepreneurs, particularly at local level. 40 In pluralist political systems in developing countries, such as Kyrgyzstan, political struggles typically comprise 'contests between factions led and dominated by members of the intermediate classes' . 41 Figures from this social class played important roles in Osh, as bazaar owners, traders and local 'authority' figures, sometimes involved in semi-criminal connections and activities. Their emergence challenged Soviet-era elites, both urban Kyrgyz from the north, who had a legacy in Soviet political and government structures, and Uzbek entrepreneurs in the south, many of whom had developed their business and trade experience in Soviet economic and trade institutions.
Khan argues that '[t]he only viable redistributive strategy for developing country political entrepreneurs in the absence of any fiscal or regulatory space is to organize enough organizational muscle to be able to capture resources through a combination of fiscal, off-budget and even illegal means' . 42 They are able to do so because they have legitimacy and superior mobilisational and organisational power. In Kyrgyzstan these local leaders had significant mobilisational power, through clientage networks or kinship groups, or through organised crime networks. In Osh many local criminal and political leaders funded 'sports clubs' , where their followers trained and congregated. Local police estimated there were nearly 150 such clubs in 2010, most of which were exclusively visited either by young Kyrgyz or young Uzbek men; they were often run by trainers with criminal records, and operated under the protection of 'either political leaders or criminal groups, or some combination of the two' . 43 A typical representative of these newly emergent 'intermediate class groups' in Kyrgyzstan was Melis Myrzakmatov, who served as mayor of Osh in 2009-2013. He graduated from Osh Technical University as an 'economist-manager' , before joining the local police force and the tax inspectorate -typical prerequisites for a successful business career in Kyrgyzstan. In 2007 he moved into politics, and was appointed as mayor of Osh in January 2009 by President Kurmanbek Bakiev. Uzbek leaders in the south were worried about his reputation as a Kyrgyz nationalist, and in interviews Uzbek businessmen claimed that the mayor's team began to undermine the informal institutions of the post-1990 political settlement, including a renegotiation of the clientage system. 44 It was not merely a potential increase in costs that upset Uzbek entrepreneurs, but the semiotic context of these exchanges: 'Until Melis' , argues one former official, 'the authorities always did deals with the Uzbeks: and they didn't deal with them arrogantly (naglo ne delali). 45 Even where existing protection roofs (krysha) were not disbanded, they were no longer conducted in a way that accorded with the other side's understanding of what was culturally appropriate. In other cases, protection was simply curtailed. One opponent later said: 'Melis did not provide a roof for business. He just forced people out of business' . 46 One of the strengths of Khan's model of political settlements is his recognition that non-material factors play an important role in patronage systems. Alongside economic wealth, '[t]he political ability to organize, the numbers of people that can be mobilized, and perceptions of legitimacy' all contribute to holding power. 47 Myrzakmatov's accumulation of economic power was inseparable from the symbolism and meaning with which these new elements of an emerging political economy were imbued. The assertion of political power against the perceived economic advantages of the Uzbek community had particular resonance within Myrzakmatov's narrative of Kyrgyz nationalism. Moreover, Myrzakmatov marked his difference from traditional Kyrgyz elites, with a populist message that resonated with many young Kyrgyz. His supporters claimed that he was a new type of politician, who did not 'steal from the budget' and 'checked the streets at 7 am every day' . 48 This well-constructed image -achieved with the help of journalists and media advisers -contributed to a narrative of Myrzakmatov as a strong Kyrgyz leader who could properly represent the interests of the ethnic Kyrgyz population in the south, even at the expense of minority groups.

Violent conflict and dispossession
Despite these underlying shifts in power at the local level, a peaceful transition to a new political settlement might have been managed were it not for the collapse of central political authority in April 2010, when President Bakiev was ousted from power by opposition protests in the capital, Bishkek. He was replaced by a weak interim government, which was unable to act as a 'guarantor' of the existing informal political settlement in the south, leaving a fluid political and economic environment in which rival groups were tempted to test their own holding power through the use of violence.
The first clashes broke out in May 2010 in Jalal-Abad between local Kyrgyz activists and followers of Kadyrjan Batyrov's Rodina party, which had long campaigned for more rights for the Uzbek community. But alongside this political conflict, there was also evidence of an emerging struggle over economic resources that frequently mapped onto ethnic divides. In late April 2010, there was a confrontation between local Uzbek entrepreneurs, importing cars from South Korea, and a local (Kyrgyz) criminal gang demanding higher protection payments. The stand-off quickly developed into clashes between Uzbek and Kyrgyz youth, but police eventually ended the fights. 49 Similar tensions also emerged in Jalal-Abad, where Aibek Mirsidikov -an alleged drug-trafficker with the nickname of 'Black Aibek' -was assassinated on 7 June. According to Neil Melvin, 'the May violence was also a confrontation between key Uzbek figures struggling to control local economic resources and … [figures] from local criminal circles whose position had advanced under Bakiev' . 50 Against this backdrop of growing tensions over economic resources and political power, serious violence broke out in Osh in June 2010 between Kyrgyz and Uzbek groups. During several days of rioting and pogroms in 11-14 June, some 470 people were killed, and around 2700 properties totally destroyed, including over 1900 in Osh city. Initial clashes involved violence from both sides, but as crowds of Kyrgyz rioters gained the upper hand, the violence turned into one-sided attacks on Uzbek communities, who lived in mahallas, compact areas of Uzbek settlement, sharply distinguished from the Soviet apartment blocks commonly occupied by ethnic Kyrgyz and other nationalities. Groups of Kyrgyz men carried out selective attacks, targeting Uzbek businesses and homes, while leaving those belonging to Russians, Kyrgyz or other nationalities unaffected. 51 An ICG report concluded that 'the mobs that roamed Osh were not for the most part mindlessly set on random destruction' , but 'were well organised and often well supplied' . 52 One rioter was quoted as saying that they were looting because Uzbeks had 'started to live too wealthily' . 53 Even as the violence died down, the redistribution of property began. Kyrgyz rioters stole properties or forced Uzbeks to sell cars at nominal prices to permit them to leave. Rioters attacked people physically to force them to leave the city, and took possessions from them as the price for safe passage. 54 The overall pattern fitted Khan's model of a conflict acting as 'a mechanism [to establish] a distribution of benefits that is in line with [the …] distribution of holding power' . 55 A local official articulated this interpretation in simple terms: 'The root of the Osh events was that the Uzbeks had taken control of all the business, and [then there was] a brutal redistribution of property [zhestkii peredel]' . 56 He went on to assert that 'A redistribution of property took place. The Uzbeks left, sold their businesses, that's a fact' . 57 During a second phase of the post-conflict period (July 2010-2011) local Kyrgyz elites seized more businesses through the process often described using the Russian term reiderstvo [raiding], meaning forced takeovers of business through the threat of violence or criminal prosecution, subsequently legitimised through formal codification of the sale. This process involved the semi-voluntary or forced sales of Uzbek businesses to new Kyrgyz owners. The extent of this process is difficult to measure, because in some cases, there was a formal appointment of an ethnic Kyrgyz as director of an Uzbek businesses as a form of protection.
One businessman explained that 'they appointed Kyrgyz [as directors] because of the nationalism and so that the police did not put pressure on them' . 58 A well-connected Uzbek leader claims that 'after 2010 the majority of restaurants and business were reregistered to Kyrgyz' . 59 Another claimed that prior to 2010 '90% of restaurants were owned by Uzbeks, but by 2013 70% were owned by Kyrgyz businesspeople' . 60 In the transport sector, one entrepreneur claimed that of around 15 companies that he knew were managed by Uzbek businesspeople before 2010, only two or three remained intact. 61 In theory those who lost their businesses in the violence were eligible for compensation, but in practice this did not always help. An Uzbek businessman said: 'Many lost their business through it being burnt down or through reiderstvo -Kyrgyz bought up the business, and then you can't say that it was raiding. If you sold the business, then you can't apply to international organisations [for compensation]. People just don't apply' . 62 Many Uzbek businesspeople left Osh during this period. Thousands fled to Russia or to Kazakhstan. Those Uzbek businesspeople who remained in southern Kyrgyzstan recounted several mechanisms they used to seek new modes of protection in this challenging environment, seeking alternative protection kryshas, doing deals with local police, gaining Russian citizenship or shifting the bulk of their business to Russia or to Bishkek. 63 In 2010-2011, however, there was still no new political settlement: several interviewees complained that during this period it was difficult to find a reliable krysha, with the police in particular playing a highly predatory role. There were multiple cases in which ethnic Uzbeks were subject to arbitrary arrest, imprisonment and torture. In such cases police typically threatened to concoct prosecutions on the basis of crimes committed during the June 2010 violence unless the accused paid large sums of money for their release. In September 2011, interviewees reported that ransom payments of US$1000 or US$1500 were typical in such cases. 64 Traditional practices of brokerage -whereby individuals with good connections in both communities could intervene on behalf of victims -appeared to no longer be functioning effectively. 65 In response to the conflict and its repressive aftermath, the Organization for Security and Cooperation in Europe (OSCE) had proposed an international police mission, which was designed to work with the Kyrgyz authorities to address underlying grievances and resolve potential triggers for future conflict. The OSCE High Commissioner on National Minorities Knut Vollebaek called for 'a power-sharing agreement' to include 'ethnic Uzbeks and representatives of other national minorities in all state institutions, including law enforcement agencies, judicial bodies and public administration at all levels' . 66 Subsequently, an international Kyrgyzstan Inquiry Commission (KIC) called for Uzbek social grievances to be addressed, including recognition of the status of the Uzbek language and 'ensur[ing] that the composition of the security forces reflects the ethnic diversity of the society' . 67 These international proposals to address minority grievances were strongly rejected by local and national authorities in Kyrgyzstan. Demonstrators in Osh, encouraged by Myrzakmatov, denounced what they termed plans for 'another Kosovo' in Osh, seeing international proposals as the prelude to inevitable Uzbek separatism. 68 In the end, the OSCE's police mission, diluted to an impotent advisory role in response to objections by the Kyrgyz government, failed to have any significant impact. 69 Instead of agreeing to international blueprints for an inclusive post-conflict process, Myrzakmatov presided over a new, highly illiberal, political settlement, which sought to manage the conflict using coercive, authoritarian practices. An ethnic Uzbek had traditionally been appointed as deputy mayor, but Myrzakmatov discontinued that practice; he also replaced ethnic Uzbeks with ethnic Kyrgyz officials in administrative posts in districts of majority Uzbek settlement. 70 Myrzakmatov continued to consolidate control over businessaccording to one local businessman, the mayor informally controlled 90% of the business in the city by 2013. 71 Although this is probably an exaggeration, another local observer claimed in late 2013 that 'business in Osh without the support of the mayor is impossible' . 72 One entrepreneur argued that the city had been divided up among two or three powerful players, but that Melis was by far the most powerful: 'Melis is god and tsar!' 73 Myrzakmatov also proposed a radical new reconstruction of the city, which would have demolished many buildings in the city centre and replaced traditional Uzbek mahalla communities with multi-ethnic high-rise apartment blocks. The mayor also proposed to transfer Osh's famous bazaar outside the city. 74 One Uzbek leader admitted that 'the Kyrgyz also suffered' from predatory practices, but that these were legitimised by local Kyrgyz leaders, who characterised extortion as a means for 'squeezing the Uzbeks' . 75 This kind of ethnicization of practices of extortion represented a form of victory for local Kyrgyz in what was widely described locally as a 'war' .

Renegotiating political settlement
Paradoxically, it was the institutionalisation of these economic practices that formed the basis of a new informal political settlement that began to emerge in 2012-2013. There was little doubt that Myrzakmatov was widely disliked by the Uzbek community. One local Uzbek businessman explained: 'Melis always had difficult relations with ordinary Uzbeks […] They [Uzbeks] will follow anybody rather than Melis!' 76 Yet during 2012-2013, Myrzakmatov began attempts to win over Uzbek leaders and entrepreneurs including -according to interviews with Uzbek businesspeople and Kyrgyz officials -providing them with new informal protection mechanisms for their businesses. 77 One Uzbek businessman, Amanyllo Iminov, who was also the deputy speaker of the local kenesh (assembly), became an open supporter of Myrzakmatov and in 2011 even joined his Uluttar Birimdigi party. Several other Uzbek businessmen become publicly identified with the mayor in the city. This close association with the mayor certainly damaged the credibility of some of these Uzbek leaders with parts of their community. Yet other Uzbeks also turned to the mayor for support. When asked why, one local politician argued that 'It is what we are used to. First we had Aidar [Akaev], 78 then Koshelek [literally, in Russian, 'wallet' , the nickname of a local member of parliament], then [the Bakievs], and then Melis came along' . 79 In this way institutionalised clientage mechanisms re-emerged, with different terms and conditions reflecting a new reckoning of political and economic power between the two communities.
The structure of this new political settlement was constructed according to Khan's model of an equilibrium of 'holding power' between the two sides. On the one hand, the conflict had demonstrated that the Uzbek community was much the weaker -Kyrgyz leaders could mobilise supporters, and had at least the tacit support of parts of the coercive state apparatus. But after the conflict, the Uzbek community began to attract international support. The government in neighbouring Uzbekistan had not intervened in the conflict, but had made clear that it would not countenance further violence towards ethnic Uzbeks in Kyrgyzstan. The Russian government also did not intervene, but showed sympathy towards the position of ethnic Uzbeks, some of whom held Russian citizenship. International financial institutions, such as the Asian Development Bank, funded much of the city's reconstruction, and therefore also had leverage. These external pressures partially constrained the ability of local Kyrgyz elites to mobilise on nationalist grounds to achieve political and economic goals, and limited the ability of local elites to use violence to maintain political and economic dominance.
Second, the Uzbek community began to regain some political and economic clout. Scheduled elections to the local Osh council (kenesh) in 2013 reminded local Kyrgyz leaders of the importance of the Uzbek vote. As one former official argued: 'He [the mayor] helped influential Uzbeks so that they would help him at the elections -he only helped Uzbeks for his own political goals' . 80 Moreover, the logic of pre-existing informal institutions began to reassert itself. The violent redistribution of property during the conflict and the departure of some Uzbek entrepreneurs provided a one-time transfer of resources and property to local elites, but damaged the local economy. The direct seizure of assets had reduced the ongoing 'clientage' payments received by local elites, and therefore a new political and economic settlement was increasingly attractive for economic reasons.
Third, Myrzakmatov failed to achieve a fully effective authoritarian political settlement, demonstrating the difficulty of achieving a sub-national political settlement while in dispute with the central authorities. According to the model of authoritarian conflict management, sustainable authoritarian responses to conflict require control over a single-pyramid patronage system. 81 But Kyrgyzstan had always been characterised by a pluralistic system of multiple pyramids of patronage. 82 Myrzakmatov attempted to construct an alternative, monolithic subnational political system, but there were many other powerful economic actors in play, including at the national level. An initial attempt by the central government to oust Myrzakmatov in 2010 had failed, but the national government continued to constrain his attempts to take total control in Osh. The government used the donor-funded State Directorate for Reconstruction and Development of Osh and Jalalabad, headed by presidential ally Jantoro Satibaldiev, to create a parallel flow of funds into reconstruction and rehabilitation projects, outside the mayor's control. 83 Above all, Myrzakmatov's influence over the security forces lessened sharply in 2012-2013, after the appointment of Suyun Omurzakov as head of the city police on 11 January 2012, and subsequently as head of a combined interior ministry command for Osh and the surrounding region in February 2013. Omurzakov was an experienced political operator, and well-known locally. 'He's a good fighter, daringhe'll break Melis' , predicted one local businessman, in December 2013, who knew both men. 84 And so it turned out. During 2013 the security services and interior ministry began to work against Myrzakmatov more openly; delegations from Bishkek arrived to check on the loyalty of the local police. In December 2013 a former official from the mayor's circle said, 'all the law enforcement agencies are working against us! The whole Republic is working against the mayor' . 85 Myrzakmatov stood in fresh mayoral elections (held among members of the city kenesh), but on 15 January 2014 Myrzakmatov lost the vote, and a new mayor, Aitmamat Kadyrbaev, took over. 86 Myrzakmatov subsequently left the country, but was sentenced to seven years in prison in absentia in July 2015 on charges of corruption and abuse of office. 87 Although marked as a sharp political rupture, the dismissal of Myrzakmatov merely resulted in a more sustainable institutionalisation of Osh's new political settlement. The informal institutions of clientage gained new patrons, but the underlying premise of the political settlement remained the same. 88 The new local leadership developed more effective engagement with Uzbek community leaders, and the overt repressions and ad hoc extortions against Uzbek citizens of 2010-2012 were largely ended. The police force in Osh became much more restrained in the sphere of ethnic relations, and there was considerable effort to improve the city's reputation through symbolic politics aimed at improved inter-ethnic relations. Beneath the surface, however, it was a renegotiated political settlement and reconfigured informal institutions of clientage that underpinned a fragile stability in the city.

Conclusion
Political settlement theory provides a productive framework to explain the outbreak of violence in Osh in June 2010, and to investigate why a new post-conflict political settlement gradually emerged after a period of violence and repression during 2010-2012. The terms and mechanisms of this new settlement were radically different from those proposed by international actors, who advocated a traditional liberal agenda of minority rights and transitional justice, overseen by an international police mission. Western NGOs warned that a failure to address these underlying minority grievances and continuing ethnic tensions would result in renewed conflict, but contrary to these warnings, the political settlement proved rather durable. NGOs and external funders promoted programmes designed to improve inter-ethnic understanding, mediation and reconciliation but there was little evidence that these were effective. 89 Rather, the real dynamics of peace and conflict were located in the precarious balance of economic power between communities, managed by opaque and corrupt informal institutions. The new balance of economic and resource distribution was forged in communal violence, and imposed through authoritarian practices. But the new settlement ultimately gained stability because it reproduced long-standing informal institutions, primarily those of clientage and brokerage, which enjoyed a measure of legitimacy among both communities.
The Osh case study contributes to the wider debate about what constitutes a political settlement. The stabilisation of the situation in Osh did not involve a deliberate elite pact; there was no conscious sense of what Kelsall describes as an 'ongoing, conflict-ending agreement among powerful groups' , or what Khan dubs a 'planned settlement' . 90 Instead, the Kyrgyz case corresponds to what Khan describes as 'an "interactive order", in which an identifiable and fairly robust social order exists, but it is the outcome of many interactions between groups and not based on any agreement or pact that can be identified ex ante' . 91 A conscious elite agreement about resolving the conflict would always have been difficult, since there were profoundly different understandings of the causes of the conflict, and no agreement on how underlying grievances should be addressed. Informal institutions in the political economy, however, could be renegotiated and reinvigorated in ways that bypassed these fundamental differences, and yet were viewed as legitimate in an historical and cultural context. In this way a political settlement was restored, although the structural violence and the grievances that it entailed remained largely unaddressed. 92 Second, informal institutions involving clientage payments and protection were central to the re-emergence of a political settlement, but they were not simply about financial payoffs and protection rackets. Instead, these informal institutions re-emerged and were renegotiated as part of a much wider set of networked relationships within the cultural context of Osh, including new and revived symbolic and discursive practices, a resurgence of patterns of political representation and electoral support, and other informal institutions and practices of brokerage and mediation, which served to legitimise and manage the new political settlement.
Third, the form of political settlement that emerged in Kyrgyzstan was not 'inclusive' in the sense widely used in the policy literature to denote inclusivity of factions in an elite bargain. 93 On the contrary, the political settlement involved the subordination of a minority group within a narrow, exclusionary settlement, characterised by authoritarian practices. Instead of a new, horizontal elite coalition, the political settlement was formed through the forced inclusion of minority groups into hierarchical pyramidal structures, characterised by vertical patronclient relations, and policed by the use of coercive, discursive and economic mechanisms. This type of political settlement is no longer adequately conceptualised by models based on inclusivity, but instead is better understood as a form of 'authoritarian conflict management' , a set of authoritarian and illiberal practices that aim to control and manage violent conflict.
The Kyrgyz case should act as a cautionary note to policymakers' increasing recourse to the idea of political settlements as an alternative framework for conflict resolution. Political settlements are almost always intertwined with different forms of violence, involve logics of both inclusion and exclusion, and are often hierarchical and authoritarian in their outcomes. Consequently, it should be recognised that Khan's political settlement theory and models of 'illiberal peace' pose a normative challenge to models of conflict resolution and liberal peacebuilding, threatening to legitimise violence as an inevitable, 'natural' mode of redressing power imbalances between communities, and justifying authoritarian responses as an acceptable means to achieve a cessation of violence.
Instead, these new analytical approaches should be used to inform more realistic and context-specific policies. A recognition of the messy reality of how order is often produced in post-conflict societies, through violence, exclusion and dispossession, is the first step towards more effective responses by both internal and external actors in ending civil wars. Conflict resolution may require alternative pathways and mechanisms to allow political settlements to emerge peacefully, for example through institutional reforms or new mechanisms for economic redistribution that address sharp imbalances between communities. Conflict prevention policies may seek to address issues of economic inclusion alongside the more traditional focus on minority political grievances. Conflict analysis needs a much better understanding of how informal economic institutions can help to constrain violence and allow viable post-conflict orders to emerge. Political settlement theory provides important insights into why many external prescriptions for resolving conflicts fail, but it can also offer a new appreciation of how policy responses informed by the complexities of local politics can encourage divided societies to move towards more peaceful and sustainable outcomes.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
Research for this article was funded by the Economic and Social Research Council Grant No. ES/ J013056 /1, Rising Powers andConflict Management in Central Asia (2012-2016).

Notes on contributors
David G. Lewis is Associate Professor of International Relations at the University of Exeter. His research is mainly in the field of peace and conflict studies and on comparative authoritarianism, with a regional focus on Russia and Eurasia.
Saniya Sagnayeva is an anthropologist, with extensive experience of research in Central Asia. She has worked as a Senior Political Analyst at the International Crisis Group, and as a researcher with the University of Exeter on the project Rising Powers and Conflict Management in Central Asia.