CIBA on Burara colline: associations as technical subcontractors

The first case-study is the intervention of CIBA on Burara colline. This colline is located near the administrative centre of Busoni commune, which links both sides of the hills surrounding the Muhembezi wetland. A large IDP camp straddles both sides of the wetland. For several reasons, Burara colline, and Busoni commune in general, have a clear ethnic segregation: a politically conscious Tutsi minority in the centre and a Hutu majority on the surrounding collines. First, Burara has a considerable group of resettled Tutsi near its centre (the IDP camps). In addition, this Tutsi group has direct links with national politics and the Tutsi elite in the capital. Several important figures within the Uprona (Tutsi) party originate from Busoni commune. Consequently, the local political struggle between the CNDD-FDD (Hutu) party, which won the 2005 elections (also on a communal level), and the Uprona (Tutsi) party was particularly intense, while ethnic tensions remained prevalent in the commune. Busoni commune therefore was an exception in northern Burundi, in which the CNDD-FDD in general held a very strong position (as was the case on the other studied collines). Nevertheless, this local political struggle did not substantially affect the functioning of the community associations related to the CIBA project. As will become clear, this can be explained through CIBA's close technical follow-up of the community associations, leaving minimal room for local political power relations to influence their functioning.

Of the three development agencies studied, CIBA had the most explicit focus on technical end results such as improved irrigation systems in the wetlands, water tap infrastructure, anti-erosion measures and road protection, improvement of local livestock by cross-breeding, and the certification of improved rice seed varieties. This was reflected in the staff policy, as technical staff (agronomists, veterinarians, engineers) outnumbered social workers. In comparison with the other studied projects, CIBA therefore followed a rather top-down approach to achieve these technical end results: each specific technical end result was closely followed and monitored by one particular technical staff member – a constructional engineer for the irrigation system, a forester for the anti-erosion measures, a veterinarian for the livestock component, and so on. As a consequence, technical staff were present daily on the collines to follow up the project activities. This ‘technical’ focus also substantially guided the constitution of the community associations, as associations were clearly linked to specific technical project objectives.

A first illustration of this technical approach are the anti-erosion associations. Because of deforestation and extensive land use, each season heavy rains cause floods, landslides and further degradation of the soil. Anti-erosion measures (contour lines and hedges) had previously been executed through food-for-work (FFW) and cash-for-work (CFW) programmes, also under the supervision of CIBA and the World Food Programme (WFP) in Burara. After the FFW programme stopped, CIBA preferred to make past efforts sustainable, raising public awareness to create one anti-erosion association for each colline, since soil protection was best organised by geographical entities. Although the associations were set up, the population did not accept these changes. Until that time, food aid compensation was the main incentive for the people involved. In their minds, associations represented a continuation of the FFW programme, as they expected to receive the same incentive (food aid) for the same work (contour lines), but delivered through a new channel or ‘gateway’ (via the association instead of individually). For them, the new associations were technical subcontractors, which should be paid according to the delivered services within the scope of the livelihood project. The fact that there was little or no coordination among the intervening actors made the situation even more complex, as other NGOs did compensate for the anti-erosion works on neighbouring hillsides. CIBA also had contradictions in its own approach. While staff tried to motivate the population to implement anti-erosion activities without payment, other staff paid the wages of the casual workers repairing the roads. Thus, when it became clear that food compensation would not follow, most of the anti-erosion associations stopped their activities and ceased to exist.

A second type of association facilitated the implementation of the rice programme in the Muhembuzi wetland. First, water user committees were established to maintain the irrigation system. Second, these committees were transformed into rice farmer groups to ease the technical support and to structure the rice programme. Third, all sixty-nine rice farmer groups were united into a single rice cooperative, to ease the commercialisation of rice production, and improve market access. These stages clearly illustrate the assumed linear progression of community associations as mentioned in the theoretical section: bonding social capital (the ability to maintain local efforts in the management of the irrigation system) is complemented by bridging capital (the ability to create bridges and unite all rice farmers in the wetland) to create linking social capital (the ability to influence the rice market through the rice cooperative). However, also in this case, associations functioned as new ‘aid gateways’. On the one hand, people accepted the new social infrastructure of water user committees, rice farmer groups and the cooperative because CIBA promised to buy up the rice production during project years. On the other hand, community associations served as temporary intermediaries to transfer knowledge, skills and technology. As one farmer argued: ‘The associations were founded because of CIBA's aid. Now, we know about modern agriculture. If the aid is working, we will use this knowledge for our own fields, but the associations will stop existing’ (focus group, 20.04.2007, CIBA project, Burara). Indeed, once the project had finished, the majority of the rice farmer groups ceased to exist.Footnote ⁸

In all cases, CIBA used associations in specific situations where collective action problems arose: natural resource management (the maintenance of the irrigation system in the wetland and anti-erosion measures on the hillsides) (e.g. Krishna & Uphoff Reference Krishna and Uphoff1999; Pretty & Ward Reference Pretty and Ward2001), and the commercialisation and marketing of agricultural production (the rice programme in the wetland) (e.g. Bebbington & Carroll Reference Bebbington and Carroll2000; Sorensen Reference Sorensen2000). However, these associations never merged with existing social capital endowments. Most associations remained exogenous or ‘alien’ institutional settings, having a reason for existence only within the scope of the CIBA project.

To a great extent, this lack of institutional syncretism (cf. Galvan Reference Galvan2007) can be explained by the fact that the associations, as new institutional settings, were designed with reference to external technological end results, rather than to local social capital endowments. In all three cases, their constitution and composition were based on technical criteria. For example, in the wetland, farmers were asked to form groups of thirty people with adjacent plots: these criteria of proximity and group size were based not on prior social organisation, but on maintaining the newly introduced irrigation system. The same holds true for the anti-erosion association: the fact that its scope coincided with the boundaries of Burara hillside was not a social but a geographical demarcation, based on the technical feasibility of the anti-erosion measures. None of these criteria, however, guaranteed the existence of a ‘social basis’ to maintain this technical collaboration among community members. Because of their close connection to technical end results, the newly created community associations were initially perceived as new ‘aid gateways’ – i.e. channels for the development agency to achieve technical project goals, or for the local population to access support from the external agency.

However, with regard to post-conflict reconstruction, the associations did succeed in creating bridging ties between the different ethnic groups. The NGO kept lists indicating the multi-ethnic and gendered character of both the associations and their committees; all associations included both Hutu and Tutsi. Some associations even had Twa members, an often forgotten and extremely deprived ethnic minority in the past. But the creation of bridging associations presented a potentially ‘false sense’ of inclusiveness. As already mentioned, the Busoni community has a segregated ethnic outlook: most Tutsi live in the centre, whereas most Hutu live in the surrounding hillsides. Because of this clear geographical segregation, and the presence of a large group of Tutsi in the centre, ethnicity still played a key role (compared with the other hills studied). As a result, most associations were Tutsi (centre) or Hutu (hills) dominated. Members openly indicated that they sought extra Tutsi or Hutu to make their association multi-ethnic because CIBA asked for it. They called them Hutu or Tutsi figurants (or ‘bit players’). Although this did not result in exclusion or ethnic tensions within associations, it does indicate that their multi-ethnic character was to some extent artificial, in order to meet CIBA criteria, and did not represent the actual situation on the ground. Moreover (as highlighted above), the potential ‘bridging’ effect of these associations was not sustainable: the anti-erosion associations ceased to exist during the project, while a considerable number of rice association ceased to exist after the project.

LITA on Tangara colline: associations of the elite

The second case study is the intervention of LITA on Tangara colline. Of the three studied collines, Tangara was most clearly a CNDD-FDD stronghold. In Tangara ethnic tensions were not as prevalent as in Burara, and the 2005 elections presented a more ‘classical’ political party struggle between Frodebu and CNDD-FDD. CNDD-FDD won these elections with a clear majority, which meant that ‘the party’ was omnipresent and had an evident militant character.

LITA's project in Tangara was set up around the construction of le hangar, a storage facility for agricultural production. It was managed by a committee representing local community associations in Tangara, and was responsible for the core activity of the LITA project: a seed credit system with local agricultural producers. LITA asked the local population to form associations, after which they received a seed credit of beans, peanuts etc. Afterwards, the associations reimbursed the credit to the committee for the hangar, which used it to supply other groups.

LITA was the Consortium member which focused most on participatory group-oriented processes, and was a strong advocate of an ‘associational’ approach: associations were identified as the main end-beneficiaries, storehouses were opportunities to create local structures of cooperation and collaboration, and great efforts were made to support these participatory processes. The end goal of the project was the participatory process rather than technical end results. As a consequence, the staff composition was the opposite of CIBA: social workers outnumbered technical staff, and local NGO staff was less visible in the field, as rigorous technical supervision by NGO staff was less necessary.

In terms of the intervention strategy, this translated into a more bottom-up laisser-faire approach in introducing community associations. In contrast to CIBA, LITA used a relatively open-ended approach for the creation and development of community associations. Their use was not linked to specific technical end results: LITA only required the associations to be agricultural, multi-ethnic and formalised (registration, statutes and so on). Apart from these requirements, all community members could create or join an association and each association could define its own goals and activities. To motivate the population, LITA offered the associations seed credits and the storage facilities in the hangar.

In other words, unlike CIBA, LITA went beyond a technical approach, and left more room to integrate the associations into community social life. In doing so, an attempt was made to merge an exogenous approach with local dynamic processes of social capital. However, LITA still encountered major problems. The reason was that associations, as new institutional settings, did not develop into ‘building blocks’ of bridging social capital. Instead, they reproduced locally ‘unresponsive’ linking capital (Putnam Reference Putnam2004: 669), manifest in personalised patron–client ties. This began with the problem of elite capture.

Elite capture of community associations had some history on Tangara colline. People explicitly referred to the breaking up of the ‘mother of all associations’, Twiyungure, to explain where and when this problem began. The Twiyungure association started in an IDP site near Rugari in Muyinga province, to which Tutsi from Tangara fled after the 1993 violence. When the security situation improved, these Tutsi returned to a resettlement camp in Tangara, and brought their association with them. At that time, it was the only association in Tangara colline, and was supported by WFP. It was subsequently audited and found to be mismanaged, before the president disappeared with the association's funds. Instead of staying unified in one association, several members of the committee of Twiyungure, who belonged to the local elite, took advantage of the situation and formed their own association, motivated by a possible collaboration with the WFP. The ‘associational logic’ in accessing aid was therefore already understood by the local elite. In addition, with the arrival of LITA's livelihood project access to project benefits was made conditional on membership in associations. As such, the local elite applied the same ‘associational logic’, many of them becoming presidents of their own newly created associations to access external aid. The strong position of the local CNDD-FDD political elite and administration was another factor which enhanced elite capture.

In the case of the LITA project, most of the associations were captured by le club d'hangar or ‘the storehouse committee’: a small group of local petty traders and members of the local administration under the leadership of the chef de colline. This committee centralised the control and supervision of project activities; in particular, all LITA seed credits were channelled through the storehouse. The powerful intermediary role of the club was based on the fact that committee members played two different roles: they not only distributed the seed credits among the associations but also received seed credits as presidents of their own associations. Therefore, the local elite occupied most leadership positions in both the associations and the storehouse committee.

This problem of elite capture had two main consequences. First of all, most associations had problems of internal mismanagement. The democratic character of the associations was minimal: presidents rarely accepted criticism, there were no information meetings, members were not informed about the financial situation, and presidents made financial transactions without any control. There were many reports of corruption and embezzlement of project funds. The question then is: why did ordinary members still accept these associations? Some associations did cease to exist, but most carried on. The reasoning of the members reflected the ‘pareto improvement’ argument of Platteau (Reference Platteau2004) in his work on elite capture in West Africa. They preferred unfair practices within their associations while still receiving some aid, rather than quitting the associations and losing all hope, as membership of the association still yielded a pareto improvement (ibid.: 227). Both parties were also dependent on each other: the elite needed farmers to populate their associations, farmers needed a member of the elite with links to le club d'hangar.

A second consequence of elite capture of the community associations was its impact on the broader community. The associations not only created internal problems, but also pitted different groups against one another within the community. Associations worked exclusively rather than inclusively. First, their unparalleled splitting and multiplication created jealousy and tensions between them. Membership was kept to a minimum in order to maximise profits. A teacher (Tangara 2007 int.) of the local primary school said:

As a consequence, new associations popped up at regular intervals, increasing competition between them to access external aid. In this respect, the situation in Tangara reflects an inverse evolution – the opposite – of the assumed linear progression of community associations. Instead of opening up to new members or seeking synergy with other associations (transforming bonding into bridging social capital), the multiplication of associations further fragmented the social arena in Tangara commune (creating more of the same bonding social capital). Second, a vast majority of poor people who simply could not afford to pay the membership fees or were not ‘invited’ by the elite to join an association felt excluded from the LITA project, as they received no benefits at all. In their perception, the associational strategy of LITA deprived them of fair and equal development opportunities. It was clear to them that if LITA wanted to include everybody, this could not be achieved via associations.

In sum, community associations did succeed in syncretising with local endowments of social capital. Unfortunately, they reproduced ‘unresponsive’ linking social capital (Putnam Reference Putnam2004: 669; Titeca & Vervisch Reference Titeca and Vervisch2008). On the one hand, the internal organisation of many associations mirrored existing vertical relations between patrons (president/committee) and clients (ordinary members), resulting in mismanagement and a lack of accountability and participatory decision-making processes. On the other, and in line with the ‘critical’ civil society literature, associations reproduced existing power structures within the community, and intensified feelings of exclusion, because of elite capture.

This in turn questions the ethnic focus when promoting bridging community associations. As the Burundian conflict is analysed as an ethnic conflict, most post-conflict activities focus on reconciling the two main ethnic groups. However, members of these associations in Tangara, when consulted, immediately mentioned that ethnic groups and women were both represented; they were even more proud that Twa were also members of their association (cf. CIBA in Burara). They knew how external actors analysed ‘their’ conflict, and therefore automatically used this discourse. Still, ethnic tensions never caused problems in the associations, and in this way they succeeded in creating bridging capital.

However, the associations did create economic exclusion. Access to project funds through community associations was a privilege of the local elite, while the vast majority of the community was excluded from membership. This exclusionary governance of associations resulted in a concrete example of structural violence, as discriminatory access deprived people of new opportunities to enhance their situation. This triggered tensions within the community, but not along ethnic lines.Footnote ⁹ The LITA case therefore illustrates how a tailored focus on multi-ethnic associations may create a ‘false sense’ of inclusiveness on the ground (the multi-ethnic and bridging character of the associations). It may also neglect or reproduce a more general, deep-seated problem of social exclusion (the exclusionary and ‘unresponsive’ linking character of the associations), because it produces economic, rather than ethnic, exclusion.

CLC on Cumba colline – a hillside without associations

The third case study is the intervention of CLC on Cumba colline. At the political level, the 2005 elections also brought a clear victory for the CNDD-FDD in Cumba: the party controlled almost all administrative positions. But in contrast to Tangara, at the time of research the party was not militant at all, and people dissociated themselves from it and mistrusted party members. The main reason was the general mistrust and fear of the ruling CNDD-FDD party, after the killing of thirty-one civilians in the region by the army and Service National de Renseignement (SNR), because of alleged collaboration with the FNL rebels (HRW 2006: 18). Therefore, the population's support and enthusiasm for the CNDD-FDD party rapidly decreased after the elections, and local administration did not hold a strong position on Cumba colline.

All three NGOs used the Livelihood Consortium to switch to a more long-term development logic, by rolling out sustainable household livelihood strategies. However, CLC on Cumba colline most clearly translated this shift into a reduction in the distribution of material resources: in contrast to CIBA's ‘technical’ and LITA's ‘associational’ approach, CLC focused on changing individual behaviour and attitudes, and enhancing personal knowledge, capacities and skills. CLC thus opted for a close and constant follow-up of ‘lead farmers’ by its encadreurs de base, who are people living among or close to the target population, and who function as the contact between the population and provincial CLC staff. In addition to the political difficulties with the CNDD-FDD party, representing an absent and unpopular local administration, this focus on individual households and the strong position of the encadreurs de base also left little room for local administration to cooperate in the CLC project. CLC's focus on ‘individual change’ explains their choice of working with individual households, instead of associations.

The CLC intervention therefore gave us the opportunity to observe what happened when no external agency was raising awareness and motivating the population to create new associations. As might be expected, almost no associations were found in Cumba. As highlighted above, the main incentive to create new associations in Burara (CIBA) and even more in Tangara (LITA) was the project aid accessed through these associations. The CLC intervention confirms this general observation, by turning the argument upside down: precisely because external intervention did not channel project aid through associations, associations were almost non-existent on Cumba colline. According to the local population, the raison d'être of associations is to receive aid. As one man explained: ‘There are not many associations because even the associations that exist did not succeed once to get help from a NGO. That's why people abandon their associations and no new associations are created’ (Focus group, Cumba 2007 int.). People also explicitly referred to the importance of patrons who could act as gatekeepers: there was a lack of people willing to introduce the associations to external agencies in order to receive aid. As there was no incentive for the elite to interfere with the creation of associations, the local elite was much less involved in associational life and no elite capture was mentioned by the population. By contrast, the population of Cumba blamed the local administration for a lack of initiative because they did not approach NGOs to gain support for associations, something which also reflected the more general deterioration of relations between the CNDD-FDD administration and the local population.

The CLC intervention thus shows how reasons, incentives and motivations to create associations on Cumba colline had become structurally adapted to the presence of external assistance. CLC was even criticised because it did not work through associations like other international organisations. This reasoning was based on various experiences. First, CLC had worked with associations in the past, and the population urgently asked the organisation to start supporting associations again. However, this ‘opportunistic’ intention to access project aid through associations was exactly the reason why CLC stopped their support. Second, the population compared their situation with other hillsides: ‘Here we don't get [aid]. We hear that associations on other hillsides are supported and get aid. Why don't the NGOs come here to support the associations? If they come, a lot of associations will be formed’ (Focus group, Cumba 2007 int.). Lastly, people compared the CLC project with other projects, for example the Agricultural Rehabilitation and Support Project (PRASAB) of the World Bank that was in its initial phase. This nationwide project was based on a grant system. Local associations could submit a grant proposal to a provincial committee to implement agricultural or livestock activities. For the population, this was a ‘promising’ project, as associations would be linked to a concrete incentive, so creating the association was clearly worthwhile. For the population of Tangara it was ultimately clear that no associations would grow without external interest; this was confirmed by the small number of ‘only’ three associations on Cumba colline.

The Cumba case illustrates the possible long-term effects of the general policy of external actors in Burundi to support the creation of associations. The externally driven character of the proliferation of associations in Burundi has already been noted (Sebudandi & Nduwayo Reference Sebudandi and Nduwayo2002). As a consequence, the widespread perception of associations as ‘gateways’ to aid highlights the general vulnerability of externally driven associations. It reflects a situation in which sustained exposure to exogenous new ‘institutional settings’ (external agencies promoting community associations) can change endogenous local dynamics of social capital building. In the long term this may reduce local dynamics of social organisation, as it becomes less likely that community associations will be nurtured without external incentives in the future (Crowley et al. Reference Crowley2005: 15).

* * *

All three case studies highlight different problems in the constructability of social capital, and more particularly, the failure of the three interventions to initiate a fragile process of ‘institutional syncretism’. First, CIBA illustrates how an exogenous process of social capital building can remain exogenous. A ‘strong’ social engineering approach led to ‘alien’ associations, i.e. externally introduced ‘institutional settings’ which never merged with locally produced social capital endowments. With its focus on technical end results, CIBA gave the local population few opportunities to adapt their social life to the newly introduced institutional settings. The associations never disassociated from the project, functioning as technical subcontractors.

Second, LITA illustrates how an exogenous process of social capital building can merge with endogenous social capital endowments; but this does not automatically bring positive effects: their ‘soft’ social engineering approach (bottom-up and open-ended) led to ‘uncivic’ associations, as they adapted towards locally produced ‘unresponsive’ forms of linking capital. Through LITA's soft approach, room was left for the local population to merge the new institutional setting of the associations with local norms and rules of social organisation. Because of elite capture, the associations became too embedded in existing vertical patron–client ties.

Finally, CLC demonstrates how endogenous social capital endowments may be adapted towards exogenous processes of social capital building. It shows how a policy of external support in creating organisations for the poor affects the broader environment and diminishes local dynamics of the poor in organising themselves (Crowley et al. Reference Crowley2005: 17).

Our critique of ‘social engineering’ (or the ‘technical fix’) of community associations does not mean that external interventions are automatically flawed. Any external development intervention is de facto about ‘social engineering’. In line with the ‘critical’ literature on civil society, the above cases rather show that using community associations as ‘universal blueprints’ leads to ‘institutional monocropping’: the replacement of local social dynamics and practices with those of more advanced societies, without tailoring these to the local context (Evans Reference Evans2004). Equally, the cases show that existing endowments of social capital need to be approached with great care, as they can have a profoundly negative effect. They should not automatically be preferred above exogenous ‘institutional settings’ simply because of their ‘local’ character. Institutional syncretism therefore is an extremely fragile exercise, which is summarised in this section.

On the one hand, there is the problem of overly ambitious external initiatives of ‘social engineering’ of social capital and community associations. First, development practice to a great extent is driven by the culture of the organisations and their attempts to present their actions as examples of authorised policy (Mosse Reference Mosse2004). The aims of all three Consortium partners in playing their role of ‘institutional entrepreneur’ and strengthening multi-ethnic social cohesion reflects their attempt to present their work as examples of contemporary post-conflict reconstruction policy. This policy prescribes the integration of economic and social reconstruction, i.e. the restoration of social cohesion and local conflict resolution capacity while bringing development (Duffield Reference Duffield2002). In addition, the organisational cultures and philosophies of the different Consortium partners substantially guided the practical realisation of this objective: CIBA's ‘technical’ focus explained the rather technical finality of most of the associations on Burara colline, while LITA's focus on participatory group processes explained their clear ‘associational’ approach, and CLC's emphasis on ‘individual change’ their choice of working with households and not with associations. Second, the three cases highlight the importance of the ‘right’ tangible incentives and the functioning of ‘incentivised voluntarism’ (Gillinson Reference Gillinson2004: 6). All three cases have at least one issue in common: community associations were primarily perceived as ‘gateways’ to aid: people's voluntarism was incentivised, as the project aid redirected members' motivations. The cases generally indicate that the incentive should be designed as an initiator but not as the most important motivator of participants' behaviour; the external incentive has to be substituted by an internal incentive (which should preferably be more attractive). Development agencies need to tailor their incentives, adapting them to the local situation. At present, they are driven more by the need to disburse funds and distribute resources, rather than focusing on the local dynamics of associational life, i.e. they need to act in keeping with policy and donor logics, more than with the needs on the ground (cf. de Sardan Reference de Sardan2005: 141).

On the other hand, the above cases show how even within the same region, three external interventions generate very different results on three different collines; and therefore show the importance of detailed contextual knowledge of prior social capital endowments. On each of the three collines there was a different interplay between the intervention and the local political settings. In Tangara, the strong position of the CNDD-FDD party went hand in hand with the elite capture of community associations during project implementation. In Cumba, the blaming of the local administration for not looking for donor aid to create local associations partly reflected the more general friction between the local population and the CNDD-FDD party. In Burara, next to the technical focus and close follow-up of the NGO, the strong (geographical) ethnic segregation revealed the donor-driven and ‘artificial’ sense of inclusiveness: in reality most associations were Tutsi (centre) or Hutu (hills) dominated, and many ceased to exist during or after the end of the project. In general, the brief historical summary shows why vibrant community associations are not self-evident in Burundian society. It highlights how: (a) patron–client ties play an important role; (b) social organisation is weak beyond the family; and (c) the civil war shattered social cohesion and organisation. It enables us to better understand why associations did not evolve as expected. This is not to say that associations, as new ‘institutional settings’, cannot become a promising element in the further development of Burundian society, as this would follow a radical ‘endowment’ view, in which societies are rigid and can only change over longer periods of time. Still, it reminds us not to underestimate the thriving force of history.

In summary, if one accepts that development projects within a context of ‘liberal peace’ are always about ‘social engineering’, the above cases indicate that the real challenge lies in developing new ‘institutional settings’ that are ‘soft’ enough to adapt to prior social capital endowments, but ‘strong’ enough to change or reinforce these endowments to bring positive results. In sum, the task is to avoid both the absence (cf. the CIBA case) and ‘excess’ (cf. the LITA case) of institutional syncretism (Galvan Reference Galvan2007: 70).

With regard to post-conflict reconstruction, the three described case studies question the ability of the ‘liberal peace’ model to produce the desirable outcomes desired by donors and the international community, precisely because this fails to reach such institutional syncretism and is instead caught in ‘universal blueprints’ of institution building – or ‘institutional monocropping’ (Evans Reference Evans2004) – to favour processes of democratisation and post-conflict reconstruction.

The general idea of multi-ethnic ‘bridging’ associations sheds some light on the matter. First, our micro-level analysis of the difficulties of building bridging associations in three communes demonstrates how the development of outward instead of inward legitimacy generates substantial problems: legitimacy is developed towards external donors rather than the ‘inward’ constituency. Concretely, the idea of ‘ethnic inclusiveness’ is developed more for the donor audience than for the ethnic groups as such: in the case of CIBA, the multi-ethnic character of the newly created associations was unrelated to the actual ethnically segregated and sometimes tense situation in the community. These processes of external legitimacy also created problems of accountability and participation within the associations: in the LITA case, the ‘false sense’ of inclusiveness even worked as a smokescreen, as these multi-ethnic ‘bridging’ associations reproduced the more fundamental problem of ‘unresponsive’ linking social capital, resulting in economic instead of ethnic exclusion.

Second, in line with the critique on community associations within the liberal peace model, it has become clear that community associations are no ‘universal blueprints’ to create social capital (e.g. Molenaers Reference Molenaers, Hooghe and Stolle2003), but are deeply embedded in their social and political context, reproducing existing tensions and power structures. Uvin (Reference Uvin2005) argues that the introduction of such institutions at a national level in Burundi has installed a liberal-democratic façade, without substance or content, as the existing system and power structures undermine the imported system from the inside out. The LITA case describes exactly the same process on a micro-scale, with a disconnection between the structural and cognitive components of social capital (Molenaers Reference Molenaers, Hooghe and Stolle2003: 114). Moreover, although a multitude of bridging community associations were present (structural component, as a liberal-democratic ‘façade’), these were disconnected from their horizontality and embedded trust (cognitive component), and adjusted towards the existing ‘system’ of vertical power relations. As such, the idea of bridging associations in the Burundian context presents us with a micro-manifestation of one of the main shortcomings of this ‘liberal peace’ model: its incapacity and unwillingness to enter into dialogue and syncretise with local institutional settings. In doing so, international donors therefore fail to acknowledge to existence of local resources and make local development dependent on international presence (Belloni Reference Belloni2007: 111).