Building Democracy from Below: Lessons from Western Uganda

Abstract How to achieve democratisation in the neo-patrimonial and agrarian environments that predominate in sub-Saharan Africa continues to present a challenge for both development theory and practice. Drawing on intensive fieldwork in Western Uganda, this paper argues that Charles Tilly’s ‘democratisation as process’ provides us with the framework required to explain the ways in which particular kinds of association can advance democratisation from below. Moving beyond the current focus on how elite-bargaining and certain associational forms may contribute to liberal forms of democracy, this approach helps identify the intermediate mechanisms involved in building democracy from below, including the significance of challenging categorical inequalities, notably through the role of producer groups, and of building trust networks, cross-class alliances and synergistic relations between civil and political society. The evidence and mode of analysis deployed here help suggest alternative routes for supporting local efforts to build democracy from below in sub-Saharan Africa.


Introduction
Despite the widespread adoption of democratic institutions and increased economic growth in much of sub-Saharan Africa, the progress of democratisation and development remains heavily constrained by the persistence of neo-patrimonial political systems and agrarian economies. This realisation has undermined the optimism within international development that democracy could be built in such contexts via external interventions and encouraged greater recognition of the critical role played by politics and power in shaping development trajectories. In response, mainstream development thinkers have focused on how processes of democratisation and development in developing countries are shaped by elite-level bargaining (Acemoglu & Robinson, 2012;North, Wallis, & Weingast, 2009). Alternative analyses of similar processes, however, place more emphasis on the broader character of state-society relations, inequality and the role of subordinate forms of political agency (Evans & Heller, 2015;Sandbrook, Edelman, Heller, & Teichman, 2007;Tilly, 2007). This latter strand of thinking is complemented by the emergence of a more politicised understanding of the local politics of democratisation (Corbridge, Williams, Srivastava, & Véron, 2005;Harriss, Stokke, & Tornquist, 2004), and of which forms of civil society are most likely to advance the interests of poor and marginalised citizens (Banks, Hulme, & Edwards, 2015).
With this paper, we seek to contribute to this second line of thinking, by using Charles Tilly's (2007) conceptualisation of 'democratisation as process' to examine the experiences of two different civil society organisations (CSOs), a professional development agency and a farmer cooperative, located in the Rwenzori sub-region of Uganda. Applying Tilly's concepts to the actual experiences of local CSOs not only helps bridge a significant evidence gap regarding which organisational forms can contribute to democratisation in particular contexts, but also helps challenge the mainstream assumption that supporting western-style CSOs institutions in poor countries will inevitably lead to liberal forms of democracy emerging. This approach fails to consider the actual mechanisms through which democratisation from below may occur or the varied forms that democratisation might take. Grounded in historical analysis of how democratisation has actually taken place in conditions of socioeconomic inequality and contested state-society relations in different countries around the world, Tilly's framework is particularly relevant to understanding the mechanisms through which democratisation may actually emerge in developing countries.
The paper proceeds by setting out the nature of Tilly's framework in more depth, before also reviewing the literature on the role of NGOs and civil society within processes of democratisation. It then explores the current context for democratisation in Uganda, and in the Rwenzori region in particular, before introducing the two case study organisations and discussing their achievements. The final section revisits what Tilly's framework has enabled us to see regarding the broader question of how different types of local organisation can contribute to democratisation in semi-authoritarian, neo-patrimonial and predominantly agrarian contexts.

Democratisation, inclusion and neo-patrimonialism
Much recent analysis of how democracy and development unfold has focused on structural conditions regarding the balance of power between contending elite factions and long-term institutional change (Acemoglu & Robinson, 2012;North et al., 2009). Alternative readings, which focus more on social than liberal versions of democracy, suggest that although macro-level processes of state formation and capitalist transitions are critical, the forms that they take and developmental possibilities that they create are also shaped by popular agency, including at sub-national levels. Sandbrook et al. (2007) establish the conditions under which social democracy has historically emerged in the global South. They chart how capitalist transformation, through the commercialisation of agriculture, helped shift the balance of power away from landlords towards smallholder farmers in ways that helped establish a 'coherent and effective state with some autonomy from dominant classes', that was capable therefore of promoting development (2007, pp. 30-31). A 'robust civil society' was required here, particularly in the form of smallholder farmers and peasant proprietors whose increased exposure to market risk incentivised them to organise collectively and put pressure on the state to provide the public goods they required for productive and protective purposes (Sandbrook et al., 2007, p. 198). Such mobilisation helped foster the emergence of parties, movements and leaders that come to form public spheres within which subordinate groups can 'organise and demand political and social rights' enhancing the likelihood that 'distributional conflicts' can be 'peacefully mediated' (Sandbrook et al., 2007, p. 208). Tilly's (2007) work is similarly grounded in an understanding of how conditions of inequality between different social groups shape the character of the state-citizen relationship and the processes of democratisation and de-democratisation that result from these ever-shifting socio-economic relations. Tilly conceives of democratisation in terms of the degree to which a breadth of citizens have equal access to binding forms of consultation with the state, and are protected from arbitrary state action (2007, p. 13). His analysis of how democratisation advances hinges on three moves: the extent to which trust networks are integrated into public affairs and policies; how far public policy is 'insulated from categorical inequality'; and the degree to which major power centres have autonomy from public politics (2007, p. 23).
Trust networks are those based on kinship, religion, or informal trading relations for example. In order to achieve breadth, equality and binding consultation for members, such networks must become integrated within the mainstream operations of the state including through taxation and participation in public services and politics, a process that requires 'a weakening of the mediating power of patrons' in neo-patrimonial environments (2007, p. 94). Relations between trust networks and public politics are critical because 'they govern the possibility of contingent consent', whereby citizens (and the trust networks they are part of) accept to be ruled only insofar as rulers 'perform in accordance with citizens' expressed collective will', and enable 'the effective translation of citizens' expressed collective will into state action (2007, p. 94).
Tilly defines categorical inequality as 'organised differences in advantages by gender, race, nationality, ethnicity, religion, community and similar classification systems', including class (2007, p. 111). Categorical inequality is created and perpetuated by exploitation and opportunity hoarding within capitalist systems, including by those controlling large amounts of land vis-à-vis smallholders or subsistence farmers, or by men in relation to household assets and decision-making. Democratisation is therefore dependent upon processes emerging which insulate policy-making from these inequalities and achieve a degree of 'material equalisation across categories ' (2007, p. 115). This might include promoting a more equal distribution of access to and control over productive resources, the 'dissolution of state controls that support current unequal relations among social categories' (such as restrictions on land holdings); an increase in the middle-peasantry following increased demand for peasant products; the introduction of 'procedural devices' such as secret ballots and free or equal access to the media; and, importantly, the 'formation of politically active coalitions and associations for crosscutting categorical inequality' and 'increases of political participation, rights, or obligations that cut across social categories ' (2007, p. 115).
The third key pillar within Tilly's analysis is 'a reduction in autonomous power clusters'such as powerful ethnic factions or business elites and those with 'coercive power' such as military cadres or warlords. Reducing their power 'subordinates states to public politics and facilitates popular influence over public politics ' (2007, p. 139), whilst also suggesting the need for states to develop higher levels of autonomy and the capacity to undertake binding negotiations with citizens from an impartial standpoint. Tilly suggests this can be achieved through 'broadening political participation', and the 'equalisation of access to non-state political resources and opportunities'. Contrary to some recent theorising, then (Khan, 2010;North et al., 2009), achieving change is not entirely dependent on the assent of the existing ruling coalition, but also on pressures from below (Tilly, 2007, p. 139).

What role for civil society, NGOs and membership organisations?
Tilly's ideas about the mutually formative relationship between social inequality, the distribution of power, and democratisation resonates with recent advances within the literature on how NGOs and civil society can contribute (Banks et al., 2015). Building on a long-established school of thinking (including Bebbington, 1997;Korten, 1990), Banks et al.'s (2015) synthesis suggests that international and local NGOs can achieve better results for low-income groups by focusing their energies on supporting the emergence and sustainability of solidaristic membership organisations, facilitating a process whereby such organisations become networked, affiliated or federated, and acting as brokers of technical assistance and of relationships that bridge the local/national/global divide. The evidence we present below concurs with this conclusion while also emphasising the need for complementary elite-led strategies which exploit Tilly's class and identity-based trust networks to create space for such membership organisations to gain a voice and audience beyond the local level.
In terms of which associational forms can contribute to democratisation from below in agrarian societies, there is a surprising lack of systematic evidence about the role that the kinds of smallholder farmers associations that Sandbrook et al. (2007) draw attention to can play in sub-Saharan Africa. 1 The wider literature, taking in examples from Asia and Latin America, suggests that such groups have a greater chance of promoting their interests when they experience a shared ethnic, class or gender identity, a shared external oppressor (such as state-based marginalisation which can promote internal cohesion), are organised around livelihoods or savings-based finance, and have a federated structure which creates mechanisms for influence at different levels of market and state operation (Kabeer, Mahmud, & Castro, 2010;Thorp, Stewart, & Heyer, 2005). These findings resonate strongly with Tilly's focus on working through trust networks; resisting arbitrary state action or the operations of autonomous power clusters; and tackling categorical inequality through greater control over productive resources and pushing for more binding forms of consultation with the state.
Typical organisational constraints raised across this literature include donor dependency; the valuing of individual over group interests, including using groups as a front for patron/client based resource distribution; and the exclusion of the poorest (Kabeer et al., 2010;Thorp et al., 2005). NGOs that are able to avoid creating relationships of dependency while creating channels for macro-level advocacy, and which have an explicit ideological and operational commitment to empowerment or social justice (Kabeer et al., 2010;Thorp et al., 2005), to tackling categorical inequality and rebalancing relations with the state, have been considered critical actors within democracy and citizenship-building processes (Edwards, 2009;Michael, 2004). Such NGOs have provided organisational development support including with the formation of federations of local associations (Ferreira & Roque, 2010;Thorp et al., 2005); linked local cooperatives to finance, technical expertise and equipment (Thorp et al., 2005); and built the capacity of members to represent their interests with state and market actors, or acted as their advocates (Ferreira & Roque, 2010). Here we reflect on the relevance of these lessons for semi-authoritarian and neopatrimonial agrarian contexts such as Uganda, by using Tilly's framework to interrogate the ways in which different types of local organisation are engaged in promoting the mechanisms through which democratisation occurs, and the ways in which they might be supported in doing so.

Methodological approach
The analysis presented here revisits the results of 12 months of field research into the politics of civil society efforts to promote democracy and development in the Rwenzori sub-region of Western Uganda, conducted by the first author between 2009 and 2011, and is further informed by interviews conducted by the second author in June and September 2014 in Kampala; as well as informal conversations between the first author and key informants in Uganda in 2016. This research found that two particular CSOs in the region had achieved a degree of success in cultivating supportive conditions for democratisation in the region, including (see also Törnquist, Webster, & Stokke, 2009) through supporting the realisation of citizenship rights, building political capabilities amongst disadvantaged groups, and encouraging elites to be more open to popular involvement in decision-making processes (King, 2015). As such, these two case study organisations, each of which sought to advance the political, social and economic inclusion of smallholder farmers within development processes, offer 'theoretical exemplars' (Yin, 2003) that can be used to test and refine Tilly's (2007) propositions regarding the causal mechanisms through which democratisation from below may emerge. That each organisation represents different associational forms, the first being a professionalised research and development NGO (RD) staffed by educated elites and the other a community-led farmer cooperative, rules out the possibility of undertaking a structured comparison between cases; rather, we focus on what George and Bennett (2005, p. 108) identify as a major contribution of case-study research to theory development, namely the use of within-case methods such as process tracing, in this instance to use both cases to examine whether the causal mechanisms that Tilly posits as important are evident here.
The methodology of the original research conducted between 2009 and 2011 sought to capture the interrelationships between structural conditions and localised interventions, with data generation involving a multi-levelled approach that engaged diverse actors operating at a range of societal scales. Data generation included a review of each organisation's internal documents, 45 semistructured focus group discussions, 139 semi-structured interviews with a range of state, civil society and political actors from the local to the national level (including politicians, NGO and donor staff and community members), and 48 sets of notes made during semi-structured observation of organisational practice. Data was coded using qualitative thematic analysis and then Building democracy from below 1587 triangulated to draw out synergies and patterns in support of argument development and to ensure validity (Miles & Huberman, 1994).

The context for building democracy from below in Uganda
Recent analyses of Ugandan political economy suggest that the shifting dynamics of power relations amongst political elites, and between elites and more popular actors, have undermined the initial gains made in both democracy and development following the National Resistance Movement's (NRM) arrival into power in 1986 (Golooba-Mutebi & Hickey, 2013;Kjaer, 2015). Having built an inclusive ruling coalition that helped ensure political stability, with the notable exception of the conflict-affected north, President Museveni faced little opposition in promoting a series of economic and political reforms that left Uganda lauded as a success story in terms of securing impressive levels of growth, poverty reduction and also citizenship involvement through it's 'no-party' system of participatory democracy. However, by the mid-late 1990s it had become clear that the driving logic of the NRM government was on maintaining itself in power rather than promoting a wider project of transformation, as was evident in the misuse of local government structures (see below) and growing neopatrimonial-use of public resources for personal and political gain (Kasfir, 2012). Under a populist leader with authoritarian tendencies (Tripp, 2010), both economic and political institutions remain in an informalised stateand remain very far from offering equal access to a breadth of citizens (Tilly, 2007). With over 70 per cent of the workforce in the agricultural sector and heavily dependent on primary commodities, Uganda has not achieved the levels of structural transformation (Haussmann, Cunningham, Matovu, Osire, & Wyett, 2014) associated with significant shifts in class configurations (Sandbrook et al., 2007). With official tax revenue flat lining at around 13 per cent of GDP for the past decade, the abolition of income tax for electoral gain in 2005 and the discovery of commercial quantities of oil in 2006, the government is not held to account by the need to negotiate with citizens for resources. State-business relations are largely collusive in nature and military leaders retain a significant formal and informal presence within government decision-making (Golooba-Mutebi & Hickey, 2013), revealing little sign of a reduction in 'autonomous power clusters' (Tilly, 2007, p. 139).

Binding forms of consultation with the state?
The use of decentralisation to extend the power of the NRM at local levels rather than accord genuine power to the grassroots had become apparent by the mid-1990s. The capacity of local government to act as a vehicle of either democratisation or service delivery was further undermined by the policy of 'districtisation' during the 2000s, which saw the number of districts double from 56 to 112 in response to a mixture of local ethno-territorial demands and electioneering (Green, 2008). Although Uganda's decentralised system does offer opportunities for grassroots participation, such as bottom-up planning processes beginning at the village council level, these forms of consultation are not binding (Tilly, 2007), and the capacity of local governments to deliver services effectively and accountably is undermined by the fact that resource allocation remains highly centralised, investment in human resource development remains low, and rent-seeking is widespread (Green, 2008;King, 2015).

Categorical inequality, segregated trust networks and local associationalism
The legacy of conflict and repression instilled during the Amin and Obote II regimes of the 1970s and early 1980s discouraged many rural citizens from engaging with the state, and political participation is further undermined by hierarchical social structures and clientelist relations between poorly educated low-income rural citizens and local politicians and civil servants protected by senior patrons (Francis & James, 2003;King, 2015). The hardships of subsistence and small-scale agriculture also leave rural communities, and particularly women, with little time to participate (Lakwo, 2009). A customary land tenure system continues to operate in much of rural Uganda leaving farmers without land titles vulnerable to land-grabbing by political and economic elites and women particularly vulnerable to claims by family members within a patrilineal inheritance system.
Prior to the economic and political collapse of the 1970s and 1980s, rural cooperatives had demonstrated potential to act as popular power bases in Uganda (Bunker, 1983). Brett explains how 'in the late 1940s, the Uganda African Farmers Union claimed thousands of members and was the most influential indigenous political organisation', and how primary cooperatives were the 'largest economic units in rural areas operating in every district ' (1992, p. 54). However, their potential was rarely realised due to the limited human capital available to run them effectively and 'the conditions imposed on them by the state after 1971' (1992, p. 55): economic collapse and political repression under Amin catalysed the growth of an informal economy across rural Uganda. Farmers began selling coffee privately, smuggling across local borders was rife and many producers switched from cash to food crops to focus on local informal markets. At the same time, private and cooperative forms of service provision emerged driven by local communities such as in the form of parent teacher associations (PTAs) running schools, churches providing health and education services and 'a multitude of local NGOs and cooperative groups […] formed, often by women, for income generation and social improvement of various kinds' increasingly supported by foreign NGOs (Brett, 1992, p. 20).
By the time the NRM came to power, agricultural cooperation had been tainted by mismanagement and economic collapse. Cooperative societies which had survived the Amin era were unprepared for the transition to private ownership ushered in under structural adjustment and Museveni's further liberalising reforms of the 1990s, both of which undermined collective action within the main smallholder sectors (Brett, 1998;ILO, 2008). However, there are signs that recent years have seen something of a revival amongst co-operatives in Uganda (Okello, 2013), despite continued problems around mistrust and an incoherent regulatory environment (ILO, 2008).

NGOs, civil society and protection from arbitrary state action
Development NGOs have not significantly filled the gap left by government in Uganda in terms of building organisational capacity among smallholders, constrained as they are by 'erratic development interventions with donor-dependent project circles' (Okello, 2013, p. 9). The early signs that certain indigenous organisations, particularly the women's movement, would challenge patronage forms of politics in Uganda were eroded over the 1990s (Tripp, 2010). The growth of populist policy reforms over the same period tended to undermine local associational oversight, as with the displacement of parent teacher associations following the introduction of universal primary education in 1997 (Tripp, 2010), thus further reducing the extent to which Tilly's 'trust networks' are integrated within state services. The space for more politicised civil society approaches has if anything become more restricted since the return of multi-partyism in 2005, which increased the incentives for government to repress dissenting voices which may be construed as 'oppositional'.

The Rwenzori Sub-Region 2
Rwenzori has been the site of repeated ethnic conflict, most prominently between the Toro and Rwenzururu kingdoms in the 1960s (Kasfir, 1970), but also concerning a rebellion in the late 1990s led by the Allied Democratic Forces (African Rights, 2001). The main ethnic conflict ended in the 1970s with the granting of district administrations to the Bakonjo and Bamba of the Rwenzori mountains and lowlands, although tensions continue to simmer with an outbreak of ethnically-driven violence as recently as 2016. The conflict was triggered by the Batoro's exploitative indirect rule of other ethnic groups, characterised by discrimination in the provision of health and social services and government positions (Kasfir, 1970) that created a legacy of ethnically-characterised 'categorical inequality'. Museveni initially abolished the traditional kingdoms but these were reinstated in 1992, reviving passions among the Bakonjo for their own Ruwenzuru kingdom (African Rights, 2001) which was finally granted in 2009.

Building democracy from below 1589
The dominance of wealthier social groups over local governance continued into the NRM era, and, although the provision of special leadership positions for young people and women engaged most people in the new system to begin with, local elders suggest that by the mid-1990s development plans were being agreed by chiefs and councillors at parish and sub-county levels. Patronage-based districtisation has been manifest in Rwenzori: three new districts have been carved out of the existing administrative structures since 2009 and new sub-counties created along ethnic lines within these districts.
The sub-region's economy is based mostly around agriculture, including large tea and coffee estates, and livestock. The majority of farmers continue to operate within a customary land tenure system based on patrilineal inheritance and occupation, with men still predominantly in control of household finances and decision-making. With population growth above the national average, land fragmentation through the traditional inheritance system, and an increasing concentration of land in the hands of large companies like the tea estates, there are increasing pressure on land and growing levels of landlessness, particularly among younger generations. Poverty levels in the sub-region are estimated to be above the national average at 30 per cent but vary considerably across the districts with conflict-affected areas suffering disproportionately. The cooperative movement in the Rwenzori sub-region has followed a similar trajectory to the rest of the country: a gradual revival is occurring but many farmers remain distrustful of cooperative endeavour, with corruption and mismanagement of savings remaining widespread.

A research and development NGO (RD)
RD was founded as a research institute in 1996 by a charismatic graduate passionate about peacebuilding and community empowerment in Rwenzori, and soon moved into implementing donorfunded projects. Its official vision has continually evolved but can be summarised as fostering greater economic, social and political inclusion within existing structures and processes rather than systemic change. In 2010, it had a budget of approximately GBP £702 thousand from 10 international donors and 28 members of staff. RD's core donor was (until recently) a Dutch co-financing agency which has been supportive of experimentation and action learning particularly at the sub-regional level. In 2011, the founder director left the organisation to become an NRM MP.
This paper examines three of RD's strategies developed between 2002 and 2011: the provision of rights and governance training and facilitation of sub-county dialogues, support to rural producer and savings groups, and the convening of new deliberative spaces for regional elites. To increase the effectiveness and accountability of local development processes, RD trains village residents, health centre and primary school management committees, councillors and civil servants from village up to sub-county level, about citizenship rights and policy guidelines regarding how local governance processes are supposed to work. RD's community level work is delivered in partnership with teams of trained community activists called 'process facilitators'. RD also convenes civil servants and service providers, local councillors and health and school management committee members at parish and subcounty level for quarterly problem-solving dialogues. 3 Staff and process facilitators use pictorial participatory action learning tools for training which RD developed in partnership with a British consultant in 2002.
In terms of capacity building, RD promotes socio-economic empowerment among farmers through savings and credit schemes and sustainable production training; grant or asset-giving; and organisational development advice. Staff initially began training 'model farmers' to support the development of small producer groups in their own communities, with some farmers then becoming process facilitators. Later, staff and process facilitators began organising rural savings and producer groups into associations for collective storage and marketing of their produce or for micro-finance. By 2011, RD estimated they had worked with over 500 farmer groups across the region. Further capacity building has involved support for community-based organisations and fledgling advocacy NGOs, linking them to international donors, and convening regional reflection workshops for CSOs on how to 'make development work'. In 2012, these workshops culminated in participating organisations adopting a 'regional development framework', an agreed set of guiding principles for making CSO interventions people-centred.
Finally, RD also convenes annual leadership retreats for the sub-region's cultural, political and social elites, enabling informal discussion and networking, and reflection on the priorities, challenges and opportunities for development in the sub-region. In partnership with a local university, it has established a regional think tank to generate evidence in support of better state and civil society development planning which works through regional stakeholder forums for community activists, CSO workers, political leaders, and civil servants who identify research priorities, deliberate findings and plan actions. A regional Leaders' Group comprising all district council chairpersons and two local MPs is tasked with ensuring that development plans build on priorities identified through think tank research. The leadership retreats and an MPs' Forum is supposed to maintain links to the Leaders' Group and think tank in order to promote regional interests in parliament, thus establishing bridges between civil and political society and alternative channels for sub-regional representation at the national level.

A micro-finance and coffee marketing co-operative
Bukonzo Joint Co-operative Union was established as an association of 11 savings groups in 1999 by a local community leader who received training from an international NGO in Frierian-style community mobilisation. Bukonzo Joint aims to improve 'the standard of living by enabling people to realise their potential and become active partners in the social, economic and political development of Bukonzo County, the Rwenzori region and Uganda at large' (internal documentation). The membership is 83 per cent female, predominantly smallholder farmers growing coffee in the foothills of the Rwenzori mountains. It has a democratic and participatory structure and ethos: each savings group has representation on a parish-level committee which links to Bukonzo Joint's Board and board members rotate every two years to ensure that leadership opportunities are distributed widely amongst members. By 2011, Bukonzo Joint comprised 3887 individuals participating in 201 registered member groups. Many are also members or clients of the coffee marketing association established in 2005 to protect smallholders from a fluctuating market and exploitative middle-men, and which now exports coffee to a buyer in London. Adding clients and members together, Bukonzo Joint has over 5000 local stakeholders including six full-time staff, six training officers, and 42 'training volunteers'.
Bukonzo Joint mobilises smallholders into village-level savings groups through household-tohousehold conscientisation and group formation. From the early 2000s, and with donor support, they have extended their work to some of the most marginalised local households, including training on improved farming methods and financial and group management. Between 2004 and 2007, a British consultant supported members to develop first, a pictorial participatory action learning system, and then, a gender action learning system (GALS), aimed at overcoming both educational and gender inequalities that could undermine trust and accountability dynamics in the household and within the savings groups (Baluku, Mayoux, & Reemer, 2009). Oxfam Novib has supported the documentation of GALS as Bukonzo Joint's core operational methodology for good practice sharing in Uganda and internationally.
In terms of influencing local governance and resource allocation, trainers have encouraged savings group members to participate in village meetings and attend sub-county budget conferences. However, when this failed to secure new relationships or benefits for members, Bukonzo Joint promoted alternative mobilising strategies that enabled group members to circumvent local decision-making processes. This involved generating political leverage beyond the local level by bringing the donor agencies funding their own local government to their sub-counties of operation to learn about their approaches and needs, engaging them subsequently in advocating for their interests within government.
Building democracy from below 1591 6. Challenging categorical inequalities and integrating trust networks 6.1. Increasing access to productive resources and political participation RD's work to increase household income among farmers, especially women, through supporting cooperative forms of saving, production and marketing, has generated mixed results. Staff generally reported low levels of ownership among members leading to poor enforcement of monitoring and accountability mechanisms and mismanagement. Farmers under economic pressure tend to sell their produce individually, undermining the collective bargaining power of associations and reducing the levels of loan capital available within partnered micro-finance associations. Women's participation within groups has remained tokenistic with men dominating decision-making.
RD has achieved more positive gains where efforts to promote economic empowerment have dovetailed with organisational capacity building, particularly through a new community-based agricultural production NGO. This locally-embedded NGO was able to work in a sustained way with farmers living in the same remote area to establish a successful rice marketing cooperative. Members explained how the increased socio-economic mobility they had experienced in response to RD's support had acted as a catalyst for political mobilisation. In at least six cases, group members had encouraged a leader to participate in local councils. Respondents also felt that RD's push for the inclusion of women had contributed to incremental shifts in gender roles, such as women selling cash crops at local markets and engaging in political campaigning. State-farmer relations were also shifting. Farmers were exercising 'contingent consent' by refusing to display signposts from the national agricultural extension service (NAADS) in front of plantations they had developed without state assistance, and leaders of the local support NGO refused to sell coffee seedlings to NAADS officials who were planning to offload them onto local farmers in the dry season.
In 2002, RD also helped to establish a new regional network of model farmers who now assist other households and farmer groups with enterprise development and broker links with larger urban-based NGOs. Trainers within this network and RD's own process facilitators are also now channelling the interests and experiences of smallholder farmers into the regional deliberative processes described above, signifying the emergence of new cross-class regional coalitions.
Bukonzo Joint community mobilisation has resulted in both the amelioration of categorical inequalities and the integration of local trust networks within the formal economy and public politics. Trainers have fostered a strong savings culture within member households and improved farming practices leading to better quality coffee, higher levels of production, better financial management and therefore increased household income and socio-economic mobility. By 2010, the organisation had a loan disbursement of just under 1.9 billion shillings, 4 and that year the marketing society collected 300,988 kg coffee with a market loan value of 1.27 billion shillings. 5 Overall, production has increased from 'about 13 tons in 2005, to more than 220 tons in 2012'. 6 With international support, including from the Fair Trade organisation Twin, Bukonzo Joint now exports Fair Trade Organic Arabica Coffee to countries in four continents.
Bukonzo Joint's economic success has been grounded in the cooperative's gender justice methodology which directly engages with categorical inequality. GALS training combined with savings-based organising has catalysed changes in gender relations in many member households so that women are beginning to benefit from productive resources more directly. 7 An impact assessment of GALS training between 2007 and 2010 found that out of 291 people sensitised about the benefits of joint land ownership, 61 households now have joint land certificates from their village council and 25 have registered customary joint land agreements. Of 1096 participants in action learning about co-operation in the household and 'in the garden', men are beginning to take responsibility for a few roles like collecting firewood in 449 households; and in 366 households men and women are sharing most or all responsibilities. In 2015, their website now states that 'the GALS process has led 67 per cent of our married farming households to have joint names on documents for coffee certification' suggesting these gains have been sustained over time. 8 Farmers linked to the marketing association report having a direct buyer for their produce at a price they can trust. Savings-group members explain that the pictorial methodology means everyone can understand how money is being accounted for and used irrespective of literacy. Socio-economic mobility is triggering the assumption of leadership positions: a member of the training team had gone on to become a sub-county councillor; two out of five members of the sub-county land committee were Bukonzo Joint members; and another member had gained a seat on the sub-county farmers' forum.
Bukonzo Joint's mobilisation may not have catalysed binding forms of consultation with the state, but it has begun to reshape power relations in such a way as to contain arbitrary action by the state and secure the interests of previously marginalised groups. In 2010, the Bukonzo Joint leadership circumvented a sub-county development plan in which the local government had decided to invest Belgian government funding into a potato-growing project rather than rural electrification to support coffee processing. There were rumours that the potato project was favoured because it would provide an easier source of rents for local government officials. With the backing of the membership, the Coordinator negotiated directly with the Belgian government, who sent representatives to look at Bukonzo Joint coffee production and marketing work, and then applied pressure on the sub-county to change their plans. Here, then, is a case of an originally informal trust network constructing a business which joins the formal economy, but exercises contingent consent when the state fails to protect their interests, and uses international connections to level the political playing field.

Trust networks and public services: building state capacity and human capabilities?
RD's rights and governance training has largely failed to secure greater influence for local citizens over the formal participatory processes that govern health and education services (King, 2015). Parents and service users were no more confident to challenge teachers or health workers and village residents did not question the absence, infrequency or ineffectiveness of village planning meetings. PTAs were defunct or tokenistic in most schools. However, RD's sub-county dialogues, which target the 'intermediate' classes that often have a significant mobilising role within clientelist environments (Khan, 2010, p. 54), has incrementally built participants' capacity to govern local services effectively, and provided an alternative (albeit non-binding) problem-solving space for state-society interaction. Teachers and health workers, management committee members and councillors reported that they had gained a clearer sense of their responsibilities for improved performance in health and education provision and that some had altered their behaviour and approach in response to the dialogues. Participants also reported that recent improvements in academic performance flowed, at least in part, from the sharing of good practice within these problem-solving spaces. In 2008, a series of these dialogues secured a new sub-county education bill aimed at reducing school drop-out rates through the more effective enforcement of sanctions. Dialogues also triggered a public apology and improved staff behaviour in a district health centre, which qualitative accounts suggest has also resulted in increased uptake of services. These findings reflect Tilly's (2007, p. 139) emphasis on efforts to facilitate 'popular influence over public politics ' (2007, p. 139), and also Evans and Heller's (2015) argument that increasing the volume and variety of spaces for state-citizen interaction is a vital ingredient of democratic developmentalism in the twenty-first century.
The widespread social mobilisation catalysed by Bukonzo Joint's model of community organising has led to increased levels of non-state social provisioning within the parishes where they have primary co-operatives. This includes the emergence of a new immunisation centre, school, and nursery, either run or financed by members. Combined with the increased levels of household income we reported earlier, this increased provision means that more parents can afford to send their children to school and is resulting, according to one sub-county chairperson, in better educational outcomes in the area. In addition to these informal 'segregated' services, members are also building the capacity of local state services and public transport infrastructure with support from Fair Trade buyers and the Belgium Development Agency. This includes the extension of local health facilities, building a new road and bridge to connect farmers from remoter areas to Bukonzo Joint's 'micro-washing station' enabling them to increase the value of their product, and building water pipelines which link farmers to Building democracy from below 1593 washing stations while also alleviating the burden on women and children to carry water over long distances for daily consumption. This integration of formerly excluded trust networks within public affairs and policies not only reflects 'the effective translation of citizens' expressed collective will into state action (Tilly, 2007, p. 94), but also further extends Bukonzo Joint's contribution to reducing categorical inequalities through improved health and education outcomes for rural citizens and greater economic inclusion for previously excluded farmer households through the mobilisation of segregated self-help networks.
RD's civil society building work has supported the emergence of a regional anti-corruption coalition as well as the regional farmer training network mentioned previously. The anti-corruption organisation, which by 2012 was operating in 30 sub-counties and had approximately 450 activists monitoring projects and services on the ground, was able to recover goods or money to an estimated value of 3.5 billion shillings between 2005 and 2010. Radio is the most accessed form of media in rural Uganda and RD also established a radio talk show that enabled civil society professionals to directly question politicians and civil servants about their use of resources in the sub-regionechoing Tilly's emphasis on access to procedural devices and non-state political resources. Amidst multiple reports of the efficacy of these shows, the most notable impact concerns the exposure of a former District Chairperson's alleged corruption over land contracts which helped push him to stand down before the 2006 elections.
The deliberative spaces that RD has convened at regional level since 2002 have culminated in the Rwenzori Development Framework, which, in ensuring that development interventions are shaped by local evidence and agenda-setting, has begun to reshape the balance of power between local farmers and civil society elites, and between such civil society elites and international donors. This process has increased the capacity and commitment of the region's elite cadre of civil society professionals to promote community mobilisation and organisational development, advocate on behalf of marginal groups and reflect more deeply on the region's development challenges and potential. Some representatives from community-based organisations have also participated in this process. Farmers within the regional agricultural training network are now acting as a bridge between these NGOs and farmers on the ground, cutting across social categories and linking community based organisations to training and grants from these larger elite-led organisations.
RD's leadership retreats have fostered more positive relationships between civil, state and political elites. This has led to non-state actors having greater inclusion within and influence over local planning and budget processes, with district planners increasingly receptive to locally-generated research by civil society actors. A critical outcome has been a rapprochement between the leaders of the Rwenzururu and Toro kingdoms, and also between ruling party and opposition politicians, with some MPs now working collaboratively to form a parliamentary caucus for the region. The think tank regional stakeholder forums have catalysed a process whereby state, civil society, politicians and farmers are all engaging in analysing locally-generated data and planning action in response. Their investment in this process and commitment to pursuing particular action points suggest at least a nascent commitment to improved development outcomes and more inclusive forms of governance. Again, while these outcomes do not speak of binding forms of consultation, they demonstrate shifts towards more responsive forms of consultation through Tilly's integration of segregated trust networks within public affairs and also a greater receptivity amongst the powerful to a more equal distribution of power (Törnquist et al., 2009).

Scaling up?
There are clear signs that these civil society groups are contributing, at local and regional levels at least, to the processes of democratisation that Tilly identifies as critical, including through challenging unequal social relations, reducing the space for arbitrary state action, and creating interfaces for greater consultation between state actors and low-income citizens. But to what extent do these incremental changes engage with the broader structural constraints that shape how democratisation and development are currently unfolding in Uganda? This is clearly a considerable challenge, and as yet there is only incipient evidence that these organisations are proving able to transcend the 'localism' that often besets such initiatives. For example, the democratisation of regional development thinking and planning has rendered such processes more subject to local demands in ways that may start to undermine 'the mediating power of patrons' (Tilly, 2007, p. 94). In a different kind of case, the regional caucus of MPs catalysed by RD has successfully secured 500 million shillings 9 a year for three years for the new regional university, but also offers a platform for securing gains more broadly. This group of MPs includes RD's former director, one among many progressive politicians in Africa that honed his political skills within a civil society training ground (Ndegwa, 1996), and who has taken a vocal stance within parliament against corruption in the country as a whole.
Both RD and Bukonzo Joint are also seeking to challenge unequal relations of power within the economic sphere. This has mainly involved promoting the greater integration of local trust networks into the formal economy by brokering relationships between local producer associations and large Kampala-based buyers to achieve better terms of market inclusion for local producers. RD has achieved this through its business forums which have brought national buyers together with local marketing associations; Bukonzo Joint, by building a reputation for efficiency, trustworthiness and high quality coffee throughout the coffee value chain. Within Uganda, they are members of the National Union of Coffee Agribusinesses and Farmer Enterprises which aims for a 'sector-wide impact on societal transformation' 10 , and are spreading learning and building solidarity through the Joint Marketing Initiative, a network of African smallholder coffee cooperatives and the UK-based Fair Trade 'pioneer' Twin. Given the strong emphasis on agro-processing and value chain approaches within Uganda's new National Development Plan 2016-2020, it is conceivable that national elites will take note of the revenue-generation resulting from such smallholder enterprises, and provide greater public support for their activities.

From theory to practice: what works?
Framing the activities and achievements of these organisations in terms of Tilly's 'democratisation as process' helps reveal the causal mechanisms through which social democracy may emerge from below in agrarian settings, and also suggests the need to refine aspects of his conceptual approach in relation to the particular challenges of promoting democracy in neo-patrimonial agrarian contexts. Here we revisit the three key dimensions of Tilly's approach, namely the integration of trust networks into public affairs, the reduction of categorical inequalities and the insulation of public institutions from powerful players, before briefly identifying some of the more practical implications that flow from this with regards to how to support efforts to build democracy from below.
Tilly's emphasis on the integration of trust networks within public affairs as a key mechanism through which societies move towards democratisation resonates with our evidence, which also suggests that CSOs can play an important role in this process. This has been most successful where the trust networks involved have focused on the smallholder producer groups that Sandbrook et al. (2007) identify as critical agents of democratisation over the long-run. The solidarity generated by common livelihood concerns, coupled with the challenge that such organisational forms make to categorical inequalities, has proved a more powerful driver of political engagement amongst citizens than working through more formal and political approaches alone. However, in contexts like rural Uganda, this move needs to be preceded by a period of capacity building within such networks and may require a focus on informal as well as formal 'public' affairs. Faced with high levels of rentseeking and corruption, Bukonzo Joint initially focused on developing the capabilities of members and the wider organisation, including through greater economic autonomy, and on avoiding direct engagements with formal political channels until the capacity to exercise contingent consent was built. RD's focus on trust networks also highlights the critical importance of focusing on informal institutions and channels of influence in these kinds of contexts, as does Bukonzo Joint's efforts to secure the support of clan and religious leaders for their efforts to transform household gender relations. That Bukonzo Joint initially focused on one particular clan before broadening their focus, offers a microcosm of how popular agency can move from the particular to the universal regarding processes of democratisation (Foweraker & Landman, 2000).
This integration of trust networks within public affairs can also help address Tilly's second concern with reducing categorical inequalities, through increased levels of social provisioning. More generally, challenging inequalities have been achieved most effectively here through Bukonzo Joint's politicised use of participatory methods and capacity building, which work from the bottom up starting with household gender relations before moving upwards to shifting relations between smallholder farmers and local elites or national buyers. This approach has had both material and discursive effects, in terms of shifting social attitudes with tangible economic evidence of the capabilities of women in the eyes of men, and of small holder farmers in the eyes of elites and state actors. This methodology, which is designed to forge cross-class coalitions through teaming upwardly mobile farmers up with those struggling on lower-incomes, flows from a commitment to promoting social justice that goes beyond a focus on poverty reduction alone.
This focus on reshaping social attitudes among wealthier groups towards lower-income groups relates not only to challenging categorical inequality but also the success of these organisations in opening up dialogical spaces that have helped counter-act the tendency for public institutions to be captured by more powerful actors in society (Tilly, 2007). RD has used research and the creation of knowledge and dialogue-based platforms to bring cross-class stakeholders into interaction with one another, which has begun to reshape the attitudes of elite classes towards the capabilities of smallholder farmers and also encouraged elites to be more open to popular involvement in decision-making processes (Törnquist et al., 2009).
Moving from theory to practice, we now briefly discuss how our evidence relates to the broader literature on how CSOs can contribute to building democracy from below in the terms set out by Tilly. At a broad level, this starts with the recognition that building democracy in agrarian contexts requires engaging with the political economy and not just the politics of democratisation. Whereas much mainstream development theory and practice focuses on a vision of 'civil society' that is glimpsed only partially in some urban settings of Africa, democratisation in rural contexts must involve building the organisational capacities and influence of the main productive groups and classes in such contexts (Sandbrook et al., 2007). The case of Bukonzo Joint shows how creating solidaristic relations of economic interdependency and socio-economic mobility can, when channelled through collective and federated organisations, move towards achieving greater political influence for subordinate groups (Patel & Mitlin, 2009). This in turn flows from CSOs developing and promoting development as part of a wider project of social justice (Kabeer et al., 2010;Thorp et al., 2005). Finally, our case-study organisations reveal an impressive capacity to build and manage relationships across a range of boundaries and levels (Bebbington et al., 2008). This has involved bridging between organised subordinate groups and political and economic elites, and between civil and political elites (Evans & Heller, 2015), and forging relationships with external financing agencies in ways that draw in skills, connections and resources, while preserving their autonomy. External agencies have been critical in terms of providing technical assistance and financial resources through normal aid channels, and also more broadly through links to trading companies and movements. However, the long-term strategic partnership funding that RD received and the fact that Bukonzo Joint's core business has remained financially and strategically autonomous (with salaries paid out of a percentage of loan interest and share dividends, and donor finance accepted only for very particular purposes), has been critical to each organisation being able to pursue their vision and remain responsive to local realities. As such, and whilst transnational actors will continue to play important roles within processes of democratisation, these engagements need to be subordinated to local imperatives and processes.

Conclusions
Tilly's (2007) notion of 'democratisation as process' provides us with a framework for understanding and capturing the specific ways in which particular kinds of association can help build democracy from below within semi-authoritarian, neo-patrimonial, and agrarian contexts. Moving beyond the current focus on how elite-bargaining and certain associational forms may contribute to liberal forms of democracy, this approach helps identify the intermediate mechanisms involved in building democracy, including the significance of challenging categorical inequalities, and of building trust networks, cross-class alliances and synergistic relations between civil and political society that can render public institutions more responsive to popular influence. Following Sandbrook et al.'s (2007) work on how social democracy may emerge in the global periphery, our evidence foregrounds the critical role that can be played by smallholders and producer groups in catalysing long-term processes of socioeconomic and political change, and starts to suggest alternative routes through which external actors can support moves towards building democracy from below in sub-Saharan Africa.