UNDERSTANDING GOVERNANCE FROM THE MARGINS WHAT DOES IT MEAN IN PRACTICE?

What does governance look like ‘from below’ – from the perspectives of poor and marginalised households? How do patterns of conflict affect that? These were the questions at the heart of the Governance at the Margins research project. Over three years from 2017-2020 we worked to explore this through in-depth study in conflict-affected areas of Mozambique, Myanmar, and Pakistan. Our research teams interviewed the same people regularly over that time, finding out how they resolved problems and interacted with authorities. In this paper we connect what we found to the realities and complexities of development practice, drawing on the input of 20 experienced practitioners working in bilateral and multilateral development agencies and international NGOs, who generously gave their time to help us think through the practical implications of our wealth of findings.


INTRODUCTION AND ABOUT THE PROJECT
What does governance look like 'from below' -from the perspectives of poor and marginalised households? How do patterns of conflict affect that? These were the questions at the heart of the Governance at the Margins research project.
Over three years from 2017-2020 we worked to explore this through in-depth study in conflict-affected areas of Mozambique, Myanmar 1 , and Pakistan. Our research teams interviewed the same people regularly over that time, finding out how they resolved problems and interacted with authorities.
Here we connect what we found to the realities and complexities of development practice. We draw on the input of 20 experienced practitioners working in bilateral and multilateral development agencies and international NGOs, who generously gave their time to help us think through the practical implications of our wealth of findings.
This document is organised around 5 key findings. After each finding there are some prompt questions for practice, and some practical responses to the finding suggested by practitioners. We don't suggest one blueprint or uniform way of responding because we think it is crucial to recognise differences between contexts -particularly different intensities of conflict and violence and their drivers. Responses will look different in different contexts.
In our discussions with practitioners a number of cross-cutting challenges and dilemmas emerged. A separate section explores these in more depth, raising some bigger questions. 1 In Myanmar our research was conducted prior to the military coup that took place in February 2021, and references to government authorities are not to the subsequent military-led regime.

About the project
The first phase of our project, from 2017-2019, was entitled 'Governance Diaries' after the innovative research approach we developed. By visiting and interviewing the same households approximately monthly over the course of 12 months we were able to identify the kinds of issues that came up for families, what actors they did or did not engage to help resolve them, and how far these issues were resolved. We included more than 160 households in this phase, across the three countries.
In a second phase of the research, from 2019-2020, we focused in on findings from the first phase. We identified 80 key intermediaries in a number of the communities from the first phase -people who households in those communities told us were important in solving local governance problems -and used the same research method of visiting and interviewing them regularly over 12 months. We looked to see what kinds of issues they became involved with, what strategies they used to resolve them, and how they navigated multiple authorities and linked into the wider governance system. We also adapted the method to investigate the roles of intermediaries and community level problem-solving after the onset of the COVID-19 pandemic.
Differing views on the legitimacy of different authorities also affect who people will engage with and on what. History, politics, conflict, and identity affect these perceptions of legitimacy.
Authorities often overlap, whether cooperating or actively competing with one another in the same areas and on the same issues. This is especially the case in Myanmar, where there are parallel authorities to the Myanmar government established by ethnic minority groups in many areas. In other places the overlaps come when different groups or organisations with influence decide on a more ad hoc basis who should get involved in particular issues, or compete to resolve problems to bolster their own standing. There are often clearly understood differences in what kinds of issues authorities are responsible for or are better at resolving. For example in some communities in Mozambique, mining companies are seen as having the authority to resolve certain specific issues. Who people see as a 'duty-bearer', relevant to solving a problem or accountable for taking action, doesn't necessarily align with any formal government mandate or control over state resources.
Multiple authorities mattered in all of our research locations, both rural and urban, although who was important varied across different places within each country. Diverse authorities still mattered even where the state or formal authorities were very visible. In some of our urban research sites, there was more obvious state-led service provision but other authorities were also important. Sometimes this was because state services were inaccessible to the poor and marginalised for other reasons.

Questions for practice
• If we are not currently working with a diversity of informal actors, why is this? Have we decided that they aren't relevant to the outcomes we're seeking? Or are there barriers to working with them?
• How do we identify and assess the local legitimacy of a diversity of formal and informal actors?
• If the formal system is weak, is this because powerful actors have vested interests in keeping the state weak, or simply a consequence of conflict, context and history?
• Considering the limited risk appetite amongst many donors and development actors, how in practice can we work with more 'political' actors such as parties, parallel authorities, and non-state actors?
• What role can we play in facilitating convergence or interaction between multiple authorities, and in what contexts?
• In what situations does it make more sense to support a plurality of authorities rather than trying to bring them together?
• What are our implicit assumptions about what 'good' development looks like when it comes to the role of non-state and informal authorities, and should we be questioning these?

Examples from our research
• In Myanmar community members in one village tract had engagements with not only the Myanmar government and the Myanmar military, but also with both the administrative and armed wings of five different ethnic administrations. These engagements included taxation, escalation of community issues, justice/conflict resolution, security, service delivery and receiving directives from above. Be explicit about ideological differences: Some of us see the primary aim of development work as building institutions and public services, and particularly the institutions of government and democratic governance processes. Others are more focused on community resilience, meeting immediate needs or cohesion. Being clear on our programme's objectives and what we think success looks like, and articulating our underlying assumptions, can guide us in engaging with a diversity of authorities.
Think through acceptable compromises: Some authorities viewed as legitimate from within the community may be viewed as 'unsavoury' or too political by development actors (e.g. armed groups, militaries, political parties, or some customary institutions). This is particularly likely to happen where national identity and the state is contested. Engaging with non-state authorities, including those that may challenge the state, requires complicated compromises and careful understanding of when there is an unacceptable risk that core principles might be undermined.

Challenge risk aversion:
Creative solutions may be needed to navigate compliance and risk management requirements when these reduce our ability to engage with non-state authorities or particular kinds of actors with influential roles. Because this may require extra resources and increase risks for those implementing programmes on the ground, strong trusting relationships with donors along with high-quality risk monitoring are key -but this may be harder to secure where donors or other development actors are also prioritising their relationships with state authorities.

How can we respond to this?
Practitioners that we spoke to gave these ideas: Assume a diversity of important authorities: Particularly in conflictaffected contexts we need to reject the default assumption that state institutions are automatically in control, and instead start from the premise that there are diverse sources of authority, many of whom have local legitimacy, and that the state is simply one actor among others.
Seek a deeper understanding of the diversity: Undertake ongoing analysis to understand the complexity of authorities and their roles on different issues. Who undertakes this analysis is critical, as local insights on various authorities can help us to move beyond looking at the alreadydocumented structures, roles and responsibilities. Being aware of multiple authorities is important for minimising harm, whether we end up working with them or not.
Create solutions to circumnavigate compliance and risk requirements that reduce the ability to engage with nonstate authorities or particular kinds of influential roles. Strong trusting relationships with donors along with strong risk monitoring is key.

FINDING 1: MULTIPLE AUTHORITIES
Intermediaries were key to deciding whether problems experienced by individuals or communities should be escalated, to which authority and how. In most research locations, and particularly where conflict was more active, community members had very little, if any, choice in which intermediary they could go to. However, intermediaries could often choose which external authorities to approach and how these contacts were navigated. Different intermediaries also focused on different issues. In contexts experiencing active conflict, intermediaries were Many development programmes and policies assume that people can approach and connect with public services and officials themselves. We found it is more common that these contacts are mediated by key individuals. This challenges assumptions about how governance and services work in practice. The informal or unofficial nature of these roles, coupled with the fact that those playing them are sometimes unusual governance actors, means that despite practitioners often knowing how important they are, they are frequently missed out when programmes map what structures are in place and who is important.
In all our research locations intermediaries were essential to the functioning of local governance, serving as the 'grease' that oiled governance systems.
There are many terms we could use for this role, but the best general term for their function is that they mediate. We mean this in two ways. First, they connect people to higher-level authorities and other decision-makers to get problems solved. They sometimes communicate high-level authorities' rules to community members, and sometimes select who benefits from resources allocated by those authorities. Through these activities they mediate relationships between authorities and community members. Second, some intermediaries have decision-making and enforcement authority themselves, so they mediate within the community -resolving problems between people at that level, coordinating collective activities or providing services. Intermediaries' behaviour and choices can amplify or confound other attempts to improve local governance and public services.

PRACTICE QUESTIONS
WHAT WE FOUND HOW WE RESPOND?

RESEARCH EXAMPLES
Intermediaries are crucial to how governance happens. They work as the 'navigators' of diverse sources of authority, work across formal and informal local governance systems, and in some cases exercise significant authority themselves as 'deciders'.
In South Sudan, we saw organically strong and legitimate groups of women becoming 'formalised' and lose their credibility once funded by an INGO -it very much changes dynamics when external actors come in. So we learnt from this and instead accompanied small informal groups, connected them with other networks, provided small seed funding and acted as a sounding board.
often particularly focused on ensuring protection and security for the community -which could involve negotiating with parties to the conflict.
Intermediaries' approaches and tactics varied across locations depending on the level of influence or control they had over local development issues. For example, in some locations Intermediaries played more of an overt public authority role, only escalating issues to higher level authorities when absolutely required. Others played a subtler 'advocate' role or used less obvious and more covert tactics. This was particularly notable among women intermediaries, who would often leverage the position of their husband or son within the community or use their own connections to other women to influence decisions outside of formal channels.
Some intermediaries spoke of undertaking their role as a duty or obligation to others, but many also had other incentives for accepting these positions. Their position as intermediaries often provided social status or supported their political ambitions. Particularly where their role was not otherwise renumerated, being an intermediary sometimes also provided financial reward through 'fees'. These fees were framed as recovering the costs of their time or travel, rather than generating substantial income.
Intermediary roles are sometimes thought of in negative terms -as being predatory, supporting clientelism, or being liable to co-optation. In our research, however, we saw a more positive -or at least neutralview from households. Intermediaries were generally seen by households as more trusted and able to deliver for communities than formal or more distant higher-level authorities. This might be because the intermediaries we studied often offered some very localised accountability -as they could be called on to explain their actions in order to sustain their position or to uphold norms about what the role entails. This more positive view might also be because we focused on the intermediaries who households had identified as important to them.
Intermediaries have different characteristics. Some are political party brokers. Some are community activists. Some are elected village leaders. Some are retired government officials. The intermediaries we studied all had close relationships to their local areas. Most often they were current or former community leaders, members or former members of community associations, other grassroots organisations, political parties or movements. They often held widely acknowledged 'informal' leadership and intermediation roles rather than official positions, although these distinctions generally didn't matter to community members. They are often seen by those they 'represent' as having a legitimate right to act and take decisions on their behalf. What gave these individuals their legitimacy and importance was different in each location. Across our locations, in addition to community social norms, this related to individuals' personality traits, abilities, histories, resources or connections. Representing a particular identity -whether ethnic, religious, or political -was also important.
• If we're choosing not to engage with informal intermediaries, why is this? Is it an oversight or are there barriers -for example, concerns around legitimacy or corruption? Or have we made a conscious choice to focus only on building the formal system?
• Does our analysis, particularly at the programme design stage, seek to identify and take into account these different kinds of intermediaries?
• How do we challenge perceptions of intermediaries only as 'gatekeepers' that can give programmes access to communities, and instead engage with them as actors who exert public authority themselves?
• How can we work with these actors whilst also avoiding creating opportunities for elite capture or reinforcing any exclusionary practices?
• How can we engage with intermediaries in a way that avoids undermining their legitimacy and the respect in which they are held by community members and/or public authorities?

Examples from our research
• In Myanmar many intermediaries were village leaders. Within the Myanmar government their roles were not formally recognised, but within many communities they were viewed as having an official position and held significant power. In some communities they were elected, in others they were appointed or rotating. Village leaders engaged with other influential people within the community as well as a variety of external authorities, such as the Myanmar government and multiple ethnic administrations. Some issues will move between multiple authorities before they are resolved.
• In Mozambique there is a very structured and hierarchical decisionmaking system of formal local governance, but intermediaries' informal contacts were key in getting things done. This included some intermediaries with strong links and relationships with ruling party, and others whose position with a community association allowed them to bypass the official hierarchy.
• In Pakistan brokers with good connections within the state and to service providers were particularly important, but so were groups that claimed to represent people and mobilised them to act collectivelysuch as social movements. In Islamabad, two intermediaries from a left-wing party working on housing rights for slum residents often dealt with the municipality on behalf of the residents. Whenever government officials came to demolish houses, residents called these intermediaries for help. When someone got sick or needed medical care, they also contacted them first.

FINDING 2: INTERMEDIARIES
Influencing norms around intermediation. It is important to examine the social and governance norms that shape who has legitimacy to mediate for or within a community. We should consider working to influence norms and expectations around how intermediaries behave, involve people, report back to them and treat the most vulnerable.

Recognise intermediaries as political actors.
Understanding intermediaries' political ambitions and incentives -as well as competition or division between intermediaries -presents opportunities for action and change, as well as for gaining a more holistic picture of the local political economy. Engaging with them is an important part of politicallyinformed programming and policy. This often happens under the radar at point of delivery, instead of being more overt and deliberate -but a more overt approach does come with a trade-off in terms of the risk of perceived or actual politicisation of development work.

How can we respond to this?
Practitioners that we spoke to gave us these ideas: Engage with intermediaries as important actors without 'projectising' them. Recognising intermediaries' roles and in some case supporting them doesn't mean formalising a relationship or funding them directly. We should resist the temptation to 'formalise' or 'projectise' intermediaries, or to try to co-opt them into external agendas. One risk is undermining their community-level respect and legitimacy, and another is concentrating or condensing power in these roles. These risks can be monitored.

Engage constructively with intermediaries' authority.
Concerns about intermediaries being predatory or 'gatekeeping' in non-inclusive ways may lead us to try to get around them, rather than engaging. But rethinking the role of these intermediaries and recognising them as important holders of public authority may open up possibilities to work with them, and indeed to challenge them to adopt more inclusive and less self-serving practices if necessary.
Understanding intermediaries' political ambitions, incentivesand competition or division between intermediaries -presents opportunities for action and change, as well as giving a more holistic picture of the political economy. Engaging with them is an important part of politically-informed programming and policy. Development programmes and policies too often assume that there is one system of governance or decision-making across wide territories. There is also often an assumption that a linear hierarchy through national, regional and local governance institutions means that policy is adopted and implemented in the same way across those territories. That isn't how real local governance works in our research locations. Many practitioners are very aware of this, but end up working with frameworks that aren't flexible or nuanced enough to respond to it.
Across the towns, villages and neighbourhoods where we worked we saw various sources of authority connected to one another in networks. Different actors became involved in different issues at different times and 'moved' issues and decisions between members within their networks. These networks were made up of different combinations of multiple external authorities and local intermediaries. Networks of public authority were not limited to the kinds of people who are usually seen as governance actors; they also involved religious authorities, people in service provision roles such as doctors, and in some cases companies or private businesses (both legal and illegal). Organised collectives were also important parts of the web of public authority. Across all three locations, it was often individuals who were identified as important and trusted to act, rather than the institutions that they were part of or represented.
The picture of which sources of authority mattered and how they were connected looked very different in each of our research sites. How public authority was built and sustained, how local governance worked,

PRACTICE QUESTIONS
WHAT WE FOUND HOW WE RESPOND?

RESEARCH EXAMPLES
Communities are governed through diverse local networks. 'Standard' local governance structures rarely apply, and there can be a lot of variation even within one region.
We can find good arguments to convince donors that iterative analysis is not necessarily costly. We have designed it into our regular programme monitoring. By doing this we can anticipate issues and put measures in place earlier when things change… these are good selling points.

PRACTITIONER REFLECTION
who was involved, over what issues, and how -all these aspects varied significantly. Factors driving this diversity included: • the presence of structures with different historical or customary origins; • the level of active violent conflict; • the level of political competition; • the extent of parallel governance systems; • the presence or threat of armed groups; • the geographical location (including proximity to powerful authorities); and • the dynamics of centre-local relationships with the formal government.
Whilst the relative importance of various higher-level authorities varied across contexts within one country, so too did the specific authorities within a network. In some locations, political party members or former local leaders were very powerful, whereas in other locations they were less so. Religious leaders and more formalised civil society groups were highlighted as part of some networks and were powerful in some locations. However, in general, they were not as prominent in local governance as some development narratives suggest. Other public sector or service delivery actors to whom development programmes rarely pay much attention, such as doctors, teachers, midwives or judges, featured as important actors within some of the networks.
Interactions within and across these networks relied on people using their existing familial, social and political capital, which has significant implications for inclusion and exclusion. As social capital is often connected with material wealth or education, these networks provide some people with more access to authorities than others -privileging local elites, and marginalising other groups. Arguably, more formalised institutional structures can compensate for a lack of social capital by guaranteeing all citizens' rights of access. However, in their absence accountability for decisions that include some people and exclude others is shaped by social norms and cultural expectations. These networks have the potential to exacerbate inequalities in social capital, meaning that people who are already doing well continue improving their lives but those without the necessary social connections have no-one to approach for help. Governance networks and experiences were different in urban and rural locations, but for different reasons across countries. The variation was driven by where political power was concentrated. In Mozambique the existence of political competition between Frelimo and other political parties in some urban areas meant that we found more contestation and multiple party-linked networks in these areas, whereas in the rural areas Frelimo tended to dominate all networks. In Pakistan the dominance of landlords and the military as core power-holders in the rural locations we worked in make a difference to the networks. In some urban parts of Myanmar physical proximity to government offices and relatively low levels of contestation of the state's authority influenced the shape of the networks.
In a project in Pakistan, we undertook in depth Political Economy Analysis and regular power analysis. Communities then did their own analysis ('power change analysis') which was also an empowerment tool. We also used the Reality Check Approach. Without this granular analysis, it would have been hard to do anything.

FINDING 3: DIVERSE LOCAL NETWORKS
• Do existing practices and standards for programmes' analytical work, particularly at the design stage, emphasise identifying networks, intermediaries and authorities from the perspective of the people whom the programme is seeking to reach?
• Do we over-generalise in assuming the importance of some actors in governance networks, such as formal CSOs? If so, does this make us less likely to look for other actors who may be equally important?
• What kinds of analysis would tell us about the crucial variations across local governance and community problem-solving, understanding these network dynamics and identifying where informal power lies? How do we find people who can meaningfully undertake this kind of analysis, and keep it up-to-date?
• How can programmes combine a sufficiently uniform design across locations with recognising that these locations may have very different decision-making and power structures?
• Does the variance in local governance networks mean that we tend to focus programming in certain areas or emphasise working with specific authority systems where there is more regularity?

FINDING 3: DIVERSE LOCAL NETWORKS
Conflict analysis has to be prioritized but power analysis is really helpful to go deep in these power relations. We find tools mapping actors, interests and attitudes and power really useful.

Examples from our research
• In Islamabad, the proximity of informal settlements to government departments means that different kinds of access are possible and important. For example people who work in government offices as cleaners have been able to build connections that allow them to raise problems or find out who can help with a problem.
• In parts of Myanmar, village committees make these networks more visible. In some locations in Kachin state, these committees consist of a ward/village administrator (head of committee), a secretary and/or associate secretary, and in some cases youth and women representatives. Parallel to this, a ward level 'traditional committee' addresses issues using customary law. Elsewhere, however, these committees do not exist, with village administrators having more concentrated power while other influential individuals within the village, are drawn on for specific issues.
Balance insider and outsider perspectives. The best analysis will come from combining different perspectives. Particularly where ethnic, tribal, political or religious divisions are strong, many people will have implicit assumptions about the 'way things work' in different places. Including the knowledge, intuition and insights from those in the locality/community alongside information from other places in the country or beyond can surface tacit assumptions on both parts, help with unearthing hidden power dynamics, and challenge received wisdom.

Design for adaption.
Having a programme design that is flexible enough to respond to ongoing analysis is key to ensuring that our activities are responsive to change. This is often a challenge when detailed programme designs are required in advance. Building a programme that focuses on overall principles and approaches -for example articulating the different types of duty-bearer we want to engage with rather than explicitly naming specific authorities -can allow the flexibility to adapt approaches and expand the range of actors we work with when things change on the ground.

Design in variation.
The aim should be to shift our focus away from uniform activities, and more towards overall principles that can work within a variety of local governance networks. The challenge of managing a wide variety of interventions to address micro-level variations in governance environments can be mitigated by grouping project locations into a smaller number of 'types' that require a particular kind of approach.

How can we respond to this?
Practitioners that we spoke to gave these ideas: Analyse local governance realities carefully. Relations between individuals and public authorities within these networks are influenced by complex and evolving power dynamics. Analysis must be grounded in people's own understandings of legitimacy and authority from within a given community, and nuanced enough to identify and map who people really go to for help. If necessary this may mean challenging biases towards 'the usual actors' and formal or official organisations and figures, including CSOs or religious leaders.
Ensure that analytical tools are fit for purpose. We need to look at existing analysis tools and critically reflect on whether they are mapping the networks in the way we need. Embedding ethnographic or social network approaches within our analysis may help to surface different connections across the network -for example family and business links. This information may already exist outside of our programmes in existing research.
Integrate regular analysis. Power and influence within networks can change quite dramatically over time. Building analysis into existing systems and processes, as an ongoing component of a programme or policy, is a way of ensuring that our understanding of these remains accurate. This can be a formally structured part of the monitoring, evaluation and learning system, or as light-touch as having informal chats with the same people every few months, which helps to build up trust. Once trust exists, these check-in points may be enough to ensure that programmes' assumptions about local governance remain valid. These low expectations combine with other incentives to avoid approaching higher authorities or using official channels. Doing so can be costly -both in time and in 'fees'. It can require social capital, language skills, or ways of behaving that are unfamiliar. It can also be seen as an escalation that sends a signal that communities are not able to resolve their own issues, with going outside of local channels being seen as 'airing dirty laundry' or embarrassing the community, especially when the community is already marginalised and facing negative stereotypes. This is also gendered -women taking their issues outside of the community can be seen as them 'exposing' themselves or sharing information that should be kept private.

UNDERSTANDING GOVERNANCE FROM THE MARGINS
Previous negative experiences of seeking an official solution also mattered -either direct experience, or common narratives. Sometimes multiple competing authorities mean that deciding to approach or not to approach any given authority is a significant political move, and one which may have negative ramifications within a conflict-affected setting. Attempts by higher authorities to deter people from escalating their concerns are also common, and may include pushing officials or intermediaries to attempt local solutions first.
Assuming a preference for state-run or centrally managed services and decision-making may not be in keeping with community views and practices and may be premised on a demand for service provision that doesn't exist at a local level. This has a range of implications for public policy and development programmes.
Across our locations, people, and many of the intermediaries to whom they turned, did not expect much from the wider governance system. Higher level authorities were often parties to conflict or seen as aggressors. Historic under-provision of services and limited experience of solutions actually being provided by those authorities were often coupled with a sense that authorities were implicated in causing the problems in the first place. Sometimes the physical distance from where offices and officials are located is a factor. Distrust of authorities, and political and social divides (including on identity lines) also play a role. In some situations, low expectations of a particular authority are due to political views about who the most legitimate duty-bearer is. Low expectations result in fewer demands being made of higher authorities.

PRACTICE QUESTIONS
WHAT WE FOUND HOW WE RESPOND?

As a result, what we term 'self-provision' is widespread. By selfprovision we mean situations where solutions are found without involving higher-level authorities or official duty-bearers.
• Sometimes this involves intermediaries or other informal sources of authority making decisions or taking steps to secure local provision of services such as electricity generation or water supply.
• Sometimes it involves intermediaries intervening between parties and coming to a decision, which is then respected by those involved.
• Sometimes it simply involves people negotiating with each other to resolve a problem that could have gone to a more formal or institutional resolution process.
In the more extreme cases, communities or intermediaries establish their own local rules, resolutions, and punishments. Community support for these kinds of self-provision solutions is sometimes linked to what is seen as a culturally appropriate and 'just' outcome, regardless of broader policy or legislation.
This doesn't necessarily mean that people see self-provision as the best solution; sometimes it is considered the least-worst result based on previous experience and political dynamics, or acknowledged as purely a strategy for survival. In Pakistan, for example, many people involved in our research felt that powerful local elites benefit more than poorer and marginalised households from self-provision, yet they feel 'locked in' to this system and have few alternatives.
In one location, people from local communities formed 'solidarity' networks to help manage influx of displaced people. Our organisation brought additional support to the community focusing on access to basic services. Tensions arose between our work and what the solidarity network was doing. So we shifted our approach, so that community members could input into criteria and be transparent. We involved intermediaries, listened directly to people, and kept our ears on the ground. This looked very different from the original service delivery approach. It took time but was more legitimate as we were really listening. We saw micro issues as indicative of systemic issues and integrated learning to help us replicate this changed modality.

Examples from our research
• People in the informal settlements of Islamabad mostly prefer to solve disputes locally, rather than involving the police. Their preference is based on their historical experience with law enforcement agencies and police brutality. For example, one of the intermediaries we followed was approached to resolve an assault linked to domestic abuse. The family members of those involved approached him rather than going to the police station and registering a criminal case.
• In Myanmar explicit 'village rules' were commonly developed and applied. Village rules range from banning alcohol on full moon days to not letting animals wander recklessly, to night-time curfews (often applied due to the likelihood of soldiers coming into the village in the evening). Breaking these rules incur community sanctions -generally agreed-upon fines or punishments involving public shaming, such as being tied to a tree in public all day.
• In one urban neighbourhood in Mozambique, 'community police' formed by groups of residents were locally recognised as having the right to solve problems of criminality, including issuing punishment. This was tolerated by the authorities -though there was an insistence that they had a duty to inform the police about what was happening in the neighbourhood.

Questions for practice
• If we support local solutions that meet people's needs directly, how do we balance this with rights-based approaches that focus on holding higher level authorities -which may be weak -to account?
• What are the ways of supporting local solutions rather than replacing them, and what are the risks in doing this?
• What are the implications for women in particular of local resolution systems and 'customary law' -are their rights infringed or are they unfairly treated by these systems?

How can we respond to this?
Practitioners that we spoke to gave these ideas: Identifying where self-reliance is a necessity. In some situations, selfprovision has developed due to unresponsive or predatory authorities, or in response to immediate needs. Where this is the case, self-reliance may reflect fear and/or be a question of survival and resilience. Identifying and working with communities to strengthen internal accountability within these local provision systems may be more appropriate than encouraging engagement with external authorities.
Understand expectations and preferences. Programmes may sometimes set out to encourage a demand for external service provision that doesn't exist at a local level. Identifying what issues there is a desire to resolve or what services are needed and already delivered at what level is the key, especially where conflict makes authorities less trusted or reliable. Strengthening accountability within a community may be combined with social accountability approaches to build trust with authorities over time, enabling programmes to respond better to community preferences.
Assess if self-provision is incentivised from above. In some contexts governments and other authorities might be perpetuating low expectations, self-reliance and self-provision to avoid accountability and demand for services. Supporting self-provision in these cases potentially lets duty-bearers off the hook. Programmes' responses to these drivers should be quite different from cases where self-provision is driven by conflict, fear or rejection of external authorities.
See self-provision as an asset. Self-provision initiatives can provide opportunities that can be built on. Some local initiatives around selfprovision may strengthen local solidarity networks and 'social capital', bringing together groups that may not have been connected before. They may also help to form group identities and support other forms of collective action and mobilisation. Shifting from self-provision to collective claim-making will depend on context, as well as on the levels of division that exist within the community. These claims might be channelled towards external authorities when conditions change for the better, for example when the level of violent conflict reduces.

Recognise risks of displacing existing solutions.
When new policies, services, or structures are established there is a risk that they may 'squeeze out' informal strategies of community protection and resilience that may have been working for many people, potentially providing a more reliable safety-net than new and often competing initiatives can develop in the short-term. Being conscious of these existing strategies is critical in minimising harm.

Be alert to inequalities or injustice sustained by self-provision.
Recognising self-provision as a common solution and in some cases as people's preference shouldn't blind us to who wins and loses from these systems. In some places women in particular may experience worse outcomes, for example through highly patriarchal 'customary justice' institutions. Other people may also be relatively disadvantaged compared to the rights they have in formal systems. Gender-sensitive analysis is crucial, as is engaging with women's rights groups to agree joint approaches to addressing any inequalities.
All three of our contexts have highly patriarchal social norms. Women reported needing to go through men for all kinds of services or to get issues resolved, including at a household level through male family members, as well as through mostly male intermediaries beyond the household level. Although we actively sought out female intermediaries to include in our research, the majority of those that were identified by households as important in resolving community issues were men.
In the three countries norms around women's leadership in formal spaces were different, but in general women holding official positions or being the 'go to' intermediary challenged established norms. In some cases when they did play these roles their authority or breadth of responsibility was challenged and undermined by men. For example, in Mozambique there has been a push for the ruling party to be seen to include women in governance roles, but we found that once in these roles women were often side-lined or not given actual authority to make decisions.
In cases where women did play important roles as intermediaries, this was sometimes limited to issues that were seen as concerned more with women's than men's 'domains'. Examples of these are domestic disputes, sexual violence, access to health services and child welfare. Some of our intermediaries worked on these issues, often quite visibly in parallel to male intermediaries, seeing their role as being a voice for women and to make sure other women were treated fairly.
The fact that women very often need to engage with or rely on men who hold greater power within both formal and informal governance structures is not particularly surprising. However, it is important to recognise explicitly if public policy or development programmes expect women to be able to access services or entitlements unaided or on an equal basis to men.

PRACTICE QUESTIONS
WHAT WE FOUND HOW WE RESPOND?

RESEARCH EXAMPLES
Women typically need to engage with local patriarchal power structures. Despite this, women intermediaries are frequently viewed as successful.
In supporting women's leadership in community roles we try to take a 'Women Lead' approach where they drive the agenda, assess and identify needs and solutions. This approach means we have less control in identifying outcomes up-front for donors. Instead we focus on the value of the process and on building the evidence that this approach results in practical outcomes that are more sustainable and transformative in their impact.

PRACTITIONER REFLECTION
Women intermediaries also used gender norms to their advantage at times, and were often seen as more appropriate leaders than men. In Myanmar women have historically taken up some village leadership roles during periods of conflict, because they are perceived to be braver and more reliable than men, but also because cultural taboos over harming women were seen to protect them more from armed groups. Similarly, one of the female intermediaries in Pakistan mobilised other women to protest specifically because they would not be physically attacked by men in public.
Some other women were able to play wider intermediary roles, beyond these issues. Often in fact they were regarded as more successful than men in these roles, and widely respected for that. On the one hand, the 'barriers to entry' for women to these roles appear higher; in almost every case they were able to establish themselves as intermediaries because they had special characteristics associated with their family, status and class or a personal history of determination and campaigning to resolve issues. On the other hand, women intermediaries have a wider reach within populations, as they do not face the barriers that male intermediaries do in dealing with women as well as men.
Women intermediaries were sometimes able to mobilise others -both women and men -to protest or take collective action. For example women intermediaries played a leading role in Mozambique in blocking a road in protest at development issues, and in Pakistan in staging an occupation of disputed land. However we also found that when women played intermediary roles, they often drew on more indirect strategies in order to navigate a patriarchal environment. For example in Pakistan women's exclusion from decision-making meant that women intermediaries sometimes relied on males within the household as their proxies. In Myanmar, some women intermediaries drew on the protection and status that derived from their association with powerful men.

FINDING 5: LOCAL PATRIARCHAL POWER STRUCTURES
Our organisation actively supports informal female roles as an entry point for facilitating their access to formal roles (if this is what they are seeking). We start where they are at and build from there.

Questions for practice
• How can we support women intermediaries where they do exist?
• How do we identify and try to break down the barriers to entry that face women seeking to exercise this kind of leadership?
• How do we work out whether to support alternative roles -such as informal 'female' roles operating alongside the 'male' system -rather than getting women into the main intermediary roles?
• To what extent should we seek to include the existing leadership, which is often male dominated, rather than creating space for new forms of leadership to emerge?

Examples from our research
• In some locations in Myanmar, Women leaders and groups were important, and had parallel structures. These, for example, brought cases of sexual assault to the attention of authorities, or were used to ensure a woman from the group was present when other decisionmaking bodies were making judgements in cases involving women.
• In Mozambique a husband and wife team led the residents' association that was competing with a Frelimo-linked group to represent one local community and resolve issues resulting from displacement by a mining company. Whilst the woman was regarded as the leader, and relied on her own personal and political connections to get things done, her role was not official, and her husband was the one who was named as director of the association.
• In Pakistan, one particularly effective female intermediary in a rural location was known to have deep connections with informal and formal authorities and to be able to get problems solved -although she often had to send her son to represent her in decision-making spaces that are customarily reserved for men.
• In the case of a group of children trafficked to China from a border town in Myanmar, their parents informed a Kachin woman intermediary who is also the wife of a leader in the ethnic armed group. This intermediary is well respected and speaks Chinese; rather than going to the ethnic administration, she secured the return of most of the children by using her influence with a contact within the Chinese police. Think creatively about overcoming elite bias. We should accept that in some contexts women who can take on leadership and intermediary roles will tend to come from privileged backgrounds or have more access to assets and resources through their families. Whilst we might prefer to see women from more diverse backgrounds in these roles, supporting those who already have recognition may provide a path for others in the future. We can also work with more privileged women leaders on broadening their approaches to inclusion and supporting others.

How can we respond to this?
Practitioners that we spoke to gave these ideas: Identify women leaders. Policy and programme analysis should focus specifically on identifying who women rely on and trust to raise their issues. Women intermediaries may not be widely recognised as leaders and may not have the same titles or roles as male leaders within a community. Particular efforts should go into identifying informal women leaders and understanding the power dynamics surrounding them.

Support women's informal leadership positions:
The characteristics that are associated with intermediaries or other authorities, and the associated public recognition and reputation as a leader, can often be gained through more informal roles. Actively supporting informal roles that women take up as representatives or through cooperatives or unions can be a useful entry point to develop potential for exercising future leadership in other roles. This also provides more representation for women within communities, and might provide some counterbalancing power to largely patriarchal structures.

SIX CHALLENGES AND DILEMMAS
CONTENTS PAGE 1 1 In our discussions with practitioners a number of cross-cutting challenges and dilemmas emerged. We tackle six of these here.
Our research also suggests that parallel systems might not end up being very different in terms of power dynamics. Focusing on new systems might also miss opportunities to make existing ones more inclusive.
One 'middle way' suggested was to engage with existing informal actors on moving towards more inclusive behaviours, which could provide an opportunity to strengthen local accountability. Associated with this is the risk that supporting localised solutions to specific issues undermines rights or lets external duty-bearers off the hook, particularly where higher level authorities have a formal responsibility to act.
What does this complexity of governance actors and systems mean for where we focus our efforts?
Our findings show how complex and intricate decision-making can be at a community level, in particular in places where there is a backdrop of conflict and a history of under-provision. It is not only that processes are complex, however, but also that the underlying politics is complex. Policy and programme actions might not result in what was intended, informal or unexpected decision-makers might appear in the way, and our actions may alter the politics in unpredictable ways.
Do we work 'with the grain' or challenge it -and at what costs for rights-based principles?
A common school of thought is that policies and programmes should 'work with the grain' and engage with ways that things are already done and with prevailing power structures in particular contexts or locations. In some ways this is a natural conclusion of our research findings on the depth and importance of these practices and institutions.
The quickest way to achieve things, and with a form of localised legitimacy, might be supporting established practices. However, we have shared a number of negative or potentially discriminatory aspects of these local governance practices, and working through existing power structures might solidify them or even undermine some core policy-making and programming principles. For example, the unequal access that women had to services and decision-making, or more predatory forms of intermediation identified, could be exacerbated by 'working with the grain'.
An alternative approach practitioners saw was to develop parallel systems that are more inclusive, or to invest in new forms of leadership and community representation to emerge -thereby attempting to disrupt established (and patriarchal) power structures. Although potentially more transformative over time, this is probably less likely to solve immediate problems.

CONTENTS PAGE
One response to this might be to side-step the complexity and instead focus efforts on more standardised, formal or predictable parts of the governance or service delivery system. Some practitioners suggested that they had already seen this happening in practice, as programmes responded to recommendations to work on issues with the greatest chances of success and traction, and the fewest obstacles to achieving change.
However, the problem of how policies made at a higher level are actually experienced by people on the ground remains, as these complex local systems will continue to shape who gets access to what, when and how. The same issue could apply to decisions not to work in conflict-affected or more fragile settings, which of course raises important issues of equity and exclusion.
Working out which issues are best dealt with at a community level rather than through engagement with a higher authority is challenging. Some practitioners noted that engaging with the granularity of local governance practices and actors can be a distraction from larger problems of public service delivery, and that there may be a risk of romanticising what can be achieved at a local level. For some issues, working directly on higher-level structural determinants might be required. Our findings suggest that this should not necessarily be a binary choice.
How do we weigh up the risks around engaging in politics?
Key to these findings is the inherently political nature of the authorities, the intermediaries and the networks through which they interact. Politics presents a challenge to many development actors, who are explicitly committed to staying apolitical. Existing approaches have often been carefully crafted to avoid politicising the work, which leads to programmes engaging largely with state officials or civil society actors who are divorced from local politics. Some of our respondents suggested that on the ground, many implementing staff incorporated working with the local politics into their practice, even if they then down-played this in formal reporting.
To create meaningful change we need to recognise the separation of development work from politics as artificial, and engage more with local politics. Doing so would enable us to bring back into our thinking the political ambitions, the incentives and the potential divisions amongst different actors -which may offer opportunities, as well as giving us a more holistic picture of the political economy we are working within.
But this does come with a trade-off around levels of risk. There are risks associated with both perceived or real politicisation of organisations and development work. Many organisations and donors have a low level of tolerance for engaging with political entities due to these risks as well the need to sustain the impartiality which is often required to be effective in other areas such as humanitarian or bilateral government engagement.
4 What does this mean for working at scale?
Concepts of working to scale and value for money are becoming increasingly important as development funding decreases. The diversity of local practices and actors that we found clearly challenges the assumption that we can pilot an approach in one locality and then simply 'scale it up'. Instead, identifying the diversity of local governance networks suggests that smaller-scale and more contextually-tailored approaches should be applied.
Questioning how we view scale may help us overcome these challenges. Starting from localised analysis and interventions in multiple locations, we can identify commonalities and trends across these localities. This will allow us to focus on issues or solutions that are common to multiple localities and can be addressed at higher levels through issue-based projects. This would also enable issue-based projects working at national or provincial levels to be informed by different ways of working at a local level.
Supporting plurality or strengthening convergence?
A diversity of authorities can lead to better outcomes for community members where it provides a degree of choice, or perhaps competition between actors. But diversity can also create confusion and reduce the scope for accountability that comes from having a clear duty-bearer.

6
There is a spectrum of options to respond to this challenge -from supporting convergence between multiple authorities as a way to simplify a governance system, to supporting strategies that allow a multiplicity of actors to coexist and interact, to focusing on linkages between formal and informal systems. It is important to consider the trade-offs in each situation before deciding at which point along this spectrum a policy or programme will focus its efforts. Since this is a central dilemma faced by many peacebuilding actors, looking at the pros and cons identified by work within this field could be helpful.
Who benefits from the status quo?
A key consideration to help support decisions on how to respond to these findings is a deep understanding of who benefits from maintaining the status quo. Powerful actors such as companies, militaries, criminal organisations or even intermediaries themselves may have vested interests in keeping the formal system weak. Supporting informal solutions may be perpetuating their power. In other situations, the status quo may be a consequence of conflict, history or other political legacies. Better understanding of the reasons why the formal system is weak will help locate our work in broader historic, political and social ecosystems.
The answer to this question can help guide decisions on how to engage and what systems we want to strengthen.