Constitution-Building in Deeply Fragmented States

This discussion paper highlights the challenges faced by states where control over the territory is contested by various groups, leading to fragmented governance. It emphasizes the importance of understanding governance structures at the local level, and their potential impact on peace- and constitution-making processes. It argues that peace- and constitution-making processes often overlook the diverse forms of local governance that exist in these contexts and, as a result, may set up arrangements that are inappropriate or ineffective in some, or perhaps many, localities.

The paper also emphasizes that the variety, complexity and fluidity of substate—mainly local but also regional—governance arrangements in conflict-affected states makes any engagement with them and assessment of their potential impact on peace- and constitution-making processes a difficult task.

The analysis in the paper is structured around six key dimensions of local government that draw on those identified by Mampilly and Stewart (2021) to describe variations in local ‘rebel’ political institutional arrangements.

1. Source of authority. This dimension examines the basis of authority in local governance structures. In stable democracies, authority typically stems from direct elections. However, in conflict-affected areas, elections are often not possible (or desired), and authority may derive from coercion and be legitimized by invoking past elections, through pre-existing or newly developed sets of rules and practices, or through charismatic leadership. The paper also discusses the fluidity of local authority in these contexts and emphasizes the need to understand changing dynamics in establishing arrangements that involve local authorities.
2. Level of institutionalization. Institutionalization refers to the degree to which governance is based on established rules and norms. It also considers the complex interplay between formal and informal governance systems at the local level, which underpins its variability and affects the predictability and consistency of governance.
3. Integration of civilian authorities and armed actors. This dimension explores the relationship between (and potential integration of) civilian authorities and armed groups. The paper identifies a spectrum of integration, from complete control by armed groups to collaborative governance. It examines the way in which the balance between civilian and armed authorities is managed in different contexts, and the implications of this for governance and stability.
4. Degree of continuity and change. Are local governance structures long-standing or newly created? The discussion paper examines how new authorities may maintain, transform or replace existing institutions. It argues that the degree of continuity and change is influenced by the level of control new authorities have over an area and their long-term political objectives and highlights the adaptability of local governance structures to changing circumstances.
5. Level of public inclusion and accountability. This dimension assesses the involvement of different social groups in decision making and the accountability of local authorities. The paper notes the varying levels of public inclusion and accountability in different contexts. It concludes that inclusive governance can build legitimacy and trust, while exclusion can lead to instability and conflict.
6. Effectiveness or responsiveness. Effectiveness measures the degree to which governance structures meet the needs of the population. The paper notes the challenges of service delivery in conflict-affected areas and examines how basic services such as security, health and education are provided (or not) in different contexts, and the impact of governance effectiveness on public perception and support.

The discussion paper illustrates the way each dimension may be used by drawing on case studies from three countries, Myanmar, Somalia and Yemen, that are currently experiencing high levels of conflict and fragmentation, and whose central states (or governments) are violently contested. By focusing on these contexts, the authors seek to draw attention to the importance of understanding local governance arrangements properly when central state-level agreements are made and constitutions are written. The paper suggests that a focus on the dimensions it uses provides peacemakers and constitution makers with a better understanding of the complexities of political arrangement in the country and thus enables better constitutional choices, informed by realities on the ground.

In cases of protracted conflict, when control of a state (or the state itself) is contested, different parts of the state may be controlled by antagonistic groups, each of which asserts authority over and (to a greater or lesser extent) governs the areas it controls.¹ Peace and any ensuing constitution-making processes usually need to engage with the forces controlling such areas. However, such processes are typically driven by central state actors and focus on central government-level arrangements (including in certain cases the relationship between substate levels and the centre) and seldom take account of the nature of governance within these areas. Nonetheless, regional and local (or, collectively, substate) governance arrangements may inform the way in which central state political arrangements will function—including their chances of success.

In other words, substate arrangements should be considered in designing both peace agreements and constitutions, as peace- and constitution-making processes would benefit if they were informed by an understanding of the diversity of governance structures at the substate level. Understanding substate structures can, for instance, lead to better-informed decisions about whom to include in these processes, as well as on the content of peace agreements and new constitutional frameworks. Hence, the focus of this paper is on how substate, and particularly local, governance may be better understood and how such an understanding could help in securing a political settlement that will endure.

The variety, complexity and obscurity of regional and local governance arrangements in conflict-affected states make engaging with them and assessing their implications for the future in peace- and constitution-making processes a daunting task. Moreover, although there are studies on governance by rebel groups, little work has been done on local and regional governance during conflict. To initiate discussion about how to ensure that issues concerning regional and local governance are adequately considered in peacemaking and constitution making, this paper offers an analytical framework with six dimensions for understanding local governance in conflict-affected states. It proceeds to illustrate the way in which the framework may be used by drawing on a number of case studies from three countries— Myanmar, Somalia and Yemen.²

Chapter 1 of the discussion paper sets out the six dimensions; Chapter 2 clarifies the terminology; Chapter 3 provides a brief introduction to the three countries from which the individual case studies were drawn; Chapter 4 contains an examination of the case studies using the proposed dimensions; and Chapter 5 concludes with suggestions for what the outcomes of the research on local governance may mean for negotiations over (state-wide) political arrangements, including the constitution.

Substate governance in conflict-affected settings is often difficult to define. The standard distinction between regional (the level immediately below the central state) and various levels of ‘local’ government may not hold, as arrangements are fluid, authority shifts as power relations change and the local is always embedded in larger regional dynamics. This paper and the case studies that inform it are, nonetheless, primarily concerned with government at the most local level and touch on regional issues only when necessary to understand local dynamics.

The six dimensions proposed in the paper to help understand local governance are:

1. The source of authority.
2. The level of institutionalization.
3. The level of integration of civilian authorities and armed actors.
4. The degree of continuity and change.
5. The level of public inclusion and accountability.
6. The level of effectiveness or responsiveness.

These dimensions draw on those identified by Mampilly and Stewart (2021) to describe variations in local rebel political institutional arrangements, which include the extent to which power sharing occurs, the depth of integration of rebels and civilians, the extent of innovation and the degree of inclusivity. The authors also mention two additional dimensions—the level of formalization and the overall effectiveness of political institutions. The dimensions, which approximate Mampilly and Stewart’s, need to be understood, just as theirs, as ‘spectrums with … groups more (or less) closely approximating two extremes of each dimension and not as binary outcomes’ (2021: 19).

As explained in the conclusion, these dimensions have been chosen as tools for understanding how substate and particularly local governance operates. The different dimensions prompt analysis that helps one understand not only the potential for democratic processes in the future, but also the ease or difficulty of changing existing arrangements and implementing broader political settlements. At the very least, these dimensions should be considered when observing substate governance, in particular when key actors attempt a remake of state structures in contexts of fragility and fragmentation.

The case studies highlight a wide range of forms of local governance, including rebel governance,³ customary or traditional governance, and state-led or state-backed local governance. This acknowledges that while some groups may oppose central state structures, others may see benefit in supporting them, and that opposition or collaboration is often strategic and volatile in nature (Kasfir, Frerks and Terpstra 2017: 259). Including state-led or -backed arrangements also allows the paper to reflect on the fact that central states or central state governments often ‘opt for a pragmatic form of engagement that allows them to govern with, or even through, those they consider rebel groups’ (Kasfir, Frerks and Terpstra 2017: 271–72; see also Joshi 2023; Fraihat and Alijla 2023). This is what has elsewhere been called the ‘mediated state’ (Menkhaus 2006, 2014; Stel 2017; Kasfir, Frerks and Terpstra 2017).

A number of concepts used in this discussion paper need further definition.

State fragmentation: The paper follows Sempijja, Chegraoui and Kiros (2024: 3) in their understanding of state fragmentation as a ‘process by which a central government loses control over parts of its territory, leading to the emergence of semi-autonomous regions or competing power structures within a state’. In their view, state fragmentation is often a result of state fragility, which allows ‘non-state actors or regional entities to assert authority, effectively dividing the state into distinct, often conflicting, entities’. Fragmentation in this context is often driven by internal conflicts, ethnic divisions and the state’s inability to provide essential public goods, leading to a de facto division of the state into smaller, often conflicting entities (see also Fearon and Laitin 2003). The impact of fragmentation on governance can depend on a number of different factors, including: (a) in conflict-affected areas, the type of conflict and level of violence, and whether there are processes in place to manage them; (b) the degree to which parties are involved in or dependent on a ‘war economy’, and in general the extent to which the economy functions through illicit means; (c) the level of government consolidation at every level of government; and (d) the degree to which an administrative system actually exists and at which level or levels.

Local governance:⁴ Most states have multiple levels of government, from the central to the local or municipal level. Constitutions that establish systems of multilevel government provide for at least two different levels of government—the central level (referred to as the federal level in some federations), and at least one substate level. Some constitutions provide for three or more levels. Names given to different levels of government vary widely. Moreover, in some countries there will be more levels in some areas than in others—asymmetry is relatively common. Generally, in this paper, the terms ‘local-level governance’ and ‘local governance’ are used to refer to the lowest levels of governance, which may be referred to as the village, traditional community or municipal level. The term ‘regions’ is used to refer to larger territorial divisions, variously called provinces, states, regions, governorates or cantons, etc. Local and regional governance may be referred to collectively as substate governance.

Formal/informal forms of governance (or state and non-state forms of government): The paper shares the understanding of institutions used by Helmke and Levitsky (2004: 727) among others as ‘rules and procedures (both formal and informal) that structure social interaction by constraining and enabling actors’ behavior’. In the countries with which the paper’s case studies are concerned, as elsewhere, there are areas governed under systems that are not designed by the central state or subnational lawmakers, but are well anchored in their communities. For want of a better term, these arrangements are referred to in this discussion paper as ‘informal’ even when they are widely recognized, understood and respected, as they are in many cases. They include long-standing customary governance arrangements with deeply entrenched practices and religious provisions (see also Seidel 2025). According to Helmke and Levitsky (2004), these informal institutions are to be distinguished from (a) weak institutions, (b) informal behavioural regularities and (c) informal organizations. Furthermore, they exercise governance functions that the state would otherwise be expected to fulfil (Khan Mohmand 2016: 6).

In 2024 the following countries made it into the top positions in the Fragile States Index—Somalia, South Sudan, Sudan, Syria, Democratic Republic of the Congo, Yemen, Afghanistan, Central African Republic, Haiti, Chad and Myanmar.⁵ All of these countries have fragile, fragmented and violently contested central states (and governments). Most of them are also subject to constitutional uncertainty, or lack of implementation of constitutional arrangements, and negotiations of varying intensity (often intermittent) over new political settlements.

By focusing on Myanmar, Somalia and Yemen, the intention is to show that, despite the central state’s inability to govern the country’s entire territory, and despite the absence of a unified central government, substate governance and governance arrangements continue to exist, although with varying degrees of effectiveness. In some parts of these countries both regional and local-level arrangements are under the control of authorities recognized as the central government—or, as is the case in Myanmar, a military junta that has captured the central state. In other areas, the substate level is under the control of revolutionary forces or emerging substate authorities. Substate authorities in these instances can have more or less tenuous links to the (proto) central state. Lastly, there are instances where rebel groups or revolutionary forces have control over parts of the state and govern with autonomy, separate from other governance arrangements, such as the Arakan Army-controlled Rakhine State in Myanmar or Ansar Allah in Yemen.

In what follows, the paper briefly summarizes the current context in the three countries with which it is concerned. An analysis of their diverse local governance arrangements follows in Chapter 4.

3.1. Myanmar

In Myanmar, an incipient decade-long transition to democracy ground to a halt on 1 February 2021 when the military, once again, removed the elected government and (unconstitutionally) declared a state of emergency.⁶ The coup d’etat was initially met with peaceful demonstrations, but, after a violent crackdown, these quickly evolved into an armed insurgency. Some ethnic armed organizations (EAOs) that had been fighting the military for decades, and controlled territories on the state’s peripheries, became part of the pro-democracy movement and gained support from the broader population. Thus, many Myanmar citizens took up arms (often supported by those EAOs that decided to continue fighting the military) or went into exile, and some set up government institutions in opposition to the military junta.

While the junta has not yet suspended the 2008 Constitution of the Union of Myanmar, its actions tell a different story, as the Constitution that it purports to uphold has largely been neglected. Shortly after the coup, a group of elected members of parliament (most from the dominant political party, the National League for Democracy), operating largely from outside the country, established the Committee Representing Pyidaungsu Hluttaw (CRPH)—Pyidaungsu Hluttaw being the (disbanded) Union Parliament—and released the Federal Democracy Charter (FDC), which includes interim governance arrangements as well as a roadmap towards a federal democratic union, and is intended to supersede the 2008 Constitution (International IDEA 2022). The FDC was drafted and supported by not only those elected members of parliament, but also a number of representatives from civil society organizations, other political parties and other (ethnic) resistance groups. As its name suggests, a central feature of the FDC is its promise of federalism for Myanmar in the future. Until now, Myanmar has been divided into seven regions, located in the centre of the country and where the majority Bamar population resides, and seven ethnic states, located on Myanmar’s periphery and populated by a myriad of ethnic groups.⁷ The FDC does not specify how many regions or states there would be in a future federal Myanmar, providing only that Myanmar will be composed of sovereign ‘states’ or ‘federal units’ with the right to draft and ratify their own state constitutions. Under the FDC, these substate constitutions will specify the powers and potential revenue sources of both regional and local government units (FDC, Part I, section ii and section iii, article 13), with the ‘subsidiarity principle’ as a key element of the future system of power sharing (FDC, Part I, section II, article 4). Beyond this, the FDC remains silent regarding the future federal design, and does not include any other provisions related to local governance.

The FDC also provides for the establishment of a People’s Assembly,⁸ a deliberative body, which, in turn, approved the formation of the National Unity Government (NUG). (The NUG had been previously established by the CRPH.) The People’s Assembly also recognized the National Unity Consultative Council (NUCC)⁹ which, according to the FDC, is to lead the implementation of the FDC, and the deliberation on and drafting of the Transitional Constitution (FDC, Part II, Chapter 3). While the timing is unclear, the People’s Assembly is intended to ratify the Transitional Constitution. In the meantime, several constitutional proposals have been developed outside the NUCC, some of which were submitted to the NUCC and some not (International IDEA 2025b). Since September 2004, however, under NUCC auspices, NUG representatives, ethnic resistance organizations (EROs)¹⁰ and ethnic political parties took a step back to negotiate the Articles of Federal Transitional Arrangement (AFTA). The idea is that the AFTA will be submitted to the NUCC, and, following a consultative process, a national conference will then adopt these principles as the federal transitional constitution (International IDEA 2025a).

At the same time, the ongoing conflict and the junta’s control of all major cities in Myanmar’s regions (including Yangon, Mandalay and Nay Pyi Taw) have meant that NUG advances in the control of urban areas—it already controls most rural areas in the country’s regions—have been slow. This has been the case despite the military support and guidance provided by various EROs. In the meantime, however, the NUG has created a Central Committee for Interim Implementation of Local Administration, responsible for establishing interim local governance structures in seven regions,¹¹ and is increasingly collaborating, particularly on service delivery, with EROs that control states’ territory. Despite the latter, stakeholders are still in the process of negotiating the future structure of Myanmar, which is as uncertain as the role of the different substate governments in the anticipated constitution-making process.

Since the coup, EAOs—of which there are more than 20, primarily active in Myanmar’s peripheral ethnic states, not all of which support the democratic resistance against the military—have significantly expanded the territory they control, sometimes into territory inhabited by other ethnic groups or the majority Bamar. Some of these EAOs, in particular those representing groups like the Kayah and the Karen, have developed complex systems of government, while others have more rudimentary structures. Those that have created governance structures have also developed their own substate constitutional frameworks, with some amending previous drafts of substate constitutions that had been developed in the early 2000s and, in the case of many, were never made public—and others developing new constitutional frameworks. Since the 2021 coup, some of Myanmar’s regional stakeholders have followed suit. The level of progress in developing these constitutional drafts, however, varies across states and regions. Some stakeholders are still engaged in negotiations to establish a drafting body (for instance, in Kachin); some have set out the drafting process but have not yet started with the actual drafting (such as in Yangon); some have already started with the drafting (for example in Magway and Mandalay); and still others have finalized the drafting and have adopted their (interim) constitutional frameworks (such as in Karenni). Sometimes several groups within the same territory have drafted constitutional frameworks in opposition to each other (International IDEA 2025b). While these regional stakeholders are for the most part closer to the local level than their central state counterparts, the relationship between the regional and the local varies significantly throughout the country. Among other things, this relationship is defined by institutional capacity at the regional level, by the regularity of interactions between the regional and the local level, and by whether the territory is located in an area controlled by the military junta, the NUG, EAOs or other groups.

3.2. Somalia

Somalia has been in a constitutional transition since the Provisional Constitution was adopted in August 2012—the third interim constitution since 2000. The Provisional Constitution was based on a 2004 elite bargain that resulted in a first (proto) federal transitional constitution (Menkhaus 2018). Building on the latter, the Provisional Constitution is the result of intermittent negotiations between Somali elites representing all major clans and factions and persistent pressure from the international community (Menkhaus 2018). It includes provisions that regulate its own review process and the development of a federal constitution of Somalia.

The Provisional Constitution establishes an Independent Provisional Constitution Review and Implementation Commission and a parliamentary Oversight Committee (to which the commission is to report). The review process has seen many delays, however. These delays led to the establishment of the National Consultative Council in September 2015—an informal gathering of the president of Somalia and the presidents of the Federal Member States (FMSs)—which managed to reach some agreements on constitutional reform despite increasing challenges to its legitimacy (Mogadishu24 2024). While some of these agreements have been put into effect, the involvement of Somalia’s FMS leaders—each associated with a different dominant clan¹²—in decision-making institutions, and the weakness of the central government, has been a significant obstacle to reaching agreements that could benefit Somalia as a whole.

As noted above, a central element of the Provisional Constitution is that Somalia should be a federal state. It provides for the formation of new states and for the involvement of existing and new FMSs in the conclusion of the constitutional review process. Except for Puntland, which was established as an FMS in 1998, the Provisional Constitution requires all other FMS to be formed from the union of at least 2 of the extant 18 regions. As a result, although not according to the procedure originally agreed, Jubaland was constituted in April 2013, Southwest Somalia in June 2014, Galmudug in July 2014 and Hirshabelle in October 2016. Under the Provisional Constitution, the new states are to be involved in determining the allocation of responsibilities between the federal government and the FMSs (article 54), and to adopt their own constitutional frameworks (article 120). Each of the newly formed FMSs has adopted its own (interim) constitutional framework, often stretching the few substantive constraints on them theoretically imposed (though mostly unenforceable) by the Provisional Constitution. The Provisional Constitution makes an exception for Puntland, providing that it may ‘retain and exercise powers endowed by [its] own State Constitution’ until such time that all FMS constitutions are ‘harmonized with the Somalia Federal Constitution’ (article 142).

The capacity of the FMSs is limited, as is the control of their governments over their territory; in some places claimed territory is itself contested by other FMSs and by networks of clan militias, strongmen, the al-Qaeda-linked Al-Shabaab (Menkhaus 2014: 142) and, increasingly, the Islamic State in Somalia. As a result, a patchwork of fluid, variable and very much contested local governance arrangements has developed, partly managing to provide a level of basic services to communities in need (Menkhaus 2014: 143–44).

In 2013, the Somali Parliament enacted Law No. 116, establishing a local government framework. Subsequently, in 2016, the federal government introduced the so-called ‘Wadajir Framework’, a non-binding guideline designed to assist in the formation of district governments. Despite these efforts, however, establishing local government institutions along the lines proposed by the federal government has faced significant challenges.

3.3. Yemen

In 1990, the Yemen Arab Republic (YAR) and the People’s Democratic Republic of Yemen (PDRY) unified to form the Republic of Yemen (RoY). The YAR had been established by the overthrow of the royalist Imamate regime in 1962 and the PDRY in 1967 when British colonial control of the territory ended. After unification, a Yemen Constitution was approved by popular referendum in 1991. Perhaps predictably, tensions rose between the leaders of the two former regimes almost immediately, leading to a short civil war in 1994 between the Sana’a-based authorities and southern separatist elements. This was decisively won by Sana’ani forces, laying the foundations for one of the current conflicts in the country. In the following years, the YAR ruler, Ali Abdullah Saleh, who became President of the RoY, successfully undermined the democratic elements of the Constitution to bring about a regime where his rule would continue largely unchecked.

The first decade of the RoY failed to fulfil the hopes of prosperity raised by unification and the discovery of oil in the mid-1980s due to a number of factors which led to deteriorating living conditions for the majority and worsening poverty (by 2014 Yemen ranked 160th out of 188 countries on the United Nations Development Programme Human Development Index (UNDP 2015: 50)). Relevant elements were the economic and financial constraints resulting from the state’s position on the 1990 Iraqi invasion of Kuwait, restrictions on the democratic processes of the regime of Ali Abdullah Saleh and the state’s encouragement of local conflicts, alongside rampant corruption. By the early 2000s, the Huthi movement rose in the far north of the country, with a series of six armed conflicts against the regime headed by Saleh between 2004 and 2010. Discontent with the new ruling processes in the former PDRY led to the emergence of a separatist movement in 2007.

These factors combined with worsening poverty, political paralysis, the postponement of parliamentary elections and Saleh’s intention to pass on the presidency to his son, leading to the popular uprising of 2011. In the course of that year, the regime was challenged both by the popular movement and by factions that had deserted it. The international community intervened to prevent the outbreak of a full civil war, and the regime headed by Saleh was replaced by a transitional one intended to bring about democracy and due to last only two years, from 2012 to 2014. The transition was to encompass a government of national unity, security sector reform and a National Dialogue Conference (NDC) whose task was to set the scene for a more representative polity and prepare for a new constitution. Following the unopposed election of Abd-Rabbu Mansour Hadi as President in 2012, the NDC ran for 10 months from 2013 to 2014 and was followed by a Constitutional Drafting Committee which had completed its work by early 2015. The federal structure proposed by the committee was constrained among other things by the requirement imposed by President Hadi that it be based on a particular territorial map; it satisfied neither the Huthi movement nor the southerners. It was the final straw for the Huthi movement, which took over the reins of power in the capital Sana’a. The internationally recognized government (IRG) relocated to Aden, which it named the temporary capital, although, for most of the period since, its main leaders have operated from exile in different Arab capitals.

Full-scale war started on 26 March 2015 with the launch of the Saudi-led coalition airstrikes intended to return the IRG to power in Sana’a. Front lines were soon established, with Ansar Allah (the Huthi movement) controlling about a third of the country’s territory but two-thirds of its population, while the IRG had tenuous control of the rest. The economic situation has worsened dramatically over the past 10 years. The international coalition’s direct military involvement was on hold from April 2022 when a six-month truce was negotiated by the UN, but US (and recently British) bombardments resumed in early 2025. Since the end of the truce, fighting on the various fronts has continued in Yemen, although in a limited way and involving Yemeni forces only.

Direct negotiations between Ansar Allah and the regime of Saudi Arabia were close to successful conclusion when the Gaza war started in October 2023, leading the Huthis to launch a series of attacks on shipping in the Red Sea and on Israel itself, in support of Palestinians. International involvement in fighting the Huthis has since shifted from the Gulf states, which have remained conspicuously silent, to the United States and United Kingdom, which launched air strikes against the Huthis in early 2024. In March 2025, US President Trump started a 52-day air offensive, which ended with a ceasefire agreed between the Huthis and the USA, mediated by Oman. Renewed expansion of ground fighting remains a serious risk.

Decentralization (but not federalism) had been on Yemen’s agenda for a long time. As Lackner (2025) records, it was the focus of political struggles from the 1990s, and in 2000 a Local Authority Law was adopted applying to the country’s 22 governorates (regions) and 333 districts. However, this law, and a 2008 amendment to it, gave little real autonomy to substate units, particularly because local authorities were not allocated the resources to implement their responsibilities (Lackner 2025). Currently, local government in the Ansar Allah areas is highly centralized. The area under the IRG is fractured, however. Thus, Aden is the stronghold of the Southern Transitional Council (STC), which, although part of the IRG, is committed to southern independence. Somewhat similarly, the Hadhramaut region’s relationship to the capitals is distant (Lackner 2025). As Lackner concludes, ‘governorates have their own specificities, which contribute to the overall complexity of the situation, further marked by the dominant political/military leaderships involved in each area’ (Lackner 2025: 23).

As can be inferred from the previous chapter, weak and contested central states may sometimes incentivize territorially concentrated groups or regional stakeholders to develop new (or adapt old) substate governance arrangements, in an attempt to both strengthen institutions and legitimize their authority, and perhaps to also bring coherence to the local level. What follows is an exploration of the diversity of local governance arrangements in these fragile and fragmented settings, using the dimensions introduced in Chapter 1.

4.1. The source of authority

In stable democracies, the primary source of local authority lies in direct elections. When holding elections is not possible (or not desired by those with decision-making power), and in the absence of a functioning state, generally the source of authority lies in the ability to coerce. This type of authority is sometimes legitimized by invoking past elections (even though they may have been held under a currently suspended/abrogated legal framework), through a pre-existing or newly developed set of (written or unwritten) rules, or through widely respected or charismatic leadership that may have a long history or a record of delivering services. At the same time, authority can be challenged and is indeed often contested, as the ‘social contract is broken, uncertain, or incomplete’ (Joshi 2023: 2).

In Myanmar, wards/villages and townships under the control of the military junta are governed by local administrations officially accountable to the General Administration Department, under the captured Ministry of Home Affairs. De facto they directly report to the regional military council. Some of the administrators, for instance elders who mostly govern according to unwritten rules, stayed in their positions from before the coup; others resigned and had to be replaced. In the latter case, to replace them, the junta appears to have chosen either individuals close to the military or (particularly in large cities such as Yangon) National League for Democracy officials with closer links to the communities they are to govern. In these areas, the military junta must balance local control with resistance to its rule. In other areas of the country, the military junta may collaborate with a select number of EAOs or with paramilitary militias (Pyu Saw Htee), which may support military-appointed authorities, or which may themselves govern areas under their control, with the connivance of the military.

In regions where the Myanmar military is not in control, local authorities are mostly subordinate to the NUG or to EROs. Forms of authority in these areas include: (a) civilian administrators representing the NUG-affiliated People’s Administration, which work in tandem with People’s Defence Organizations (PaKaPhas) and People’s Security Forces (PaLaPhas), and which at the village level will be elected in public meetings—at the township level authorities will be directly selected by the NUG Ministry of Home Affairs; (b) civilian administrators—often elders but sometimes also younger political activists—who govern supported by either EROs or (local) People’s Defence Forces (PDFs), or both;¹³ and sometimes (c) customary or traditional administrators, particularly in areas inhabited by smaller ethnic groups. In many instances, territorial competition between different armed forces—and principally between the military junta (and some of its satellites), and EROs and PDFs—leads to complex and sometimes conflicting governance structures, where civilian authorities may be in control only nominally while de facto it is armed groups that control (or fight for control over) both government and territory. The NUG has different degrees of control over both local administrators and their armed supporters in areas under its formal authority. This may lead to difficulties in resolving disputes between different levels of government, or between armed and civilian actors vying for control of an area.

In Somalia, legislatures and executives at the FMS level have been indirectly elected since their establishment. These elections are underpinned by unwritten clan agreements, whereby clan elders choose their representatives for the state parliament and the state parliament then chooses the (state’s) president (ICG 2024a). Sometimes, as in Puntland, the presidency rotates among the dominant clans. Because most FMS governments are not in control of swathes of their own territory, at the local level there have been pilot direct elections for certain local councils only—for example, in Puntland in October 2021 (expanded to 34 districts in May 2023). For the most part, the local level is governed by clan elders (or their appointees), clan warlords (leading clan militias) or, indeed, Al-Shabaab. In some local communities, administrators are appointed by the respective FMS president or by the president of Somalia, as appears to be the case in the Banadir region surrounding the capital Mogadishu (Davies et al. 2023). The most common source of authority is hence the acceptance of clan authority, but where non-state armed groups like Al-Shabaab are in control, their ability to use force establishes their authority.

In Yemen, the latest parliamentary elections took place in 2003, leaving current authorities with a clear democratic deficit. Nonetheless, Ansar Allah, controlling the capital, uses the members of parliament based there to endorse its decisions through standard parliamentary procedures, and even held elections to fill vacant seats in 2019. By contrast, the members of parliament affiliated to the IRG have been unable to meet regularly in the ‘interim’ capital Aden, as the separatist STC has prevented them from doing so.

Ansar Allah inherited the ministries and administrative institutions of the RoY and uses them and their staff to maintain its claim to legitimacy. However, to enforce its ideology and priorities, it has appointed ‘supervisors’ at all levels of the administration, who have full authority to control the work of officials and civil servants. Its authority depends on loyalty to Ansar Allah, and it is widely decried as being unqualified and lacking in commitment to the institutions it effectively controls.

In the IRG-controlled area, the deep divisions and rivalry between different leaderships reflect the fragmentation of the country along social, political, economic and sectarian lines. They affect the administration, and generalizations about the justification for power are difficult in the absence of meaningful ideological positions and political programmes. Remarkably, however, and presumably seeking some form of legal legitimization, throughout the country ruling factions claim to be following the 1991 Constitution, and provide various excuses for not implementing its procedures, usually based on the political, economic or military crises.

In summary, in the localities that the case studies examined—as in most, but not all, situations in which the state is fragmented—elections are often neither possible nor desired by those in control. Thus, rather than drawing their authority from a democratic process, governing authorities rely variously on often long-standing traditions that confer authority on local elites, on deals brokered among contesting groups, simply on the threat of coercion, or on a mix of these. At the same time, governance at the local level amid instability and conflict is, of course, fluid in the sense that in many cases groups frequently gain or lose control of territory, in turn having to vacate or build new local governance structures.

4.2. The level of institutionalization

This dimension is concerned with the degree to which local governance takes place through established rules and norms that are systematically and consistently implemented, sometimes though not always through distinct organizations (see Kahn Mohmand 2016: 9). In the framework used, institutionalization is absent if authority is exercised through the sheer, unconstrained exercise of discretion by someone or some group, even if this is mediated by ad hoc engagement with social groups or elites when it is necessary (or desirable) to build legitimacy. The level of institutionalization goes beyond the distinction between formality and informality, as defined above, in that informal norms can be regularized and predictable and can therefore be the basis of a highly institutionalized system.

In the case of Myanmar, while all military-controlled areas, including Yangon, allegedly function under the 2008 Constitution, as well as old or new legislation and executive orders, enforcement of the legal framework is not consistent or systematic. While the state legal system ‘was never fully accepted and entrenched in the wider society’ (Kyed 2024), it seems now to be en route to further ‘de-institutionalization’. When possible, as for instance in Yangon, administrators will strategically implement rules and regulations, balancing residents’ needs with orders from the military junta. This is evident when looking at the implementation of the military-junta-sponsored conscription law. Despite the fact that in the space of one year over 25,000 people have already been conscripted (MOHR n.d.), the conscription rate varies across wards depending on the individual administrator. The administrator’s flexibility often hinges on the community’s trust in the administrator, and in turn on the junta’s need to maintain community satisfaction. There are also administrators that are perceived to simply disregard rules and regulations for their own benefit—for example, by taking bribes for the resolution of land and water disputes, with the connivance of the military junta. Martial law is also only enforced in some parts of Yangon, particularly where the junta fears that the revolutionary forces may gain control. And wealthier parts of Yangon appear to suffer from fewer restrictions than others. At the local level—wards or villages—disputes are likely to be solved through negotiation and mediation by household heads, village or ward leaders or local administrators, as was the case before the 2021 coup.¹⁴ Although disputes that involve parties from different villages are more difficult to solve, they are equally unlikely to be referred to the captured state judiciary, which lacks citizens’ trust (Kyed 2024). In general, the arbitrary enforcement of rules and regulations, as well as unilateral and opaque amendments, contributes to a pervading sense of lawlessness and rampant corruption.

In parts of the country that are not controlled by the junta, there are attempts to institutionalize local governance, some of which are fairly successful. Generally, before the process of liberalization began in 2010, ethnic armed and non-armed groups had already developed their own institutions and codified rules, and established their own parallel justice and governance systems, with varying degrees of institutionalization (Kyed 2024). Since the coup, in some places those rules have been re-adopted and are being enforced. Many ethnic groups have hence turned to, and amended, previously drafted constitutional frameworks—most of which had been drafted in the early 2000s—or have developed new constitutional frameworks to regulate governance at the substate level (or within the areas they control).¹⁵ Indigenous people in control of smaller areas within Myanmar substate entities have done the same.¹⁶ One of the added difficulties in the post-coup period is that increasingly, as ethnic (armed) groups expand the territory under their purview, they may start to govern areas no longer exclusively populated by their own ethnic group. This may have implications on new institutions and processes, and their level of responsiveness to the needs and demands of other groups and individuals.

Those areas under the control of NUG-affiliated People’s Administrations have tried to maintain the structures and institutions that existed before the coup. The internal conflict, however, with the consequent displacement of people, has resulted in serious gaps in terms of human capacity. Beyond this, the NUG has little control over People’s Administrations (and PDFs) on the ground, some of which take advantage of their positions of power to run their own patronage networks. In May 2024, the NUG adopted guidelines for People’s Administration teams, but their ambiguous and vague language and the lack of enforcement mechanisms hinder their effectiveness in regulating local governance.¹⁷

In terms of dispute resolution, for the most part pre-coup customary-based rules, and mediation approaches, have continued to be the ‘first instance’ at the local level (Kyed 2024). In some instances, it appears that authorities at the local—village and township—level engage in discussions with household heads and elders on the development of new (or the adaptation of old, sometimes customary) rules and procedures aimed at improving governance at the local level. Their success in bringing these customary norms into line with human rights, as the FDC expects (Part 2, 49), varies. When the case is more serious, or relates to complaints against the administration at the local or township level, it will be referred to the NUG-appointed township or district court, but according to interviewees in Sagaing and Chin State, such complaints and subsequent referrals happen rarely.

Somalia’s local-level governance is generally based on clan structures; clan elders back local authorities and deal with disputes by using both customary practices and Sharia law. Sometimes, rival (sub)clans establish parallel local administrations in an attempt to increase their share of power (Wahiu and Abebe 2025). Some villages and districts are controlled by warlords (sometimes leading clan militias), and others are controlled by Al-Shabaab. Both warlords in general and Al-Shabaab in particular sometimes use clan elders as intermediaries with the population and sometimes also higher government levels (ICG 2022: 27). As noted above, FMS governments have only limited control over their territory, and while both the central level and federal units—which constitutionally have the prerogative of dealing with local governments—have developed guidelines and legislation to regulate local governance (e.g. to elect local councils directly or to redraw district boundaries), these are mostly disregarded. At the same time, the influence of customary practices on state institutions, and vice versa, is significant. Generally—and, depending on the context, with varying degrees of influence from central state or FMS government officials—clan elders select state representatives at all levels from among specific (sub)clans following intra- or inter-clan agreements. These positions in turn give clan representatives both political influence at the central (and federal unit) level, and the possibility of building patronage networks with members of their clans. Clan elders’ control over appointments also gives them the ability to undermine institutions from within without affecting their power over clans and territories (Wahiu and Abebe 2025). In other words, state and non-state institutions, including state law, customary law and Sharia, are tightly interwoven, and affect each other in their functioning and development.

In Yemen, in both Ansar Allah and IRG areas, local governance formally takes place on the basis of the RoY’s 2000 Local Authority Law. In practice, the situation is very different between the two ruling entities. In Ansar Allah areas, as noted above, centralized authority in Sana’a prevails, with governorate-level authorities subordinate to the central government; actually, rather than the formal ‘government’ with a president, prime minister and ministries determining policies, it is the Ansar Allah movement and its own ideological leadership that control the supervisors and therefore the actions of local administrative authorities. The practices adopted are not institutionalized and, as noted above, decision making is arbitrary, seemingly dictated by political whims.

The situation is very different in the IRG-controlled areas, largely as a result of the fragmentation referred to above. With a very weak, dispersed and divided central government, and the divisions both within the ruling Presidential Leadership Council of eight members, and the formal government under the prime minister, the quality of local governance varies from one area to another. The most important differences in IRG areas are between governorates. Despite the 2008 amendment to the Local Government Law that required the election of governors by governorate-level elected councils, they are once again appointed by the president. The offices of the different institutions and ministries remain in place, though with limited power and influence. At the district level, they are largely absent, as was the case in the pre-war situation. In this part of the country, informal authorities, particularly influential tribal leaders and other civil society figures, can have considerable influence and play an important role in the provision of justice and security. Informal governance in the form of community organizations, many of them tribally related, is nowadays the mainstay of whatever little stability exists at the community level in IRG areas. How institutionalized these are is difficult to determine.

In conclusion, across the case studies, the degree and form of institutionalization of local governance can be seen to vary considerably, always informed by the complex interplay between formal and informal governance systems at the local level. Local authorities frequently develop or adapt their own systems and rules, combining customary (and occasionally religious) practices with new legal frameworks. Initiatives meant to regulate governance through central state (or regional) institutions, however, often face challenges due to the unstable power relationships characteristic of volatile conflict-affected contexts, their limited capacity to agree on governing practices and enforce regulations, and the relative strength of groups with entrenched power and significant patronage networks.

4.3. The level of integration of civilian authorities and armed actors

In considering government by armed groups, Mampilly and Stewart (2021: 19) distinguish two dimensions—the degree of ‘power sharing’ between the armed group and civilians and the level of integration between them. These two dimensions are considered together here, as, for the purposes at hand, they are two sides of the same coin, and this paper is interested in the level of integration between civilian and armed authorities on certain issues of governance, however governance is secured (see also Kasfir 2015: 35–36). Arrangements potentially range from cases where there is no integration and the armed authorities are entirely in control, staffing institutions with their own people and managing all civilian affairs directly (in Mampilly and Stewart’s terminology, no power sharing at all), through to situations in which civilians play a real part in governing or are entirely in control. Full control by either armed or civilian authorities would mean the lowest level of integration between the two including no power sharing, while collaboration between them to manage institutions and other government services would point to a higher degree of integration.

As in Mampilly and Stewart’s study, the case studies in this paper show a wide range of arrangements among the different countries and within them. In Myanmar’s military-controlled areas for the most part there is no integration between civilian and armed authorities, although, as noted above, there are situations in which the military needs the support of both non-state armed groups and non-military actors. The military sometimes appoints civilians to administrative positions, and formally all officials are accountable to the military. In practice, however, the situation can be more nuanced in that the military must accommodate the population; hence, some of the appointed officials have more leeway to act than others. In areas controlled by (armed) revolutionary forces, civilians are sometimes used to run the administration, but responsibility for the most part lies with the leadership structures of the relevant armed group. It is only in some areas populated by smaller (sub)ethnic groups that civilian authorities seem to have control over armed groups, and decision making happens in a more collective manner. In contrast, in areas under the control of NUG-affiliated People’s Administrations, there is at least a nominal separation of duties between the PaLaPhas and PaKaPhas, even though administrators appear to carry weapons, and there are no known rules or regulations that would effectively separate the duties and responsibilities of People’s Administration Bodies (PaAhPhas), PaLaPhas and PaKaPhas.

As described above, in Somalia’s clan-based society, the lack of strong state structures results in informal arrangements based primarily on clan agreements. Additionally, the influence and power base of clans and subclans can differ significantly, creating different local government dynamics (Wahiu and Abebe 2025). Even in Al-Shabaab-controlled territory, which is marked by centralization and coercion of local populations, the group tends to maintain a level of connection with clan elders and other customary institutions in particular regarding ‘first instance’ dispute resolution, as long as they do not violate Al-Shabaab’s strict interpretation of Sharia law. Al-Shabaab also tends to appoint elders in replacement of those that resist their rule, though the former struggle to earn the trust of the community (ICG 2022).

Arrangements in Yemen are equally diverse. The area in which Ansar Allah is the de facto authority may be considered to be run as a theocracy directing both civil and military institutions. Civilians, not soldiers, staff the administration but they are tightly controlled by the movement. The arrangement could not be described as power sharing or integration—power is not shared with local leaders, but some local leaders can be incorporated in decision making if they adhere to Ansar Allah ideology and, ideally, are also members of the sada social category.¹⁸

In the areas formally under the control of the IRG, arrangements are more fluid. First, at least in the cases surveyed in the Yemen case study (Lackner 2025), the relationship between the military elements and militias, on the one hand, and the local administrative authorities, on the other, is complex and not formally regulated. Civil administrations attempt to provide services and carry out their responsibilities, sometimes with the backing of the local military factions, sometimes without any connection. The military factions are both complementary, such as the different southern separatist ones, and in competition with each other, such as the ‘National Shield Forces’ led by the president competing with the southern separatist forces, particularly in Aden.

Civil servants operate on what might be described as an ad hoc basis in the context of salaries that are only occasionally paid and insufficient to finance a basic (acceptable) living standard and receive administrative guidance likely only from the governor’s offices. They often have to be paid by individual citizens for performing services, a situation which was already prevalent in the pre-war period. In Mudhaffar, a district in Taiz, power sharing is implemented in practice: the director of the council ‘implements a miniaturized version of the political quota system used at the governorate level but appoints civil servants from two more parties, including the Nasserist and Socialist parties, at his discretion’ (Al-iriani 2024: 13).

In summary, in considering the relationship between civilians and armed groups, again there is great variety both across the countries that were studied and within them. Notably though, in all three countries, there are instances where local authorities are fully subject to specific state and non-state (armed) groups and instances where there is some level of integration between civilian and armed authorities. The broader conflict dynamics and ability and willingness of groups to use force seem determinative in this regard; they are also a reminder that all these relationships may be very unstable, changing rapidly as both broader and local conflict dynamics change.

4.4. The degree of continuity and change

This dimension is concerned with whether the institutions that are present have a long history or were newly created, fully or partially displacing what may have existed in the past. It also seeks to understand how and why governing arrangements have (or have not) changed. For instance, are changes meant to fill vacuums, to transform pre-existing political institutions and practices deliberately, or to respond to rapidly changing circumstances on the ground? Mampilly and Stewart’s study (2021: 20) concludes that when a rebel group captures new territory, it may choose to maintain the status quo more or less completely or it may transform or replace existing institutions and practices, perhaps driven by its ideological precepts. The case studies suggest, however, that the degree of change is largely determined by the level of control new authorities have over an area, the mechanisms they believe will maintain their control and, perhaps, their ‘long-term political objectives’ (Mampilly and Stewart 2021: 27). However, in general, both old and newer forms of local governance are in constant flux and adapt to new and changing circumstances, and are also influenced by external factors (see Boege et al. 2008: 7).

In areas controlled by the Myanmar military, it appears that there has been rather little formal change. While the military junta formally at least maintains all extant structures, institutions, laws and other regulations, as noted above, where officials resigned from their posts it has had to appoint new (military or military-close) officials with little experience in managing government structures or institutions. As also pointed out above, officials and institutions are subject to central control.¹⁹ In contrast, local forms of governance in areas held by revolutionary forces have for the most part been recently established, although sometimes they appear to build on existing customary practices and many, perhaps most, use customary structures to manage some aspects of local governance such as dispute resolution. In some cases, presumably drawing on their inherent flexibility and adaptability, customary forms of local governance have been adapted and their purview expanded to respond to the changing context (as well as to the lack of reach and enforcement of formal institutions and rules and regulations). Beyond this, revolutionary forces, in particular ethnic armed groups that had in the past established systems of governance for territories mostly inhabited by their own people, have also had to adapt these systems, particularly when they have occupied territories inhabited by other ethnic groups.

In Somalia, while customary practices have been maintained and adapted over the years to the rapidly changing context, new legislation has emerged at both the central and FMS levels to regulate local governance. This legislation has often been at least partly disregarded, however. So, while institutions established by such legislation may exist, clans and subclans have seen the significance of using existing customary channels to increase their representation in those new institutions, thereby enhancing their political power and patronage networks. Thus, clan negotiations and resulting agreements are key in defining the composition of local councils, as they are for parliamentary bodies at both the central and the FMS levels. These agreements are frequently renegotiated to accommodate new requests for inclusion of particular subclans. While these renegotiations reflect high levels of flexibility and adaptability in governance arrangements, they can also destabilize local government frameworks, due to the frequently changing compositions and power balance. This instability is possibly inevitable in unstable political contexts in which politics are underpinned by the use of force. In Al-Shabaab-controlled territory, by contrast, there is a level of centralization and coercion of local populations, with little flexibility or opportunity for local innovations.

The Yemen case studies also suggest that, at least formally if not usually in practice, structures and rules have been maintained unchanged, because both authorities claim to be legitimate and therefore the upholders of constitutional rule. Regional and local administrations established before the conflict continue to exist in some form or another but are bent to respond to practical realities. In Ansar Allah controlled territory, there is significant centralization, with little opportunity for even the local initiatives permitted by the law. In IRG areas, regional governors who were always powerful now have yet more authority, going largely unchallenged, unless there are competing factions in the area (Lackner 2025). As noted above, the arrangement introduced in 2008 for governors to be elected has fallen away, however, and they are again appointed by the president of the IRG. Elected councils seem to have stopped functioning across the country, although in Ansar Allah territory some council committees appear to function (Lackner 2025). Customary arrangements that managed various affairs in the past presumably continue in place, subject to whatever political and military forces exercise power in the area.

4.5. The level of public inclusion and accountability

This dimension is concerned with the degree to which local government structures involve different social groups in decision making (Mampilly and Stewart 2021: 20–21; Kasfir 2015: 34–35). It depends largely on the degree of embeddedness of local government structures in the communities they govern. The more embedded local government structures are, the more likely there is to be a higher level of public inclusion and accountability. And, it can be argued, the more the public is included in decision making, the higher the level of accountability of decision makers or local authorities, as the more engaged the public will be (see also Huang 2016).

In the case of Myanmar, the military junta does not seem to be inclined to involve the public in decision making or to be in any way accountable to the public, while in other areas controlled by revolutionary forces there is a range of situations, from fairly autocratic, authoritarian administrations to governance that is striving to be inclusive and accountable. In the latter cases, the motivation for greater inclusion and accountability may be to build democratic structures, as appears to be the case in Karenni State (Karenni Civil Society Network et al. 2024), or, more opportunistically, to increase support among the population and their acceptance as legitimate administrators. Sometimes inclusion and accountability mechanisms do not extend beyond sharing information in public meetings, but in some instances administrators will ask people to communicate their needs and demands. Nonetheless, in these conflict-affected areas, lack of trust and general fear may prevent civilians from requesting goods and services or expressing any criticism. As mentioned above, in some areas populated by tribes and other smaller ethnic groups, decisions may be taken collectively at council meetings held periodically. But even in these cases, it is unclear to what extent the wider public is actually involved in making or shaping decision making.

When considering the inclusion of the public, again Yemen mirrors Myanmar to some extent. As Lackner emphasizes, ‘community level involvement in politics and administration has been a major form of participation in Yemen for many decades’ (Lackner 2025). Predictably, however, in the territory now under Ansar Allah control, the public is not involved in decision making and civil society is strictly controlled. Nonetheless, even in those areas, there are instances where ‘the community has been able to influence decisions made by the local authorities, such as the removal of some governors and security directors in Taizz over the past nine years’ (Al-iriani 2024: 28). This is not public inclusion, but a sign of some level of attention to local desires. By contrast, the case studies in areas formally under the authority of the IRG record that in those areas ‘community committees and other mechanisms have been developed to encourage civil society involvement in local governance’ (Lackner 2025). Sometimes community committees have been established by the (regional) governor, which presumably facilitates interactions with authorities. Overall, as the Yemen case studies show, ‘the extent of … [the] influence and efficiency [of authorities] varies widely: they are important actors in some areas’ and less so in others (Lackner 2025: 38). Nonetheless, as Lackner concludes, ‘overall, community committees have the long-term potential of becoming the foundation stones for a more democratic and equitable post-crisis Yemen, an important contribution to the future’ 2025: 39).

It is not possible to assess the depth of public participation such mechanisms promote in Yemen—the likelihood is that the level of involvement varies from place to place and issue to issue, and over time. Lackner also notes that in Taiz, for example, while there are formal mechanisms for feedback and meetings, civil servants are also alert to the need to ‘balance the interests of the political party which has given them a seat, and other stakeholders higher up in the chain’ (Al-iriani 2024: 13).

In Somalia, the clan-based model of selecting representatives to local councils and FMS institutions is based on patronage and relies on a narrow base. Wealthy clan representatives, or those with wealthy sponsors, are usually the ones selected for local, FMS or central state positions. Only in those cases where local administrators are supported or elected from among local communities, are there examples of the administrator exhibiting a level of accountability as they attempt to respond to their communities’ demands (Davies et al. 2023). But, as International Crisis Group reports, generally representatives are not accountable to the public, and there are few mechanisms for the public to sanction representatives who do not deliver. This is increasing the level of distrust of the political, or clan, system (ICG 2024a). Beyond this, and as may be expected, Al-Shabaab’s rule is not inclusive of society at large. While Al-Shabaab may seek connections with clan elders to enable its rule of local communities, its own accountability towards local communities is even more limited, and the allocation of revenues raised through tax collection is shrouded in secrecy (ICG 2022).

4.6. The level of effectiveness or responsiveness

The level of effectiveness of extant governance structures may be the most important factor to consider when rebuilding a country or making constitutional decisions. Effectiveness can be measured in many different ways. For instance, Mampilly and Stewart (2021: 38) focus on the goals of rebels and refer to the level of effectiveness as the ‘degree to which political institutions fulfil the role they are assigned’. They note that ‘[r]ebels have divergent reasons for why they create their political institutions with different agendas driving the institutional selection process. As a result, even political institutions that do little while existing primarily on paper can be effective if the intended goal is primarily performative for external observers’ (Mampilly and Stewart 2021: 38). The interest of the authors is broader, as it is concerned with what peacemakers and constitution makers need to consider in decisions about what parts of substate systems deserve protection and strengthening in the longer term, and where weaknesses lie.

Effectiveness concerns the degree to which governance is responsive to the needs and interests of the population. In conflict-affected areas, security is usually a primary concern, but health and education, infrastructure, humanitarian aid and dispute resolution are all relevant. It is often argued that this can be measured by investigating whether people desert the area—if governance is weak and, particularly, insecurity high, populations are likely to move (voting with their feet).²⁰ But, instead of using such extreme measures, which may also be difficult to ascertain, one may assess the degree to which people’s safety is secured, taxes are raised, services delivered and authorities respected.

Our case studies could not conduct the granular cross-community studies required for a full assessment of effectiveness. Nonetheless, in Myanmar, public services appear to have significantly suffered in the aftermath of the coup, in particular in the health and education sectors. While these services have to a great extent been taken over by private service providers and charities, in military-controlled areas service providers have been increasingly limited, and bribery has become pervasive. Beyond this, while the junta has access to revenues (and taxes), allocation is not transparent and service delivery is mostly focused on infrastructure development or repair, in response to the military’s needs for transportation, for instance, but also to sustain different patronage schemes, with public funding going to military-controlled or military-friendly businesses. In general, services do not appear to be intended to respond to the population’s needs and demands, and citizens do not appear to have any role in deciding how resources are allocated or what services are provided.

In areas outside the control of the junta, revolutionary forces attempt to provide services in a variety of ways, including digitally in fields such as education and healthcare.²¹ The quality of service delivery in areas under the control of revolutionary forces varies depending on a number of factors, including access to skills and general capacity, but also: (a) access to revenues and goods; (b) the extent of the territory they control—control of larger areas can reduce the cost of procuring goods and services; (c) collaboration between public and private (national and international) providers of goods and services; and (d) fluidity of conflict dynamics. According to interviews held in Chin State (February 2025) and in Sagaing (May 2024), services are generally very limited, although their provision appears to be steadily increasing, in particular in the health and education sectors. For other services citizens often need to make use of brokers or political fixers who are able to provide services for a fee, thanks to their connections with military authorities in areas controlled by the junta. In Chin areas, private initiatives partly funded by diaspora remittances or non-governmental groups provide some services in the education and healthcare sectors. In Sagaing, services are mostly funded through local taxes, public donations, NUG grants or international development cooperation. However, the relationship between local revenue raising and revenue allocation appears not to be made public, and there appear to be no mechanisms of accountability for decision makers.

There is little information on Somalia. It appears likely that service delivery at the local level is mostly provided by non-state actors, and local governments have little capacity to provide them or to oversee or regulate non-state actors. This in turn may limit the relevance and legitimacy of the state in the eyes of local communities: ‘Weak regulation also means that there is significant inequality in the access to and quality of services. The effectiveness of local government is also hindered by significant mandate contestation with the FMS and the federal levels of government, particularly in areas where state control is limited to urban areas’ (Davies et al. 2023: 11).

In Yemen, the provision of services is the main bone of contention between local populations and their administrations. Only half the medical facilities in the country are functioning and large numbers of schools have been either destroyed or turned into temporary shelter for internally displaced people. Both the health and education sectors have been kept more or less afloat thanks to international financing, including from UN agencies and the World Bank but also from bilateral funders, which have contributed to salaries, equipment and consumables, as well as the reconstruction and repair of infrastructure. Both the IRG and Ansar Allah claim to be providing services but also to be constrained by financial difficulties, despite finding the funds for their respective military sectors. There is no doubt that access to services, either free or at low cost, is the first criterion for citizens’ assessment of their ruling entities, and on this measure both Ansar Allah’s and the IRG’s reputations are low. Recent dramatic cuts in international assistance will certainly worsen the situation in all of Yemen, and may have even more impact in Ansar Allah areas, as aid there has already been significantly reduced since early 2024.

Generally, as one might expect, in conflict-affected areas in fragmented states, local-level government struggles to provide basic services. The problem is compounded by the inaccessibility of many of the areas, which means that international aid and services are difficult to deliver. What is not known is how this affects the attitude of the people to local governance and what might be expected in the future.

There is a body of research on the impact of wartime practices on future political settlements and democratization (Huang 2016; Dudouet and Lundström 2016).²² Little work has been done, however, on the way in which institutions at a substate level might impact on the longer-term stability of the state and why they should, therefore, be considered in constitution making. This discussion paper is an initial attempt to understand this, based on studies in countries where conflict is ongoing, rather than on concluded conflicts. Further research is needed to complement the knowledge of how the durability and effectiveness of constitutions might be strengthened by paying greater attention to local governance practices that operated during conflict. This paper takes a first step by offering a way of understanding local dynamics and institutions so that those drafting a new constitution can develop a more nuanced understanding of the contributions and challenges local arrangements may present.

As noted in Chapter 1, the six dimensions that are used draw particular attention to the relationship between the population of an area and those that wield power at any level of government. In particular, they direct attention to issues of legitimacy and the rule of law. Understanding the source or basis of authority at the substate level (for example, brute force or some level of agreed arrangement and its democratic quality) (dimension 1); the degree to which governance practices are institutionalized and thus predictable and consistent with a system committed to building the rule of law or not (dimension 2); and the degree to which authorities engage with the population and are accountable to it (dimension 5) provides insight into the degree to which existing practices may evolve to become more representative, transparent and accountable. The tenacity of established institutions, both formal and informal (dimension 4), and their effectiveness and responsiveness (dimension 6), can signal how cautious any process of change and democratization should be. Finally, a proper understanding of the role of armed forces at the substate level and the relationship of civilian authorities and armed groups in local administrations (dimension 3) should provide peace- and constitution makers with insights into the challenges of implementing new arrangements that will work on the ground and should prompt them to strategize ways in which armed groups, also at the local level, would be willing to accept civilian oversight.

Although peace- and constitution-making processes are always constrained, involving many international interests along with the interests of parties to the conflict, understanding the diversity of regional and local arrangements in the territory concerned and responding appropriately could strengthen outcomes, potentially helping to avoid further fragmentation and contributing to better government across the country. At the same time, these considerations may help in nuancing the design of the negotiation process and better assessing the extent to which different parties may be motivated to engage constructively or may need to be (at least temporarily) excluded.

The single most obvious conclusion to draw from the analysis of the case studies considered in this paper along these six dimensions is that there is huge variety. Unsurprisingly, governance in each locality is largely determined by the nature of the conflict and conflict parties, the number and nature of contesting groups in a territory, access to resources, and past governance arrangements. Thus, governance varies significantly across the cases, with some groups interested in building governance institutions, others more interested in co-opting local communities and some intent on undermining state-building processes.

Overall, it can be seen that governance at the local level can range from arrangements operating with minimal rules, practices and service delivery, to situations in which more or less incipient forms of democratic governance may offer ways for communities to engage and have their needs and demands met, such as perhaps in Karenni State (Myanmar) or in Puntland (Somalia) (Arjona, Kasfir and Mampilly 2015: 1). It is also clear that old institutions tend to endure, adapting to changing circumstances and needs, and, in particular, that customary institutions often operate in a complementary way to other, more formal practices.

What might constitution makers make of this? Three interrelated points seem relevant. First, it should not be assumed that localities that have experienced significant conflict and forms of coercion do not have strong democratic instincts. In The Wartime Origins of Democratization, Huang (2016) argues that the more communities participate in their own governance, at the local or regional level, the more responsive, transparent and accountable governments are likely to be. The other side of the coin is that ‘when rebel groups demand goods, labor, and loyalty of people in their milieu, people will come to expect returns in the form of greater political rights’ and that ‘by extracting resources from them, rebel groups will become beholden to popular demands and expectations … In the case of domestic popular support, furthermore, this loss of autonomy will be felt both in the short term—people will want protection, order, and provisions in return for their efforts—and in the longer term, when they will expect the post-war regime to deliver more goods and rights to them than did the previous regime’ (Huang 2016: 182). This suggests that citizens at substate levels of government may be able to contribute to the design of accountable and responsive government in their areas and that a national constitution should not foreclose this type of democratic innovation. It also suggests that different localities and regions could learn from one another and opportunities to share and exchange experiences should be considered. In other words, supporting those communities at the local or regional level that have managed to create governance structures that are accountable to the citizens may be a good starting point, and one that other actors in different parts of the country can emulate.

Second, the existence of authorities at substate levels of government can potentially both strengthen and challenge attempts to (re)build the state. Recently there has been a push to ‘work with the grain’, to strengthen existing and emerging local and regional administrations while ensuring ownership by involving local communities in governance (see Meagher, De Herdt and Titeca 2014: 1; ICG 2024b: 3). The argument is that strengthened local and regional administrations will contribute to building the state even if the broader state-building process is top down. Equally relevant in this sense is that disregarding existing local institutions (and communities) is unlikely to be effective but will rather antagonize both regional and local elites and communities. This does not necessarily mean that every form of local authority and governance should be accepted and legitimized. But it does suggest that, in designing new institutions, constitution makers need to pay attention to existing institutions and relationships.

Simultaneously, significant levels of distrust of the (emerging) centre can reduce the motivation of local communities, and established and generally accepted local governments, to participate in the development of central state institutions (see Boege et al. 2008: 16), particularly if new central state institutions are considered bound to recentralize power. Drawing substate-level stakeholders into discussions aimed at (re)framing state institutions may help to build trust and contribute to reducing the parochialism of substate entities.

Third, one size does not fit all. The great variety in local (and regional) institutions will have legacies and it will seldom be possible for those arrangements to be replaced seamlessly by some national model. The challenges of regulating and bringing coherence to substate-level governance will be yet more difficult when central state institutions are weak and lack control of events on the ground. Indeed, diversity at the substate level may be necessary to meet different communities’ needs and demands, at least initially and possibly for a long time. Key questions are then the extent to which substate governance should be constitutionalized, and with what degree of flexibility—what can be agreed for the entire country and what should be left for substate communities and authorities to determine—and perhaps how to incorporate regular review processes. It may be that minimal governance standards should be agreed, but, beyond that, overall diversity in governance should be supported.

When a fragmented country is attempting to transition from conflict to peace and to stable democracy that can operate in a unified way, supporting the development of strong substate institutions that will contribute to state-building and strengthening democracy is important. Difficult questions inevitably arise concerning which institutions should be retained and strengthened and which dismantled. Often the central issue will be how to build democratic substate institutions from extant institutions that have operated with varying degrees of accountability, and where practices have been designed to manage the conditions of conflict. As this paper demonstrates, needs across each country will be different, so constitution makers need to pay attention to regional and local concerns and devise arrangements that promote effective and accountable government in a way that is sensitive to those differing needs, allowing for those arrangements to evolve and strengthen over time.