The dynamics of donor and domestic elite interaction in Mozambique: formal decentralisation and informal power structures

Abstract This paper analyses the interaction of domestic political elites and external donors against the backdrop of Mozambique’s decentralisation process. The empirical research at national and local levels supports the hypothesis that informal power structures influence the dynamics of this interaction. Consequently, this contributes to an outcome of externally induced democratisation different to what was intended by external actors. The decentralisation process has been utilised by ruling domestic elites for political purposes. Donors have rather focused on the technical side and ignored this informal dimension. By analysing the diverging objectives and perceptions of external and internal actors, as well as the instrumentalisation of formal democratic structures, it becomes clear, that the ‘informal has to be seen as normal’. At a theoretical level, the analysis contributes to elite-oriented approaches of post-conflict democratisation by adding ‘the informal’ as an additional factor for the dynamics of external-internal interaction. At a policy level, external actors need to take more into account informal power structures and their ambivalence for state-building and democratisation.


Introduction
In the context of the third wave of democratisation, research on democratisation has increasingly focused on the factors that can promote or impede formal democratisation processes and state-building in post-conflict societies. While the state-building literature has emphasised the influence of external actors on the construction of democratic institutions, transition theories have largely neglected external actors, concentrating instead on the role of domestic elites. 1 Recent elite oriented approaches to the study of democratic transition have identified the role of domestic elites and the interaction of domestic elites and external actors as important factors in post-conflict democratisation. 2 However, insights into the nature and dynamics of this interaction remain scarce. 3 In light of the intensifying discussion of 'defective democracies' 4 the debate on neopatrimonialism has gained new prominence over the past two decades, in parallel to the debate on transition. 5 Within the neo-patrimonialism debate, several authors have identified informal power structures used by domestic elites as a decisive factor shaping democratisation processes. These scholars explicitly take into account the interrelation of formal and informal power structures. 6 However, systematic research on this issue in the context of post-conflict democratisation needs further attention.
More specifically, elites-oriented studies have thus far largely neglected the role of informal power structures in the interaction between domestic elites and external actors and the consequences thereof for externally induced democratisation processes in post-conflict settings. In line with the objectives of this special issue, the main argument of the present article, therefore, is: the existence and instrumentalisation of informal power structures influences the dynamics of the interaction of external actors and domestic elites in post-conflict democratisation processes beyond formal negotiations, capacity-building and institutions. This sheds light on the phenomenon, why externally induced democratisation processes may work out differently than originally intended by external actors.
While proponents of neo-patrimonialism research have primarily focused on African countries, scholars of transition theories have placed emphasis on the countries of the former Eastern Bloc. 7 Mozambique, a Southern African country and former socialist one-party state with a history of conflict and transition to peace and democratisation, represents a highly internationalised setting that has been profoundly influenced by external donors. It may, therefore, serve as a paradigmatic case for the analysis of the interaction between external actors and domestic elites in the context of post-conflict democratisation.
Although Mozambique was once considered a 'donor darling' by the international community due to its supposedly successful transition to peace 8 and democratisation, 9 the country's decentralisation process as part of democratisation seems to have diverged from the original objectives of the donors. New clashes between the main opposition party's militia (former rebel movement) and the government's army since 2013, amongst others because of issues of the devolution of power and resources from central government to the provinces, hint at shortcomings of post-conflict democratisation in the realm of decentralisation.
Previous research has examined the development of the relationship between the Mozambican Government and donors on the national level; 10 democratisation, local government and conflict transformation; 11 the development of the decentralisation process and political settlement; 12 the role of elections and inter-and intra-party struggle; 13 local governance in the districts; 14 and political decentralisation and municipal development in Mozambique. 15 However, the role of informal power structures for the dynamics of the interaction between external actors and domestic elites at national and local levels and its consequences for post-conflict democratisation have not yet been explicitly addressed empirically. 16 Whereas elites and neo-patrimonialism-oriented studies focus mainly on the national level, this article seeks to contribute additional insights by showing how phenomena of informal power structure usage at the national level are reproduced at the local level. Furthermore, it takes into account the mutual perceptions of the involved actors on the external-internal interaction.
The main research questions, therefore, read as follows: which objectives and interests do external donors and domestic elites in Mozambique pursue regarding democratisation and decentralisation? How do the actors perceive the respective other's objectives and acting? How does this influence the dynamics of their interaction and consequently the nature of the democratisation process?
The empirical evidence is based on a focused ethnographic methodology of qualitative data collection. 17 This includes more than 100 interviews with national and international experts as well as political, civil society, business and donor representatives at all levels of government that were conducted during field studies in 2009 and 2010. 18 The results of the interviews and participant observations at the district level -specifically, in the provinces of Manica and Inhambane -were triangulated with secondary literature, focus group discussions as well as expert debriefings to ensure criteria of reliability and validity were met. 19 Departing from further theoretical insights and an overview on the context of decentralisation and donor involvement in Mozambique, the empirical analysis (1) shows which common and diverging objectives and interests domestic elites and external donors pursue within their interaction regarding the decentralisation process and how those are perceived; (2) addresses how domestic elites deal with the results of donor-elite interaction in terms of the decentralisation process and how this consequently (a) shapes the structures as well as (b) influences their acting within those structures, both at national and local government level. The conclusion then points out the ambivalence of informal power structures for the dynamics of external-internal interaction as contribution to further theories and policies of post-conflict democratisation.

Theoretical insights
The promotion of democracy and good governance by donors since the 1990s has been accompanied by expectations of improvements in security, economic development, the rule of law, the safeguarding of human rights and citizen participation. 20 This includes the expectation that democratisation, with the promotion of democratic institutions such as elections as its central condition, represents an indispensable pillar of post-conflict state-building. 21 The promotion of decentralisation is regarded as an important component of democratisation and stabilisation processes in post-conflict contexts. 22 However, externally induced democratisation in post-conflict settings may lead to a variety of perceptions and subsequent reactions by the domestic elites involved. 23 Elite oriented approaches to the study of democratic transition have pointed out that external actors and domestic elites may follow different objectives. 24 In practice, the implementation of democratic institutions might fulfil expectations and satisfy interests of domestic elites other than those intended by the external actors. 25 This has influence on the shape of formal democratic structures. As Zürcher et al. for example show, domestic elites welcome the resources stemming from external actors but are less willing to adopt democratic norms and institutions in case they view their interests threatened. 26 Studies on neo-patrimonialism with its strand of political clientelism contribute to this debate by highlighting the role of informal power structures: ruling political elites using formal democratic state structures for the interest to safeguard their privileges and to maintain power. 27 The informal dimension, however, also applies to external actors. Studies show that they may promote democratic institutions by acting as an additional power outside of domestic formal democratic decision-making structures and thus suffer legitimation deficits. 28 Conflicting interests and objectives apply to informal influence on the shape of formal democratic structures as well as to the informal action of domestic elites within formal democratic structures. Studies on post-conflict democratisation have pointed out to the aspect of the willingness of domestic elites to play according to democratic rules. 29 In the context of democratisation processes, scholars of neo-patrimonialism have observed that the state has been captured by elites who are primarily concerned with their own maintenance of power through clientelistic networks of patronage. 30 Therefore, research on neo-patrimonialism again contributes to the debate on the interaction of external actors and domestic elites by highlighting the role of informal power structures, namely the clientelistic relationships of exchange and reciprocity, based on resources of power within formal state institutions. 31 Additionally, Migdal and Schlichte assert that many international and transnational actors have propped up the state as all-important mediator between global actors and the domestic population, enhancing domestic political elites' power over society through the control of key distributional mechanisms such as foreign aid, loans and investment. 32 This applies all the more in the context of post-conflict democratisation, with its specific challenges, such as peace-building, stabilisation, reconciliation and often imposed institutional structures, accompanied by heavy external resource input, compared to development contexts. 33 One point of intersection among proponents of elites, transition and neo-patrimonialism research alike is the role of agency (in terms of the values, interests and objectives of the involved actors) in negotiations over norms and institutions. 34 Additionally, some authors link the question of power to agency as decisive for structuring the interactions of groups of people. 35 Taking the elite concept proposed by Higley and Burton as a starting point, elites are defined as […] persons who are able, by virtue of their authoritative positions in powerful organizations and movements of whatever kind, to affect national political outcomes regularly and substantially. […] [N]ational elites can be defined as top position-holders in the largest or most resource-rich political, governmental, economic, military, professional, communications, and cultural organizations and movements in a society. 36 The focus of this paper is on the interaction of donors as a specific group of external actors and 'political elites' in Mozambique. Following Macuane, the latter is referring to those elites -representatives of political parties, the administration, the private sector, or civil society -who are somehow linked to the ruling party in Mozambique and have an influence on political decision-making, taking into account the blurred boundaries between different groups of domestic elites and various interchanges (e.g. between politicians and the administration). 37 The empirical analysis focuses on political elites linked to the ruling party and donors as a specific group of external actors. Their interactions in Mozambique at national and local levels is analysed against the backdrop of decentralisation within the broader frame of democratisation.
Decentralisation generally implies the transfer of legal and political authority over planning, decision-making and administrative management from the central government to local administrative units, combined with the necessary transfer of resources. 38 It is commonly differentiated according to devolution (i.e. political decentralisation) and deconcentration (i.e. administrative decentralisation). 39 'Informal' in this article applies to structures and acting that are not formally regulated (e.g. through contracts, rules, institutions) and that are available within and outside formal democratic structures. Relevant for the characterisation as informal is the respective perception of structures and acting as informal by the involved actors.

Donors and decentralisation in Mozambique
The post-conflict order in Mozambique after 10 years of struggle to liberate the country from Portuguese colonial rule -independence in 1974 -and 16 years of conflict between the Liberation Front of Mozambique (Frelimo) and the Mozambican National Resistance (Renamo) 40 -General Peace Accord of Rome (GPA) in 1992 and first multi-party elections in 1994 -has been decisively influenced by external actors.
According to the Human Development Index, Mozambique continues to number among the four poorest countries in the world. 41 There has been an inextricable link between donor support for the Mozambican peace process and aid conditionality. 42 The country's relationship with donors has been characterised by its extreme dependency on external aid, amounting to about 50 per cent of the national budget. 43 Over the history of donor involvement in Mozambique after 1992, policies and attitudes have shifted several times. Funding evolved from a project-based, bilateral process to budget support and increasing donor co-ordination (G19) in accordance with the Paris Declaration in the Programme Aid Partnership (PAP). This policy ties budget funding to government performance in accordance with the Poverty Reduction Strategy Paper (PRSP). 44 The entire decentralisation process since 1994 has been dominated by the concept of 'gradualism' , meaning the gradual devolution of power and resources to the local level. The ruling elites support a dual process of administrative decentralisation -the deconcentration of resources to nationally ruled districts -and democratic decentralisation -the devolution of political power and resources to municipalities with elected local governments (starting with 33 municipalities in 1997, extending to 43 in 2008 and to 53 in 2013). Thus far, no official decentralisation strategy or law has been enacted. 45 On average, less than five per cent of Mozambique's national expenditures are currently transferred from the national to the local government level (districts and municipalities). 46 The legal frame of decentralisation consists of Law 3/94, the amendment to the Constitution of 1996, the Local Authority Package (Pacote Autárquico), 47 and the Law of Local Organs of the State (LOLE), Law 08/2003, including its regulation and the guidelines on the organisation and functioning of the consultative councils (CCs) at the district level. 48 Additionally, since 2006, the national government has promoted a district development fund (FDD), known as the 'seven million' . 49 With the advent of the CCs and the FDD as part of the deconcentration process, new formal decentralisation structures have been established at local level, offering representatives of district populations the opportunity to formally take part in local political decision-making processes. 50 However, as experience in the field shows, a variety of phenomena suggest that these structures function differently than intended by donors in terms of promoting democratisation. 51

Diverging objectives
The main objective of external actors such as the International Monetary Fund, the World Bank and bilateral donors 52 has been to build new institutions and support democratic state reform in post-conflict Mozambique. 53 Within this broader frame, donors and domestic elites together have promoted formal decentralisation structures with the objective to address questions of stabilisation, poverty reduction and participation at different levels of government. 54 Originally, donors mainly supported democracy and governance at the central level -electoral processes, political parties and civil society -but the focus has shifted to local levels in more recent years. Decentralisation has been on the donor agenda since 1994. However, together with good governance, it has begun to attract additional attention from donors, including the G19 group of budget support donors. 55 This has been a consequence of the perceived insufficiency of the democratic quality of the parliament and opposition parties, as well as the ruling party. 56 National and international experts point out that the ruling political elite's main objectives involve defending the historically predominant role of Frelimo and the unity of the party, maintaining control of the economy and promoting strategic national and international alliances to achieve these objectives. 57 At the same time, there are external experts that believe that the Mozambican political elites want the population to remain poor in order to justify continued donor funding, especially in light of expected natural resource rents and diminished donor contributions to the government budget. 58 While formally embracing the decentralisation process, there is widespread scepticism at all levels and from the perspectives of various domestic elites (politicians and the administration) regarding external influences on national policies. As one national consultant stated: 'The big changes in the country came because of donor pressure (decentralisation, anti-corruption, good governance). But those are not our agendas. The government accepts it because it needs the money' . 59 The donor presence has also been criticised for being overly paternalistic and conditional: 'We are no longer donkeys; there is no need for programmes for 'savages'' . 60 An international expert in the Ministry of State Administration (Ministério de Administração Estatal -MAE) explained: 'Mozambique is technically broken; it is a conditioned democracy' . 61 Directors of Provincial Directorates voiced similar perceptions: ' After all, it was the Europeans that told us, 'You have to be democrats' , that's why we changed the constitution' 62 and 'They [donors] do what they want, not what we want' . 63 Donors are accused by domestic elites of being primarily interested in their own visibility and the justification of their practices. 64 There is also the perception on the part of domestic elites and external experts that donors would turn a blind eye to democratic deficits in order not to challenge the apparent 'success story' of their intervention in Mozambique. 65 The interaction of donors and domestic political elites is characterised by factual and perceived diverging objectives. Domestic elites accuse donors of following their own interests of legitimating their continued role in post-conflict democratisation. Donors focus on technical issues of formal decentralisation of policies and structures ignore the hidden and informal objective of power maintenance by domestic elites by using decentralisation structures amongst others.

Shaping structures
At the national level, several authors have identified phenomena of (political) clientelism, limited access to decision-making and rent-seeking as part of the political elite's strategy of maintaining power. Specifically, they point to the 'partidarisation' of the state. 66 Interview respondents from many different backgrounds and perspectives describe the political system as characterised by a dominant party and blurred boundaries between the state, government, economy and ruling party. 67 Statements by national consultants such as 'Government is the state' 68 and 'The country returned to the one-party state' are typical. 69 An international expert claims, 'The state in Mozambique is detached, dominated by the political elites, instrumentalised and absorbed' . 70 This observation is supported by independent external researchers, which assert that Mozambique could be labelled a de facto (not de jure) one-party state. According to them, this phenomenon can be traced back to two sets of informal power structures in Mozambique that have shaped state-party-society-economy relations at different levels: (1) a political culture influenced by the legacy of the one-party state and (2) informal arrangements of Frelimo with the major opposition party Renamo in the postwar setting. 71 Thus the blurred lines between various Mozambican political elites include arrangements between the leading figures of Frelimo and Renamo. 72 There is growing alienation between the government and society, but the weak opposition allows the ruling party to dominate all spheres of society. 73 There are some independent forces such as certain national intellectuals and the media; in general, however, the opposition, civil society, the private sector and traditional authorities have not yet effectively challenged the government or acted as significant change agents. 74 The ruling political elites have managed to maintain the balance of power in their favour through a process of constant negotiation and circumstantial coalition-building with state and non-state actors. 75 Weimer et al. discuss this phenomenon of 'political settlement' by the ruling political elites in post-conflict Mozambique. 76 Such settlements have been possible despite the influence of and dependency on external donors and the radical changes the country has undergone 77 due to various means such as instrumentalisation and co-optation of formal democratic structures to ensure compliance. 78 Members of the political elites control access to and use of government resources including donor funds assigned for the building of democratic institutions, which they employ to promote the ruling party's interests of maintaining power as well as personal ones -following partisan, ethnic, regional or kinship criteria. 79 The decentralisation process in Mozambique falls in line with the above described interest of the ruling political elites of maintaining power. Decentralisation was not part of the 1992 GPA signed by Frelimo and Renamo, but the process has been negotiated and implemented since 1994 with the support of external donors. On the side of domestic political elites, decentralisation was formally accepted primarily because of the need for stabilisation and economic recovery. However, from the very beginning, there has also been resistance within the ruling domestic elites to an excessively far-reaching political decentralisation process that could constitute a threat to their hold on power and to national unity. 80 Although donors tried to push for a more comprehensive model of political decentralisation involving all government levels, the Frelimo government elites succeeded in limiting decentralisation to a dual and gradual process, 81 and in shaping its implementation in practice. This has resulted in a system of governance that runs at two different speeds. 82 Political decentralisation is limited to selected municipalities that have been granted the status of autonomous local government (autarquia) and hold regular elections for the municipal president, council and assembly. 83 Deconcentration primarily involves the decentralisation of tasks and limited power, but scarcely any fiscal decentralisation. The 'deconcentrated' districts with their localities and administrative posts continue to operate under central government rule with centrally appointed district administrators.
Weimer highlights the 'critical juncture' of the peace process of 1990-1994 and the strategic choice made by the ruling elites to redefine the initial model of decentralisation into a two-pronged model of devolution and deconcentration. This entailed high political and administrative costs, producing two classes of citizens and opening the door for future re-centralisation. Weimer concludes that decentralisation is merely 'part of the game' as long as it can be informally instrumentalised by political elites to distribute rents to local 'clients' via deconcentration and thus to secure the political and economic predominance of the ruling party. 84 Plagemann argues that the discrepancy between the formal planned decentralisation supported by donors and the informal implementation by domestic political elites ultimately created unstable formal democratic institutions. 85 The result is an informal recentralisation process taking place within a formal decentralisation framework. 86 While accepting the externally introduced decentralisation process as such, political domestic elites succeed in shaping formal decentralisation structures according to their interests of maintaining power. Donors turn a blind eye on the persistence of informal structures of the former one-party state used to side-line formal structures of democratisation and decentralisation. This indicates how the ruling domestic elites have managed to maintain and extend their power base -a far cry from what was originally intended by donors promoting democratisation together with domestic elites in post-conflict settings.

Informal acting within formal structures
Within the system of deconcentration, the CCs and especially the FDD are widely regarded as political instruments of the ruling party used to strengthen its electoral support at the district level. 87 The process has not been implemented as originally intended by the administration of the Ministry for Planning and Development: 'It got politically hijacked' . 88 There is the widespread opinion that the introduction of the 'seven million' [FDD] has contributed significantly (but not exclusively) to Frelimo's local electoral victories in 2009, even in former opposition strongholds. 89 These criticisms with regard to the CCs and the FDD have been voiced not only among the oppositional and administrative elites but also among donors. 90 Some external donors have resisted supporting the FDD, arguing amongst other that the government may not act as banker. Furthermore, donors suspect that these funds have been misappropriated from donor budget support, side-lining formal procedures. Unease with the FDD and allegations of fraud during the elections led to a 'donor strike' at the beginning of 2010, in which the G19 withheld budget support. 91 This serves as an example of a case in which the national political elites have ignored donor preferences and external donors themselves have not spoken with one voice. 92 Domestic political elites may formally welcome donor activities, but empirical evidence shows, however, that donors themselves are perceived by the domestic political elites to act informally in the sense of an additional sociopolitical force existing outside formal democratic structures. 93 National political elites in Mozambique have been more accountable to external donors than to their own parliament and population. This kind of informal influence may run counter to the unconsolidated Mozambican democracy by reducing the space for participation of the parliament and civil society. 94 Furthermore, there are critical questions regarding the adequacy and sustainability of the implementation of donor-influenced and -supported decentralisation programmes. Due to donors' shifting priorities (e.g. municipal versus district level) and the fragmentation of decentralisation programmes (island projects), the carrying out of such programmes often is seen as unsustainable projects that can lead to informal parallel structures. 95 International and domestic scholars have suggested that the decentralisation process as a whole might fail because of the resistance of the political elites and because of the donors' focus on the technical side, ignoring informal power structures and the political nature of decentralisation. 96 By strengthening the state capacity, donors also strengthen the ruling party's grip on the state: 'Decentralisation will continue to be an avenue for the party state to extend and consolidate control over territory, people, and resources, and also a means to increase its legitimacy' . 97 External donors also often lack consciousness about the political instrumentalisation by domestic elites within formal structures of decentralisation, such as clientelism. In case they get aware of those, there are single examples of donor reactions, for example the 'donorstrike' . However, in the end the formal structures, like for example the CCs, continue to get external support. Domestic elites in turn perceive external donors as acting informal, in the sense of side-lining formal democratic structures, such as the parliament. This perceived legitimation deficit of donors may contribute consciously or unconsciously to resistance to externally introduced democratisation by domestic elites.

Diverging objectives and structures
At the local district level, donors have established and supported a variety of decentralised formal structures with the objective to enhance democratisation. 98 There have been several years of 'island' donor projects of CCs in selected districts, especially in the north of the country. With the introduction of LOLE, the Mozambican government has followed donor objectives and issued a directive to introduce the CCs as a formal space of local governance in all districts at the respective levels -locality, administrative post and seat of district government.
These CCs are composed of leading local figures that are tasked with discussing and approving the district economic and social development plan (PESOD) and to decide on the allocation of resources of the FDD. 99 Currently, the CCs use the fund to dole out small credits for local projects of food production and job creation, submitted by individual local citizens or associations. 100 The FDD is perceived as political instrument to enhance the ruling political elites power at the local level, as well, by donors and local experts. 101 While implementing the CCs as formal decentralisation structures at the local level, the representatives of domestic political elites in the provincial and district secretariats complain that the entire decentralisation process has been donor-imposed. 102 In the local context, too, donors get perceived to pursue their own interests. As a national consultant at district level points out: 'The international organisations also do not have an interest in changing the situation; they want to secure their own existence' . 103 Domestic political elites have taken up donor introduced local decentralisation structures, such as the CCs and integrated them in national policies. However, local formal decentralisation structures, too, get used for the maintenance of power of the ruling political elites, as the FDD shows. Again, donor objectives are also perceived by domestic elites as following self-interests apart from the formal objective of promoting decentralisation and democratisation.

Informal acting within formal structures
CC representatives linked to the ruling political elites use their own direct informal channels within and outside the CC to influence decision-making in their favour. Although formally established above party lines, 104 party-politicisation plays a role when it comes to membership, access and resources concerning CCs and the FDD.
Despite the existence of formal criteria to promote democratic participation, CC members are often selected according to the preferences of the district administrator. 105 Statements such as 'We appoint the people democratically' 106 or 'We know whom to select' are symptomatic. 107 Criticism is seen as attack, and there is a deliberate exclusion of members of opposition parties. As one district official stated, the goal of CCs and the FDD is to 'help the party … I mean, the government!' 108 There is also indirect exclusion, as the local population tends to choose representatives for CCs whom they expect to have the most influence on the ruling government -hence, those linked to the ruling political elites. 109 The language, symbols and way of applauding in CCs stem from the one-party-state era. 110 The notion of 'we and the others' , 111 referring to the ruling party and the former militia, now the main opposition party, prevails. Representatives of the ruling party often have more influence and power than elected representatives. They take part in local decision-making processes despite their lack of a formal mandate. 112 Respondents accordingly speak of the 'partidarisation' of the state. 113 In addition to the distribution of privileges, control, sanctions and fear are also used as informal political instruments to discourage participation. 114 Numerous accounts describe people receiving FDD credit because of personal and party relationships, not because of merit or the viability of their project proposal. 115 As one district official stated: 'The money [FDD] does not satisfy a majority, but a certain group, that's logical' . 116 Donors support the CCs via technical assistance by supplying operational costs, infrastructure and consultancy at provincial and district levels. The empirical evidence shows a somewhat ambivalent interrelationship in terms of the attitudes towards and perceptions of donors by the political elites at the local level. The support of donors is acknowledged by political elites, as otherwise the CCs would not function properly. 117 However, there are examples of CC sessions not taking place because 'we depend on [donor X] to be able to hold the CC session' . 118 This raises questions regarding the dependency on and sustainability of external intervention. 119 Nevertheless, the donor staff is also perceived positively by domestic elites as serving as an informal broker (without formal mandate) between the local and provincial/central government and between the sectoral ministries involved. 120 Respondents bemoan the fact that donors have been excessively technocratic and focused on the technical side of implementing local decentralisation structures, as well as their use of overly ambitious impact indicators oriented toward donor interests of legitimising their actions. 121 Other respondents criticise the fact that donors have created informal parallel support structures that are not formally integrated into existing state structures -e.g. technical councils at the provincial level to advise the CCs. 122 However, some informal structures established by donors, such as thematic working commissions within the CCs, have later been included in the official national guidelines for the functioning of CCs. 123 There have been donor attempts to increase the accountability of the district administration towards its population. In practice, however, the accountability of political elites is directed at donors and higher government levels rather than the local population at the local level, as well. 124 At the local level, as well, external donors focus mainly on technical issues and largely ignore (party-political) clientelism used within formal decentralisation structures for the purpose to enhance the ruling political elite's power maintenance. However, there are examples of local informally implemented features within formal CC structures by the action of individual donor representatives that got adapted by domestic elites and integrated in national policies.

Winds of change
Despite the challenges described above, however, the CCs and the FDD have also had positive effects in terms of promoting local democratic processes. For the first time, districts have formally institutionalised spaces for participation and at least some degree of autonomy regarding the use of resources. As a high-ranking official in the MAE stated, the 'seven million' was foremost a courageous instrument of the ruling elites. According to him, despite the acknowledged technical failures, the pedagogic effect at the local level has been decisive in increasing commitment to participation: 'It is a democratic process, but not in the European sense' . 125 The consultation of the population has become obligatory. Thanks to CCs at various local levels, localities perceive the broader picture of the priorities within the district. Members of the CCs speak of more self-awareness and new possibilities to raise issues. Their commitment is evident as they for example endure long commutes to participate in CC sessions. There is also the potential to increase pressure on local government from the grass-roots level. 126 As a foreign technical advisor at the Ministry for Planning and Development explained, the CCs might not be a substitute for far-reaching local democracy in the sense of elected district governments; however, as long as democratic elections at the district level are not on the political agenda, it is good to have at least some form of participation in the meantime. 127 The FDD functions as a strong incentive for participation, as decisions are taken concerning tangible projects. 128 In summary, the CCs and the FDD can be manipulated and instrumentalised by local domestic elites. This is accomplished by using informal power structures such as personal and party relationships to influence local decision-making processes. Even though their intention is to support democratisation, donors fail to prevent the ruling political elites from co-opting formalised structures and using them to maintain their hold on power. The varying and ambivalent dynamics of interactions between donors and domestic elites found at the local level in terms of diverging objectives, shaping formal democratic structures and instrumentalising them from within reflect the findings at the national level. However, CCs also contribute to raising awareness of participation and accountability opportunities, even though they might not function as donors intended.

Conclusion: informal power structures influence the dynamics of externalinternal interaction
After 10 years of struggle for independence and 16 years of 'civil war' , Mozambique has embraced a path of democratisation with the support of external donors. Within this process, decentralisation has been promoted by domestic elites and external donors as one element of addressing broader objectives of stability, state-building and democratisation. However, empirical evidence at the national and local levels and specifically with reference to the decentralisation process shows that diverging objectives and informal power structures influence the dynamics of the interactions of domestic elites and donors. They affect the shape of as well as the acting within formal decentralisation structures and influence, consequently, the nature of democratisation processes.
Agency in terms of the values, interests and objectives of the domestic elites and external actors involved is used to negotiate norms and institutions at a formal level, but the question of power affects this interaction also at the informal level. As soon as the actors involved perceive the actions of the respective 'other' as informal, the use of informal power structures becomes even more frequent -it is seen as normal. This has consequences at both theoretical and practical levels: In line with the empirical evidence, informal power structures become manifest at various levels: (1) Domestic political elites formally agreed to externally induced decentralisation after the GPA of 1992, but have transformed the formal decentralisation structures according to their own informal political objectives of maintaining power. They have managed to do so by resisting far-reaching political decentralisation policies at the national level and by co-opting formally institutionalised spaces of participation at the local level. In light of the blurred boundaries between the state, government, ruling party, private sector and civil society, political elites use both, formal and informal means of access, resources and relationships to influence decision-making within formal decentralised structures of democratic participation.
(2) Domestic political elites' perception of external interference is ambivalent. Formally they accept co-operation with donors, donor-funding is welcomed; informally, however, there are many voices that perceive external intervention such as donor-funded decentralisation programmes as imposed, paternalistic, unsustainable, lacking adequacy and accountability and as reinforcing dependencies. Domestic elites tolerate external intervention as long as the maintenance of their power is not questioned through the implementation of formal structures of decentralisation.
(3) Donors are perceived by domestic elites as an additional political force influencing government policies. Furthermore, external donors are perceived as pursuing their own interests, such as visibility and providing the image of a functioning peace and democratisation process. Donor influence (e.g. the G19) on the national level is perceived as mostly taking place outside the countries' formal democratic structures and therefore exhibits legitimation deficits; however, this is seen as normal. These perceptions may serve as argument for the use of informal means of side-lining and co-opting formal structures of decentralisation -not only as deliberate strategy but also, consequently, as reaction by domestic elites to externally induced democratisation and the acting of donors.
(4) Donors do not sufficiently take into account the use of informal power structures by domestic elites when it comes to shaping decentralisation structures. They do not prevent the domestic elites' instrumentalisation of the decentralisation process for their objective of maintenance of power. Therefore, processes of democratisation may function different to 'Western' approaches intended and introduced by external donors. Yet, there are members of the domestic political that share common objectives of democratisation. Those need to be identified and supported as change-agents by donors.
(5) Especially the analysis at the local level shows that the establishment of formal decentralisation structures and incentives for participation, such as the CCs and the FDD, may contribute to increased awareness and pressure from below. This may lead to a change of perceptions and attitudes of the involved actors. Consequently, there may be increasingly change-agents at different government levels conciliating the objectives of domestic elites and external donors and modifying decentralisation structures from within towards more space for democratic participation in the long run.
At a theoretical level, these findings correspond with discussions in elite-centred transition studies on the importance of taking diverging objectives, as well as the perceptions and attitudes of domestic elites into account in post-conflict democratisation processes. These findings furthermore are in accordance with approaches of neo-patrimonialism, especially those linked to the micro-level of acting, such as (political) clientelism and rent-seeking. They also reflect discussions in the literature on the deliberate use of informality and on how donors may even support domestic elites' strategies of power maintenance. Thus this article contributes to elites-centred theories on post-conflict democratisation by showing that additional insights are gained when the informal dimensions gets taken into account in analysing the dynamics of the interaction between domestic elites and external actors. Additionally, there are indications that this analysis could also be further linked to the broader discussion of authoritarian regime legitimation and the maintenance of power, specifically with regard to the role that interactions with external actors play in the legitimation strategies of domestic elites in post-conflict settings. 129 Further empirical research is needed to substantiate this argument.
At policy level, the normal of the informal constitutes a general phenomenon that should not be neglected by external actors with regard to democratisation in post-conflict societies. The formal consensus (Higley) between domestic elites and external actors on post-conflict democratisation, therefore, should not be thwarted by an informal consensus on the perpetuation of mutual dependency-structures of domestic elites and external actors or by diverging objectives regarding the implementation of formal democratic structures.
The 'informal' does not necessarily have to limit state-building and democratisation. It seems to be normal and functional for stakeholders -politicians, civil society, the private sector and donors -to use informal structures of influence. The interrelationship between formal and informal is ambivalent, and the two are mutually dependent. Development practitioners in particular must reflect on why this is the case: it may be due to the weakness of formal structures, or the practical experience of more room for manoeuvre to assert one's interests or those of one's group. Informal decisions may be taken to the formal realm to be legitimised. There is an evident recourse to informal processes when formal structures are not regarded as effective and do not facilitate certain decision-making procedures. Any analysis of the interaction between external actors and domestic elites beyond the case of Mozambique should examine who has access to informal means and power structures, whether such instruments are used for the benefit of democratisation, and what might be the intended and unintended impacts in terms of development for the majority of the population.  The Making of Democrats, which compares democratic state-building in post-war Bosnia, El Salvador and Mozambique. 8. However, the 'success story' must be questioned, as decisive aspects of the GPA have never been implemented. Amongst other reasons, this has led to Renamo unilaterally revoking the GPA, with resulting violent clashes, attacks and destabilisation since 2013. The terms of a new agreement were still being debated at the time of writing. 9. In terms of the democratisation process, transitional multi-party elections were held in 1994, followed by general elections -1999-and municipal elections -1998, 2013Manning and