Abstract
State legitimacy - particularly its alleged potential to counter state fragility – has received increasing attention in academic and policy literature concerned with African development. Service provision can substantially influence such state legitimacy. Services, however, are mostly provided by a multiplicity of (state and non-state) providers. This article therefore specifically explores how joint service delivery by multiple providers shapes the attribution of state legitimacy in Burundi by means of two qualitative case studies. Empirically, the article demonstrates, first, that the process of stakeholder interaction, rather than the output of this process, most distinctly shapes state legitimacy and, second, that there are substantial variations in legitimacy attribution by different stakeholders and for different state institutions. Epistemologically, the article suggests three specific challenges that merit attention in further empirical investigation of state legitimacy in fragile settings: the diversity of people's expectations; the artificiality of state/non-state distinctions; and the personification and politicization of state institutions.
State legitimacy is referred to more often when it is imperiled or lacking than when it is thought secure (Barker 1990: 197). It is no surprise, then, that the legitimacy of African states has received increasing attention from policymakers and academics (Abbink 2009: 19–20; AfroBarometer 2012; Brinkerhoff 2010: 66; OECD 2010). This attention follows from a renewed academic interest in state legitimacy as a prerequisite to mitigate (violent) conflict and fragility (Brinkerhoff 2011: 138; Gilley 2006: 499). Mcloughlin (2010: 9) shows that most scholars now recognize that state resilience depends on both capacity and legitimacy. Despite this acknowledgement, state legitimacy has often been poorly understood by actors attempting to support it; it is often assumed, for instance, to be an automatic by-product of effective governance (Mcloughlin 2010: 60). In their meta-analysis of state-building in postwar contexts, Jones and Elgin-Cossart (2011: 18) see “legitimacy” as one of the most pertinent knowledge gaps.
Appreciations of the state partly depend on the extent to which it is able to guarantee services (Brinkerhoff 2011; GSDRC 2009: 1; Hoffmann and Kirk 2013: 32; Sriram and Zahar 2009: 30). In Burundi, basic service provision by the government is seen as the most essential characteristic of democracy by one-quarter of the people (AfroBarometer 2012: 30). 1 Clearly, service delivery is just one element of state legitimacy, and issues such as political representativeness and the provision of security and rule of law are at least as crucial in determining people's perceptions of the state (AfroBarometer 2012: 36–38; Bratton and Chang 2006; Chabal and Daloz 1999; Hern 2013; Milliken and Krause 2002). In this article, we do not contend that it is only or even mostly service delivery that matters for state legitimacy; we merely argue that it also matters and devote this article to exploring how it might do so.
Democracy can certainly not be equated with legitimacy, but in this context democracy indicators provide useful insight into the relevance of service provision for state legitimacy.
This is especially pertinent considering that states in fragile settings are characterized exactly by the lack of political will and/or developmental capacity required to provide services (Roberts 2008: 546). In such situations, non-state actors – ranging from private entrepreneurs to traditional institutions, community-based organizations, international aid providers and criminal organizations – provide alternative initiatives to deliver services, some preceding and others responding to “state failure” (Booth 2011: 13; Englebert 2009; Titeca and De Herdt 2011). Donors have become increasingly concerned with how states can perform a “stewardship” role vis-à-vis such non-state service providers in order to “do no harm” to state legitimacy (Mcloughlin 2010: 84). The question of how such multi-stakeholder service delivery reinforces, or fails to reinforce, state legitimacy nevertheless remains unanswered (Bellina et al. 2009: 37; Jones and Elgin-Cossart 2011: 19). The research on which this article draws sought to address this gap. 2
See www.kpsrl.org for research outputs (theoretical framework and case-study reports).
Burundi is emblematic of the above dilemma. Despite a history of indigenous state formation (Daley 2006: 662) and relatively elaborate state penetration (Ndikumana 2005: 8), colonial divide-and-rule and subsequent episodes of civil war generated a protracted crisis of state legitimacy rooted in authoritarianism, patronage politics and lacking socio-economic development (Ngaruko and Nkurunziza 2000: 405; ICG 2006: 7; Rubli 2013: 6; Uvin 2009: 71). While characterized by an emphasis on political reconstruction and security reforms, post-conflict state-building also invested heavily in service delivery as a “peace dividend” and embraced the idea that state legitimacy follows from service provision (Specker and Briscoe 2010; Vandeginste 2009). Considering the lack of state capacity, and following development paradigms advocating multi-stakeholder service provision, service delivery was often organized as a joint project including public, private and civil institutions (Vervisch and Titeca 2010: 486).
We studied two such local partnerships between state and non-state institutions that aim to improve water and sanitation/hygiene (WASH) services in Burundi. Our case studies suggest
that the projects had only a modest positive effect on the legitimacy of participating state institutions,
that this effect often related particularly to the process (rather than to the output) of service provision and
that the effect differed significantly per group of stakeholders and per state institution.
Beyond these findings, however, the academic significance of our study predominantly lies in the divergence between the challenges we encountered in empirically exploring local state legitimacy and the theoretical representations of state legitimacy on a national level prevalent in the literature. We highlight three epistemological puzzles – the multiplicity of stakeholders’ expectations and interpretations; the overlap between state and non-state institutions; and the politicization and personification of “the state” – and examine these in light of ongoing theoretical discussions on state legitimacy. We argue that studying the relations between service delivery and state legitimacy in contexts characterized by a fragile political setting and multiple state and non-state service providers demands an approach that recognizes the subjective, changeable and relational nature of people's appreciations of and compliance with state institutions. As such, we move away from a normative and essentialist understanding of state legitimacy and underwrite a more constructivist – empirical and relational – approach. Ultimately, we contend, what matters for state legitimacy is not only what state institutions are, do or are capable of doing, but also what they are seen to be and perceived as doing.
The article is structured as follows. We first deal with our methodological and conceptual approach. Then, we present our main findings, which are subsequently discussed and problematized. Following that, we synthesize our empirical findings with our epistemological challenges and relate these to theoretical debates on state legitimacy. We finish by delineating our conclusions. 3
We thank the Peace, Security and Development Network (PSDN) for commissioning the research underlying this article. Gratitude is also extended to ACORD Burundi for its support during fieldwork and to our respondents who so generously shared their time and vision. Irna van der Molen, Alies Rijper, participants of the 10th Development Dialogue and two anonymous reviewers gave inspiring feedback on earlier drafts.
Studying State Legitimacy: Methods and Concepts
Our working definition of state legitimacy follows Papagianni (2008: 49–50), who sees it as “the normative belief of a […] community that an institution ought to be obeyed”, which manifests itself in “the willingness by domestic political elites and the public to support state institutions and to pursue their interests through these institutions”. Defining a concept, however, is easier than researching it. Sidanius et al. (2001: 330) note that “while political theorists have been aware of the importance of legitimacy for the stability of social systems since at least the time of Thucydides and Plato”, state legitimacy “has only recently begun to be conceptualized in ways that make it accessible to the empirical methods of modern social science”. Eloquent theories on state legitimacy abound, but actual operationalizations allowing scholars to explore state legitimacy empirically are still exceptional (Sharma and Gupta 2006: 8; Zelditch 2001b: 36–37).
That state legitimacy is an inherently political concept does not limit its study to the discipline of political science. Considering that state legitimacy is increasingly featuring in development studies (Andersen 2010), its empirical study demands a translation to a more applied social science perspective. This requires three main feats:
disaggregating “the state”, considering there may be differences in the legitimacy of, for instance, local and national or executive and legislative institutions;
disaggregating “the people”, as legitimacy may differ between, for example, elites and masses or groups from different regions; and
shifting the focus from the national to the local level, because it is at the latter that the majority of people “come into contact with ‘the state’ and this is where many of their images of the state are forged” (Gupta 1995: 376).
We operationalized “the state” by breaking it down into “state institutions”: individual organizations that make up the state (ranging from ministries to departments to technical services). Two main distinctions helped categorize the multitude of state institutions constituting a state. First, the level: We differentiated between national and local state institutions. Second, the nature: We distinguished between “general” state institutions involved in generic administrative tasks (for instance, municipalities, village councils and the government) and “specialized” state institutions concerned specifically with WASH (such as ministries and technical or “field” branches). Drawing on work by “anthropologists of the state” (Ferguson and Gupta 2002; Olivier de Sardan 2008; Sharma and Gupta 2006) and by scholars concerned with hybrid stateness (Boege et al. 2009; Hagmann and Péclard 2010), 4 this typology helps transcend the abstraction of the state as an “ontologically coherent entity” and acknowledges the variety of state institutions that are seen as together constituting “the state” (Gupta 1995: 383).
Who, in turn, stand on the shoulders of “classical” political sociologists theorizing the state such as Dahl (1961), Easton (1957) and Nettl (1968).
Just as “the state” is not a unitary entity, “the people” who attribute legitimacy to it are not a monolithic body. Barker (1990: 124–125) finds that “legitimacy is differently composed for different persons and different groups [and] changes both over time and over geographical and social space”. This means we need to go beyond “the distinctive methods of political sociology, from public opinion polling onwards, [which] are adapted to studying the attitudes and behaviour of large, accessible and compliant populations” (Abrams 1988: 65–66). Turning to the work of Tyler (2001) and other psychologists working on political legitimacy, we included both inter- and intra-group dynamics and distinguished between two categories of stakeholders: beneficiaries of the services produced by the projects studied, and organizations participating in the project (public, private and civil service providers and their donors).
Traditionally, state legitimacy was conceived of as a macro phenomenon and discussed as a unitary characteristic on an aggregated national level (Lund 2006: 689; Sharma and Gupta 2006: 8; Zelditch 2001a: 40), and work on state legitimacy still tends to overvalue national conceptualizations. Yet often it is the accumulation of local experiences that makes up a national “state legitimacy” (Olivier de Sardan 2008: 4; Zelditch 2001b: 50–51). For this reason, we deliberately opted to explore the nexus between state legitimacy and multi-stakeholder service provision by studying service provision on the “commune” (sub-provincial) level. 5 Moreover, as development programmes provide important arenas “where representations of the state are constituted and where its legitimacy is contested” (Gupta 1995: 383), we explored the effect of service provision on the legitimacy of Burundian state institutions via two qualitative case studies in the rural WASH sector. 6
Burundi knows many “locales”, ranging from province to colline. We adopted the commune perspective because the projects studied listed the commune as the main implementation level.
We focused on the WASH sector because it is widely recognized as a key area of development. It is both emblematic for the multi-stakeholder nature of service delivery in Burundi and characterized as having a distinct role for the state to play (Castermans 2009). We recognize that multi-stakeholder involvement, and hence legitimacy effects, will be different for other, non-utility service sectors, such as education. Cases were chosen based on an initial mapping by our partner ACORD Burundi and, thereafter, on geographical accessibility and stakeholder willingness. For an elaborate discussion of our case studies and sampling, refer to FN 2.
First, we examined a project to improve the access of marginalized communities to drinking water in Mwumba Commune (Ngozi Province) that included the communal (municipal) administration and the provincial water coordinator. Second, we studied a project to rehabilitate hydraulic infrastructures in Buhinyuza Commune (Muyinga Province) that included the communal administration, the provincial water coordinator and a ministerial WASH department (Programme Eau et Assainissement, PEA). In both cases, the role of the communal administration in the project was to mobilize the population to contribute to the building and maintaining of water points, to support the local water committees in their tasks, and to oversee construction. 7 The role of the provincial water coordinator was in practice limited in both cases, but mainly consisted of providing technical advice to local water committees (Régies Communales de l'Eau, RCEs) and communicating between local and national institutions. The PEA coordinated and implemented reconstruction and provided trainings on water point maintenance.
Rural water provision in Burundi is organized through local water committees assisted by the communal administration. These committees are supervised by a provincial water coordinator working under the auspices of the Direction Générale de l'Hydraulique et des Energies Rurales at the Ministère de l'Energie et des Mines.
Primary data were collected by means of 35 informal meetings with WASH experts and 49 semi-structured in-depth interviews and six focus groups with beneficiaries and organizations that participated in the projects. Respondents were selected through a stakeholder analysis and subsequent snowball sampling. Secondary data encompassed project documentation, sector documentation and academic and “grey” literature. Findings were corroborated by means of a validation workshop with respondents.
Findings
We have presented our cases and results in detail elsewhere (Stel 2011). Here, we limit ourselves to an overview of our main findings. Overall, legitimacy attributed to state institutions based on the projects was modest. Clearly, the projects were just a minor issue in people's (re)consideration of appreciations of state institutions. The effects the projects did have on legitimacy tended to be positive (as we explain below); however, this has to do more with low initial expectations than with remarkable appreciations. We found, first, that the effect of the projects on state legitimacy appeared to be determined more by the process of stakeholder interaction than by output in terms of services. Second, different stakeholder groups attributed legitimacy in different degrees and based on different considerations. Third, the extent and nature of legitimacy attributed as a result of participation in the projects varied substantially per state institution.
The Prioritization of Process over Output
The OECD (2010: 8–9) defines input or process legitimacy as related to procedural concerns and output or performance legitimacy as stemming from efficiency considerations, and emphasizes that the interaction of these different sources of legitimacy is crucial. Yet despite their intertwined nature, input and output affect stakeholders’ perceptions in distinct ways. In both of our cases, it was the process of interaction, coordination and joint implementation rather than the projects’ concrete effects on service delivery – increased or improved drinking water – that impacted people's perceptions of the state.
This seems counter-intuitive because, in a poor country like Burundi, one expects that the fact that WASH services are provided is more important for people than the way in which they are provided. Indeed, our findings do not deny that this is the case. In a situation of meeting basic needs, people care more about outcomes than about processes. And it was only when we started probing into a specific element of service delivery – namely, the role of the state and people's evaluations of it – that process started to surface as the key determinant. It was when we zoomed in on how experiences with joint service delivery affected people's perceptions of the state, rather than, for instance, their general appreciation of their standard of welfare, that the process was referred to more than the actual outcomes.
We suggest that this perhaps paradoxical emphasis on process legitimacy is related to expectation patterns. Respondents did not expect the state to arrange services directly, but rather looked to RCEs, non-governmental organizations (NGOs) and donors for WASH services. The state was being held responsible for how the projects were being run and implemented more than for what they eventually yielded. As a result, process-related issues concerning decision-making, communication, accountability and responsiveness were mentioned more often as reasons for changing perceptions of state institutions than were matters regarding output. As noted before, this is not to say that respondents did not value, or even prioritize, the performance of the project over its process. Indeed, a reasonable output seemed a prerogative for the attribution of any kind of legitimacy. However, when it came specifically to the judgement of state institutions, process-related indicators took precedence.
This can be explained in that while service-related expectations towards the state varied per case, region and group, respondents emphasized the state's responsibility in terms of facilitation. People found that a “good state” would provide services, indicating an ideal revolving around the performance of service delivery, but when asked for their concrete expectations from state institutions, such expectations more pragmatically reflected a supportive role for the state in terms of process management. Indeed, according to the AfroBarometer (2012: 36–37), only 11 per cent of Burundians responded that WASH should be the government's first or second priority. This suggests that expectations are shaped by both an “ideal-type role definition” between state and “civil society” and an understanding of the perhaps utopian nature of such a division of tasks (Brass 2012: 213). Respondents’ faith in the capacity of state institutions to provide material rather than managerial resources was weak across the board. This is not surprising in light of the conclusion that more than half of the people in Burundi think the government does a fairly or very bad job in providing WASH services (AfroBarometer 2012: 40). 8 In line with this, expectations for the communal administration with respect to WASH by and large revolved not around implementation, but rather around mobilizing funding and donors, supervising contractors, supporting the RCE by mobilizing the population to contribute labour and material, and enforcing law and order. Similarly, the provincial water coordinator was supposed to supervise and technically support the RCE, in addition to lobbying national policymakers. Thus, respondents put less emphasis on the state's responsibility to “build water points” than on the state's role in mobilizing, supervising and facilitating those organizations that were expected to “bring the water”.
The AfroBarometer (2012: 26) shows that most Burundians would prefer to pay fewer taxes even if this means fewer government services.
In Mwumba especially, more positive associations with the local state followed not so much from increased quantity or quality of drinking water – by and large seen as the success of the NGOs participating in the project – but from the fact that “people from the administration no longer demand money or beer before they address an issue” and that while “before, you could not go early in the morning to see the chef, because he kept on sleeping, currently, you can go even during the night and he is ready to help”. 9 Legitimacy was associated quite directly with accountability of the state towards its citizens, which subsequently led to paradoxical situations. The project in Buhinyuza, for example, generated significant social unrest (ranging from competition among NGOs to accusations of land-grabbing) that undermined its performance. Such tensions, however, at the same time gave local authorities the chance to acquire legitimacy by successfully dealing with complaints.
Focus group, Mwumba, 17 November 2011. This corroborates the findings of the AfroBarometer (2012: 32) on corruption.
The above challenges the resource-oriented perspectives that posit legitimacy as the direct outcome of capacity and utility and that stress the importance of “objective” distributional outcomes for awarding legitimacy (Brinkerhoff 2010: 71; Hameiri 2007: 136). Hayward and Dumbuya (1983: 651) seem to underline our findings in the conclusion of their study of political legitimacy in West Africa: “Many of the efforts to bring about concrete improvements in performance failed”, yet regimes found other “ways to increase public support”. Moreover, recent psychological studies show that “the roots of legitimacy lie in people's assessments of the fairness of the decision-making procedures used by authorities and institutions”, and that “the outcomes themselves have little impact upon people's evaluations of the legitimacy of those authorities or the institutions they represent” (Tyler 2001: 416–420). Referring to social identity theory, Tyler (2001: 422) demonstrates that “a person's relationship to organizational authorities is to an important degree an exchange of status-relevant information rather than an exchange of resources”.
Stakeholder Variation
Whether and in what way respondents changed their perceptions about state institutions based on the projects was closely related to differences in expectations from and experiences with state institutions on the one hand and differences in knowledge of the projects on the other.
Beneficiaries tended to be quite critical of state institutions. However, because their expectations of the state were quite low, negative evaluations of state institutions appeared not so much to decrease state legitimacy for beneficiaries as to confirm their already bleak reputation. The state, so to speak, had little to lose in terms of legitimacy. Conversely, even minimally positive evaluations could potentially enhance state legitimacy in the eyes of beneficiaries – although, as we show below, positive behaviour of state institutions was often either attributed to non-state institutions or demeaned in light of existing scepticism. Generally, beneficiaries appeared relatively ill informed about the division of responsibilities within the projects and hence evaluated participating state institutions in a rather generic sense. However, beneficiaries who were more involved in the projects – because they participated in trainings or were employed as workers, or merely had an interest in following-up developments – had a clearer understanding of what state institutions should do and thus held more specified benchmarks for evaluation. Overall, however, beneficiaries were not all that concerned with state institutions anyway. Their priority was getting water and they associated this with the projects’ NGOs rather than with state institutions. Beneficiaries did not seem to perceive the state as a direct implementer of WASH projects, but rather a more distant spoiler or enabler of it. Respondents mentioned the importance of the state maintaining law and order so that WASH projects would be able to function in the first place and saw “attracting donors” that would send “their” NGOs to implement WASH projects as a key responsibility of state institutions. 10 State institutions did not have to necessarily “do” WASH themselves to acquire legitimacy, but could meet expectations by making sure WASH was “being done” by others. A provincial water coordinator said that “when a non-state actor provides water, this is a success for us, too”. 11
Focus group, Muyinga, 8 December 2010.
Muyinga, 2 December 2010.
The private and civil society institutions participating in the projects had a different relation with state institutions than did beneficiaries. Their more intimate working relations resulted, in our case studies, in a relatively high awarding of legitimacy to state institutions by non-state project partners. 12 A representative of an NGO participating in the Mwumba project, for instance, was convinced that “the time we spent with local government actors has not been lost!” 13 In Bunhinyuza, an RCE technician praised the “good team management” of the PEA. 14 Thus, for project partners, despite their different relationships with state institutions, legitimacy attribution hinged as much on the management role of state institutions as it did for beneficiaries. NGO representatives in the Mwumba case study consistently stressed that the state should be seen as a “coordinating entity existing by the grace of cooperation with the implementing actors”. 15
This result is probably slightly inflated by strategic considerations: Project partners tended to sometimes perceive our study as a “project evaluation”.
CARE, Ngozi, 8 November 2011.
8 December 2010.
AVEDEC, Ngozi, 7 November 2011.
Differences across State Institutions
Legitimacy was mostly attributed to state institutions participating in the projects. However, as we discuss in the next section, some state institutions that were not part of the projects still reaped legitimacy from them and, conversely, many state institutions that were officially part of the projects did not. Moreover, the legitimacy of local state institutions was considerably more affected than that of national institutions. The effect on “general” state institutions versus that on specialized, WASH-related, state institutions varied per stakeholder group. These findings are in line with Sabatier's (1991: 148) observation of the substantive differences in implementation and evaluation dynamics across policy fields (see also Mayntz 1998).
Generally, people attributed legitimacy to that which they knew. This is why, with regard to beneficiaries, the projects mostly affected the legitimacy of the – very visible – communal administration (a local, general state institution) even if this does not necessarily reflect the rather modest role the communal administration played in the projects. Beneficiaries were not aware of national ministerial structures nor were they familiar with the complicated local WASH governance system. They did, however, experience the conduct of the communal administration throughout the projects and noted that the members of the administration tried to set good examples by cleaning water points and working to limit and sanction theft of materials for water points.
Following a similar logic of “unknown is unloved”, national state institutions involved in the projects did not gain or lose legitimacy for beneficiaries. Yet, the provincial water coordinator, a local specialized institution, in some instances gained some legitimacy for project partners who took the state WASH structure as their main reference frame. On a national level, specialized state institutions such as the ministry did not receive much legitimacy on behest of the projects studied. Local state legitimacy thus need not “trickle up” to broader conceptions of state legitimacy (Thawnghmung 2003: 19). The claim that “decentralised state structures that effectively provide services and security can enhance government legitimacy” (Brinkerhoff 2011: 139) might thus be too confident. The PEA, a national state institution specialized in WASH and directly involved in the local implementation of the Buhinyuza project, however, was a partial exception. Project partners indicated they had changed their opinions on the PEA and hence, we assume, altered their legitimacy attribution as a result of the project. However, for beneficiaries, as we show below, this was different, because they were mostly unaware that the PEA was in fact a state institution.
A pertinent example of how far removed “the state” is from a monolithic entity and how contradictory the effect of service delivery on state legitimacy can be is constituted by an episode from Buhinyuza. There, while the quantity of water available was too small to satisfy the needs of all beneficiaries in the commune, the police, a key state institution, demanded a share of water that exceeded what would be a fair distribution. The communal administrator, another crucial state representative, tried to resolve the problem by granting the population access in the morning and the police in the evening, a measure met with, if not approval, at least consent. This episode had the paradoxical effect that, on the one hand, “the state” lost legitimacy due to the conduct of the police, while, on the other, it gained legitimacy through the demeanour of the administrator.
Epistemological Reflections
Rothstein (2007: 12–13) eloquently illustrates some of the problems that scholars interested in state legitimacy face:
The complexity of political legitimacy as a concept means that it is very difficult to operationalize and measure it in empirical research. Surveys are often used, but when a person states that she does not have confidence in a political institution, this is not necessarily a good measure of lack of political legitimacy. The reasons for having low confidence in, for example, the Parliament may be a) a sign of the type of healthy scepticism against authorities that we often think is a democratic virtue, b) that you dislike the current majority but not necessarily the institution as such c) that you support the majority and the institution but are upset about some recent decision or scandal.
Our Burundian cases highlight three specific manifestations of the complexity of state legitimacy that are related to the three findings outlined above. The first concerns the multiplicity of expectations and experiences. This is related to the variation in legitimacy attribution among stakeholder groups. The second issue regards the overlap between state and non-state institutions, which corresponds to variations in awarded legitimacy across state institutions. The third issue, captured under the “politicization and personification of state institutions”, loosely reflects our finding that the process of project implementation is in some ways more important in respondents’ evaluation of state institutions than project output is.
The Multiplicity of Expectations
Variations in legitimacy attribution across our stakeholder groups could mostly be explained by differences in what each group expected from the state and had previously experienced with the state. The prioritization of process issues over performance indicators in respondents’ assessments of state legitimacy was tied to specific expectations reflecting a post-conflict reconstruction logic (meaning, NGOs and entrepreneurs build water points, and state institutions “attract” NGOs and manage and control implementation). Different stakeholders had different levels of information and knowledge of the role of state institutions in the project and thus arrived at different expectations and evaluations of the state. Hayward and Dumbuya (1983: 650) similarly find that “expectations about the leaders and their anticipated policies […] set standards for the validation or justification of a particular regime”. Bellina et al. (2009: 3) note that “whatever processes a state may organize, and whatever amount of goods and services it may deliver, a central point is the […] expectations that people have of the state”. However, contrary to what Johnson et al. (2006: 72) hypothesize, we showed that these expectations are not widely consensual (see also Hern 2013: 480).
Differences in expectations and evaluations, moreover, are not limited to stakeholder groups. We found that people in different localities have different experiences with, and hence expectations of, state institutions (Bellina et al. 2009: 13). Expectations vis-à-vis “the state” were more elaborately defined in Mwumba than in Buhinyuza, probably because it is President Nkurunziza's maternal commune. In Buhinyuza, poorer than Mwumba and characterized as marginalized by the state, expectations towards the state were more modest. Past experiences thus constitute a “legitimacy threshold” (Stel et al. 2012: 54) that determines how the role of the state in concrete service provision activities is interpreted and might affect state legitimacy. This can generate a cycle of self-reinforcement wherein state institutions already considered legitimate are better positioned to reap legitimacy from specific development projects which in turn will boost existing legitimacy reservoirs. State institutions lacking legitimacy to begin with, conversely, are less likely to convert involvement in development projects into broader legitimacy (even if this involvement is positive). In our cases, negative performances of the projects were often intuitively attributed to state organizations involved, while positive performance was not associated with the state, but with other partners. In Mwumba, for instance, respondents’ accounts suggested that when water points worked properly, the RCE was thanked, but when they malfunctioned, the administration was blamed. A WASH expert said, “When there are problems regarding water, people blame the administrator, not the president of the RCE.” 16 The appreciation of the PEA also exemplifies how existing perceptions of state institutions colour interpretations of state conduct in specific projects. One of the reasons the PEA was not assumed to be a state institution (but considered an NGO, as we discuss below) was exactly because its field officer was doing a good job in establishing relations with the population, which was considered atypical for state officials.
International Committee of the Red Cross, Bujumbura, 19 November 2010.
In analysing state legitimacy, the key question is thus whose expectations are taken as guiding (Sharma and Gupta 2006: 27). “To weigh the views of all citizens equally in measuring legitimacy”, as Gilley (2006: 501) proposes, or to see “what is legitimate in the eyes of citizens”, as suggested by Hameiri (2007: 136), is less straightforward than often assumed. Fault lines might include those between domestic and international groups (Papagianni 2008: 50); between people who are to varying degrees professionally affiliated with the state (Barker 1990: 110; Brinkerhoff 2011: 146); and between elites and the populace (Hoffmann and Kirk 2013: 29; Sriram and Zahar 2009: 30). While these distinctions seem instinctively obvious, “very little research has been done on the relationship between diversely located groups of people and their employment of […] varying resources of cultural capital in imagining ‘the state'” (Gupta 1995: 390).
The Artificiality of the Distinction between State and Non-State Institutions
When taking subjective and multiple expectations as the premises for exploring state legitimacy, another theoretical-methodological challenge surfaces: What is the state? The differences between degrees and forms of legitimacy attributed to local and national and general and specialized state institutions underline the need to distinguish between de facto statehood (those components of the state recognized as such by stakeholders) and de jure statehood (the amalgamation of institutions legally constituting the state) (Brinkerhoff 2011: 139). What matters for state legitimacy is not whether an organization or individual is formally “public” but whether it is recognized as such by the constituents that can award it legitimacy.
Overlap between state and non-state institutions is usually discussed in terms of “traditional” institutions (Englebert 2002; OECD 2010: 43–44; Rubli 2013: 15). In Burundi, for instance, the government co-opted the Bashingantahe (“wise men”) to enhance postwar unity (Ntsimbiyabandi and Ntakarutimana 2004). The intertwining of state and non-state institutions, however, is also relevant concerning “modern” institutions (Lund 2006: 686), especially in light of the ascendancy of New Public Management and neoliberal public policies (Brass 2012: 209). The postwar “mushrooming” of civil society organizations (CSOs) in Burundi has had considerable repercussions in terms of people's perceptions of what constitutes the “public sector” and “civil society” (Vervisch and Titeca 2010: 493). We repeatedly came across CSOs perceived as state institutions as well as state institutions regarded as NGOs. This shows how the conduct of some officially non-state institutions can directly influence state legitimacy and how the behaviour of particular formal state institutions might not have any effect on state legitimacy.
A telling instance of state/non-state conflation concerned the local water committees participating in both projects. While these RCEs cooperate closely with the communal administration, they are not part of the state and are officially CSOs. Many of the professionals working in the WASH sector are aware of this, but the majority of beneficiaries – and some local state authorities – did not distinguish between the RCE and the communal administration and saw the RCE as a public provider. This had to do with prosaic considerations such as the RCE's use of communal stamps and its holding office in the same building as the administrator – an observation that highlights the significance of “banal techniques of representation” for attributions of state legitimacy (Sharma and Gupta 2006: 15). Thus, the conduct and performance of the officially non-state RCEs – for better or worse – directly affected people's image of the local state.
Another state/non-state mix-up amongst respondents is constituted by the PEA, the ministerial programme implementing hydrological reconstruction in Buhinyuza. For various reasons, such as the PEA's project-based working modality and its lack of a permanent field office, many beneficiaries considered the PEA an NGO. The stately nature of the PEA was even considered ambiguous by PEA representatives because the PEA is concerned with implementation, not with designing policy, and because it is only partly state-funded (receiving most of its funding from, and owing accountability to, UNICEF). As such, while the PEA is officially part of a ministry and a public institution, it is doubtful whether the demeanour of the PEA has a direct influence on people's perceptions of the state (besides the general credit awarded to the local state for “attracting” such an “NGO”).
The above examples go beyond “confusion” among our respondents. Rather, they suggest that “we must take seriously the elusiveness of the boundary between state and society, not as a problem of conceptual precision but as a clue to the nature of the phenomenon” (Mitchell 1999: 85–86). Bratton and Chang's (2006: 1075) “officials of the state” and “leaders in the community” might not be as dichotomous as they suggest. Brass (2012: 211) convincingly documents how the “blurring boundaries” between NGOs and the state through the integration of former NGO leaders into the government makes the “lines between the public and the private” quite indefinable for beneficiaries (see also Bratton 1987). Our examples are an extension of Lund's (2006) “twilight institutions” that have “fluctuating degrees of stateness”. As also suggested by Brass (2012: 221) for Kenya, the often-heard fear, also expressed by Burundian policymakers, that NGO involvement “crowds out” state institutions by reinforcing citizens’ perceptions of state incapability seems too simplistic (Rosser 2006: 11). Rather, in our cases, state legitimacy was often derived from association with NGOs, as state institutions were credited for “bringing in” these bagiraneza, or “benefactors”. 17
Decentralization expert, Ngozi, 9 November 2010.
Politicization and Personification
Respondents’ conceptions of state legitimacy were further complicated by symbiotic relations between ruling party, government and state, which might partly account for the premium our respondents put on the behaviour of state institutions and the process of project implementation (Englebert 2002: 101; Vandeginste 2009: 67). 18 In Burundi, according to the International Crisis Group (2006: 7), the ruling CNDD-FDD 19 made far-reaching changes in government arrangements that resulted in a situation in which “the state and the party are two faces of the same coin”. As also noted by Bratton and Chang (2006: 1061), this makes it challenging to differentiate between state and government (or even party) legitimacy as the OECD (2010: 31) urges us to do. Patronage and clientelism are no new phenomena in the discussion on state legitimacy (Englebert 2002: 101; Olivier de Sardan 2008). Yet, the implications of politicized and personalized legitimacy have largely escaped the methodological discussion on state legitimacy (Bellina et al. 2009: 4).
Abbink (2009: 11) describes a similar logic for Ethiopia.
Conseil National pour la Défense de la Démocratie – Forces pour la Défense de la Démocratie.
Bringing to mind Booth's (2011: 13) “populist policy initiatives, especially by presidents before and during election campaigns”, the president's campaign to construct 25 new water points in each commune is the only WASH “policy” beneficiaries were aware of. People furthermore referred to situations where obligatory communal works to maintain water points were organized back-to-back with CNDD-FDD gatherings. This confirms observations by Van de Walle and Scott (2009) that “the design of public services is a far more political matter than is often recognized”. Legitimacy is then inherently personal rather than institutional (Hoffmann and Kirk 2013: 24). Respondents consistently referred to either positively or negatively appreciated personal conduct by local state authorities when asked about state legitimacy. Moreover, people spoke of the state in terms of individuals rather than structures: the “minister” rather than the “ministry”, and “the administrator” instead of “the administration”. Similar dynamics are documented by scholars working on “street-level bureaucracy” (Lipsky 1980; Berenschot 2010), leading Thawnghmung (2003: 5) to conclude that legitimacy and popularity are more interchangeable than dominant theoretical conceptions of legitimacy claim.
Towards an Empirical and Relational Approach to State Legitimacy
If there is one central inclination in the above it is that legitimacy, in the field of service delivery on which we focused, is not attributed based on “reality” but on perception. We wholeheartedly agree with Hoffmann and Kirk's (2013: 35) conclusion that “legitimacy has to do with people's norms, beliefs and everyday interactions”. Thus, as Mcloughlin (2010: 84) aptly stresses, “ensuring that the state's role in service delivery is clearly communicated is key”.
On the one hand, such communication was often under-emphasized in our cases. The role of state institutions was mostly not communicated plainly to beneficiaries. NGOs’ and donors’ public relations efforts in some instances overshadowed state involvement (see also Bellina et al. 2009: 31; Lake 2007: 42). On the other hand, Burundian state institutions knew how to shape people's interpretations of service delivery: Local authorities often “explained” a specific project in a way that exaggerated the state's involvement in it (a dynamic corroborating Sabatier's [1991: 148] theories on the dynamics between “policy elites” and the public). A respondent elucidated, “You have to understand the logic of the communal administration: When working in the WASH sector, you always need the support of the chefs des collines; they are crucial for communicating with the population.” This sheds more light on the Afro-Barometer's (2012: 10–12) finding that while most Burundians are (moderately) interested in public affairs, the majority also sees government and politics as too complicated to understand and adheres to a vision of the government as “a parent [that] should decide what is good for us”. Reflecting the importance of ceremonial rituals as described by Sharma and Gupta (2006: 18) and Hayward and Dumbuya (1983: 649), an expert explained that when the president donates materials or when ministers or members of parliament attend inauguration ceremonies of projects, the local population will associate the benefits of such programmes directly with the government. In this way, projects can boost the legitimacy of state institutions that do not even participate in them.
Thus, for people to positively change their appreciation of the state as a result of a service provision project, they must 1) recognize the role of a specific organization in improved services, 2) acknowledge the stately nature of this organization and 3) regard the organization positively (Stel et al. 2012). Crandall and Beasley (2001: 77) describe this attribution process through a mechanism they call “unit relationship”, whereby people come to see acceptable and unacceptable actions as intrinsically linked to the actor they hold responsible for it (see also Sabatier's [1991: 148] notion of “substantive policy information”). This is captured by the dual ideas of “responsibility” and “controllability” that stipulate that only when people perceive a certain actor as obliged and able to perform a certain role or to provide a certain resource will they judge them regarding this role or resource. Our cases show that in many situations the establishment of a “unit relationship” between service delivery and state institutions does not take place, because people are either not aware of or not interested in the role of a certain actor in a certain process, or because they do not perceive the necessary responsibility or controllability.
In emphasizing the importance of perception, persuasion and communication, our case studies thus underwrite an empirical rather than a normative approach to legitimacy and substantiate relational rather than essentialist conceptualizations of legitimacy. The most fundamental dichotomy in the literature on state legitimacy is that between analytical-descriptive and normative-prescriptive visions of state legitimacy (Andersen 2010: 3). Whereas the empirical school opts for “an understanding of legitimacy that directs attention to diverging beliefs and perceptions”, the normative line of thinking sees state legitimacy as a set of relatively universal measurable ethical standards (Bellina et al. 2009: 6; Hern 2013: 477). Our findings lend substance to the empirical approach to the analysis of state legitimacy that allows for what the OECD (2010: 34) calls “normative pluralism” because these findings illustrate the centrality of constituencies’ context-dependent expectations and perceptions (see also Olivier de Sardan 2008: 15; Lund 2008). We demonstrated how the framing and visibility of state institutions in joint service provision arrangements shaped notions of state legitimacy more directly and profoundly than the de facto roles or de jure mandates of these institutions. As such, a distinction between “objective” (actual) and “subjective” (perceived) legitimacy (Gilley 2006: 501) proves deceptive: “State legitimacy itself depends on what states mean to their citizens” (Sharma and Gupta 2006: 12). Our insights confirm that potential sources of legitimacy are effective only to the extent that relevant constituencies consider them to be so, and not to the extent that political theorists or policymakers determine they should be (Bellina et al. 2009: 3; Lake 2007: 21).
With the exception of structuralist accounts of state legitimacy – such as Englebert's (2002: 4), which sees state legitimacy as “a historical, structural condition” – the literature now takes the position that legitimacy is continually constructed and contested in the state's interaction with other societal actors (Lund 2006: 693). Research has repeatedly shown that “legitimacy does not follow from the office or position of the ruler but from a bargain between ruler and ruled”; it is not a condition, but “a variable that exists in greater or lesser degrees in different times and places” (Gilley 2006: 501; Lake 2007: 19–21). Weber himself acknowledged that the state's legitimacy is more dependent on tax collectors’ relations with citizens than on tax policies, emphasizing the importance of relations over regulations (Rothstein 2007: 21). The divergence between legitimacy of national, local, general and specialized institutions that we noted confirms the importance of this relational component of state legitimacy: Institutions all have different relations with their constituents and are hence awarded different forms and degrees of legitimacy. Conceptions of legitimacy as a property of “the entire state apparatus” (Englebert 2002: 4) seem untenable in this light.
State legitimacy, regarding the domain of welfare, moreover, is not just relational, but relative – dependent on comparisons over space and time (Lund 2006: 693). The OECD (2010: 15) notes that an institution is legitimate to the extent “that people regard it as satisfactory and believe that no available alternative would be vastly superior”. Indeed, in Burundi, it is exactly the existence of such alternatives (that is, non-state services providers) that lends salience to the question of state legitimacy. In our cases, the quantity and quality of the alternatives (RCEs and NGOs) at hand directly influenced state legitimacy, not least because state legitimacy was often derived by proxy via the work of NGOs or RCEs.
Conclusions
We explored the ways in which Burundian multi-stakeholder WASH projects have impacted the perceptions and appreciations of the state on the part of different groups of stakeholders. The importance of how joint service delivery projects are implemented (vis-à-vis what they yield) and the variation in legitimacy attribution per stakeholder group and per state institution was emphasized. Subsequently, we identified three epistemological challenges of empirically studying state legitimacy as a factor of multi-stakeholder service provision in fragile settings that both surfaced in our Burundian cases and are insufficiently addressed in the literature. These are: the diversity of people's expectations, and thus evaluations, of the state; the overlap between state and non-state institutions; and the politicization and personification of state institutions.
In short, we argued that state legitimacy depends on the degree to which specific societal expectations are perceived as being met by a specific state institution in the eyes of a specific group of people in the context of a specific process in comparison with specific alternatives. As such, our findings question the often overly aggregated and generalized references to “state legitimacy” that proliferate in academia and policy – ranging from state fragility and state-building indexes 20 to calls for cross-national datasets on state legitimacy and replicable ways to measure it (Bratton and Chang 2006: 1080; Berger et al. 1998; Englebert 2002: 7; Gilley 2006: 521). Our insights support the nascent move towards a more constructivist approach to state legitimacy drawing on ethnographic understandings of the state and psychological approaches to legitimacy. Our cases illustrate the importance of local perceptions of what constitutes legitimate socio-political practice (OECD 2010: 11, Lund 2008) and underscore the pertinence of empirical research on the “complexity, variety, ambiguity and modernity of the behaviour of state agents in Africa” (Olivier de Sardan 2008: 1). Legitimacy, intrinsically, is in the eye of the beholder (Bellina et al. 2009: 34).
Such as the “State Fragility Index” (George Mason University); the “Fragile State Index” (Carleton University); the “Failed State Index” (Fund for Peace); and the “Index of State Weakness” (Brookings Institution).