Ethnicity matters: Development in Burkina Faso between Symbiosis, Favouritism and Exclusion

(1)

Ethnicity matters: development in

Burkina Faso between symbiosis,

favouritism and exclusion

Max Lawaly Kuipers

Master thesis for the Research Master in International Development Studies

Graduate School of Social Sciences

University of Amsterdam

Supervisor: dr. A.F.M. Zaal

Second reader: dr. N.R.M. Pauw

6 July 2020

(2)

1 ABSTRACT

This research explores inter- and intra-ethnic group relations and their effects on development interventions in Burkina Faso, a country with a large variety of ethnic groups. The country, housing over 60 different ethnic groups has a long history- of international and national development initiatives being performed. Much of the literature involving ethnicity, ethnic groups and development conclude that ethnic diversity is in many cases counter developmental. How does this assumption hold up in Burkina Faso? Through a convergent mixed-methods design, consisting of unstructured and semi-structured interviews and a survey, data was acquired from former- and current development professionals with experience in Burkina Faso. During a fieldwork period of four months, interviewees were acquired through a method of snowballing, while surveys were distributed online and in person within seven months. Two narratives with regards to the effects of inter- and intra-ethnic group relationships were found. The first narrative describes how ethnic groups, and especially the largest present in the country, tend to favourite members of their own groups in the selection of recipients and the corresponding allocation of resources stemming from development intervention. The second narrative, conversely, denies that biased selection of recipients and allocation of resources is a result of inter- and intra-ethnic group relationships but rather a consequence of corrupt individuals that select recipients and allocate resources along their own personal networks. Another finding stemming from this study is that discrepancies exist between perspectives of research participants along the lines of their age and nationality. Overall, it was concluded that intra- and inter-ethnic group relations affect development initiatives in terms of exclusion from participation to developmental initiatives, from access to resources and through various forms of preferential treatment within and between groups. The findings and conclusions of this research contribute to the body of literature dealing with the relationship between ethnicity and development. The results seem to be relevant to policy makers, development professionals and others engaged in pursuing development goals in Burkina Faso and elsewhere. Keywords: Ethnicity, Ethnic groups, Inter-ethnic, Intra-ethnic, Development, Burkina Faso, Fulani, Peulh, favouritism, exclusion, symbiosis.

(3)

2

Acknowledgements

This thesis was made possible with the help of many. I would like to thank everyone who took the time to fill in the survey, circulate it around their networks and thus ensured that so many respondents participated. I would also like to express my gratitude to all those who shared their knowledge and took the time for an interview, in Burkina Faso, in the Netherlands and over Skype.

I am grateful for all the informal, casual and sometimes heated conversations and discussions with friends and family on this topic – both in Burkina and the Netherlands. It was clear that the subject of my thesis did not only interest me, but that many had an opinion about it and wanted to discuss. A million thanks to Annewies for supporting me throughout and to Fiona and Dirk for reading with me and correcting and strengthening my arguments where possible. Lastly but most importantly, I would like to thank my supervisor Fred Zaal for his useful and straight to the point feedback, and his unwavering enthusiastic support.

List of acronyms

CA – Correspondence analysis

CDP – Congrès pour la Démocratie et le Progrès [political party] CVD – Conseils Villageois de Développement

FCFA or CFA – Communauté financière africaine franc [official Currency of Burkina Faso] IWRM – Integrated Water Resource Management

MPP – Mouvement pour le Peuple et le Progrès [political party] NGO(s) – Non-Governmental Organisation(s)

ODA – Official Development Assistance

RSP – Régiment de Sécurité Présidentielle [former presidential military unit] UN – United Nations

UPC – Union pour le Progrès et le Changement [political party] WASH – Water, Sanitation and Hygiene

(4)

3

List of figures and pictures

List of pictures

Picture 1: 'Ne vote pas Ethnie SVP'. [Don't vote based on ethnicity please]. Graffiti in Ouagadougou. Picture by author, Ouagadougou échangeur de l'Ést, October 2019

List of figures

Figure 1: Age distributions among respondents of the Full-, Burkinabé- and non-Burkinabé sample Figure 2: Pie chart of distribution of nationalities in the Full survey sample

Figure 3: Pie chart of distribution of ethnic groups among Burkinabé survey respondents Figure 4: Percentage of respondents in specific samples indicating their development sector Figure 5: Percentage of specific sample indicating working in a specific organisation type Figure 6: Distribution of respondents across function within their organisation

Figure 7: Distribution of responses among full sample to the question: ‘My organisation briefed me with information about inter-ethnic group relationships existing in Burkina Faso’.

Figure 8: Distribution of responses among Burkinabé sample to the question: ‘My organisation briefed me with information about inter-ethnic group relationships existing in Burkina Faso’. Figure 9: Distribution of responses among non-Burkinabé sample to the question: ‘My organisation briefed me with information about inter-ethnic group relationships existing in Burkina Faso’.

Figure 10: Distribution of responses among full sample to the question: ‘I took/take account of Inter-ethnic group relationships in my work’.

Figure 11: Distribution of responses among Burkinabé sample to the question: ‘I took/take account of Inter-ethnic group relationships in my work’.

Figure 12: Distribution of responses among non-Burkinabé sample to the question: ‘I took/take account of Inter-ethnic group relationships in my work’.

Figure 13: Biplot displaying attitudes towards: ‘I took/take inter-ethnic group relationships into account in my work’ and how they relate to specific origin categories respondents belong to.

Figure 14: Distribution of responses among full sample to the question: ‘My organisation briefed me with information about intra-ethnic group relationships existing in Burkina Faso’.

Figure 15: Distribution of responses among Burkinabé sample to the question: ‘My organisation briefed me with information about intra-ethnic group relationships existing in Burkina Faso’.

Figure 16: Distribution of responses among non-Burkinabé sample to the question: ‘My organisation briefed me with information about intra-ethnic group relationships existing in Burkina Faso’.

Figure 17: Distribution of responses among full sample to the question: ‘I took/take account of Intra-ethnic group relationships in my work’.

Figure 18: Distribution of responses among Burkinabé sample to the question: ‘I took/take account of Intra-ethnic group relationships in my work’.

Figure 19: Distribution of responses among non-Burkinabé sample to the question: ‘I took/take account of Intra-ethnic group relationships in my work’.

Figure 20: Distribution of responses among full sample to the question ‘Inter- and intra-ethnic group relationships affect the allocation of resources stemming from development interventions.

(5)

4 Figure 21: Distribution of responses among Burkinabé sample to the question ‘Inter- and intra-ethnic group relationships affect the allocation of resources stemming from development interventions.’ Figure 22: Distribution of responses among non-Burkinabé sample to the question ‘Inter- and intra-ethnic group relationships affect the allocation of resources stemming from development

interventions.’

Figure 23: Biplot displaying responses to the question ‘Inter- and intra-ethnic group relationships affect the allocation of resources stemming from development interventions’ and how they relate to specific origin categories respondents belong to.

Figure 24: Distribution of responses among full sample to the question ‘Inter- and intra-ethnic group relationships affect the allocation of resources stemming from development interventions out of sight of its initiators.

Figure 25: Distribution of responses among Burkinabé sample to the question ‘Inter- and intra-ethnic group relationships affect the allocation of resources stemming from development interventions out of sight of its initiators.’

Figure 26: Distribution of responses among non-Burkinabé sample to the question ‘Inter- and intra-ethnic group relationships affect the allocation of resources stemming from development interventions out of sight of its initiators.’

Figure 27: Biplot displaying responses to the question ‘Inter- and intra-ethnic group relationships affect the allocation of resources stemming from development interventions out of sight of its initiators’ and how they relate to specific origin categories respondents belong to.

Figure 28: Distribution of responses among full sample to the question ‘Inter- and intra-ethnic group relationships affect the selection of recipients of benefits stemming from development interventions’. Figure 29: Distribution of responses among Burkinabé sample to the question ‘Inter- and intra-ethnic group relationships affect the selection of recipients of benefits stemming from development

interventions’.

Figure 30: Distribution of responses among non-Burkinabé sample to the question ‘Inter- and intra-ethnic group relationships affect the selection of recipients of benefits stemming from development interventions’.

Figure 31: Biplot displaying responses to the question ‘Inter- and intra-ethnic group relationships affect the selection of recipients of benefits stemming from development interventions’ and how they relate to specific origin categories respondents belong to.

(6)

5

Chapter 1 - Introduction

Since 2012 the Sahel - Mali, Niger and Burkina Faso in particular - has seen a steep rise in insecurity resulting in inter-communal attacks, killings and confrontations between government forces, jihadists and self-defence groups or militias such as the Burkinabé Koglweogo. Longstanding tensions and conflict between livestock- and crop farmers, gold miners and mining companies, and wildlife rangers and poachers over land and natural resources intersect with inter-communal rivalries, jihadi attacks, local settling of scores, criminal activities and banditry in the border regions of Mali, Burkina Faso and Niger (Ammour, 2020). Recent reports state that violent attacks in the north have led to the displacement of over 800,000 people (Pedneault, 2020). Amnesty International’s most recent report (June 2020) treats the insecurity in Burkina Faso as a non-international armed conflict. However, as noted by many before, the security crisis in the area is more complex (Kitissou, 2020).

The decision of the Burkinabé parliament in January 2020 to back, fund and train local vigilantes, or self-defence militias, in response to the security crisis is a dangerous one, although in line with long-standing policies (Ammour, 2020). The self-defence militias, primarily made up of members of the Mossi ethnic group, have on many occasions been accused of attacking Peulh villages and murdering hundreds of inhabitants (Huong, 2020). It is unlikely that these volunteer forces are allowed entry to minoritarian ethnic groups supportive of armed groups making that “…volunteers may end up being mobilized on the basis of their ethnic affiliation”, which could possibly reignite or create new conflicts along ethnic lines (Pedneault, 2020). Human Rights Watch even stated how “…the practice of sub-contracting defence and security responsibilities to abusive militias has fed a vicious cycle of violence and deepened ethnic tensions” (Brown, 2020). Conversely, the Peulh ethnic group in the region and, to an increasing extent nationally, are perceived as the source of the unrest and as a group that is supportive of the jihadists.

The complexity of the security crisis also stems from social-, economic- and developmental disparities in the region which contribute to people’s grievances against the central [national] authorities in the Sahel. In Africa Renewal Harsch (2018) notes how development in Burkina Faso’s northern provinces lags 50 years behind the rest of the country and how it has been marginalised in terms of public services, state facilities and development projects. Although investments of the government in health care, education and job creation have been promised since 2017, this might (have) come too late. As a result, many people feel abandoned and distrust the government’s security forces. That distrust is deepened by the fact that only few officials come from the Peulh ethnic group that predominates in the north.

Understandably attention is drawn to the current security crisis in Burkina Faso and its neighbouring countries since it escalated and turned extremely violent, yet it is necessary to keep in mind the underlying complexity and tensions in terms of (inter)national development (interventions)

(11)

10 and ethnic group relationships at large. To what extent do development (interventions) and ethnic group relations affect one another on a day-to-day basis and how do inter- and intra-ethnic group relationships influence processes and institutions of development? While the current conflict mostly highlights the Peuhl, one might wonder what happens amongst the more than 60 other ethnic groups in the country, who coexist peacefully (Huong, 2020). Is the geographical diversity of ethnic and religious groups across the remainder of the country, resulting in a high-contact setting between various ethnic groups, combined with an interdependent lifestyle lived by many Burkinabé,

responsible for the calmness in the rest of the country (Hart, 2014, p.172)? Or does it have something to do with more equal distribution of various development interventions in these 'calm' zones?

Next to the above described developments in Burkina Faso, in responding to growing tensions closer to home, the Netherlands and other European countries decided to focus more on the Sahel (Government of the Netherlands, 2018) by supporting various development projects. Europe’s increasing presence in the Sahel, both militarily and in development aid (and trade) terms, as well as the efforts of the Burkinabé government to more equally distribute development efforts, calls for a broader understanding of the inter-and intra-ethnic group relationships in Burkina Faso and how these relationships are understood by development actors. Therefore, this research aims to answer the following research question:

How do inter- and intra-ethnic group dynamics and relationships in Burkina Faso affect local and international NGO development intervention(s) and distribution in terms of recipient selection and resourceallocation?

1.1. Social and academic relevance

A large volume of scientific literature on ethnicity, ethnic groups and their existing relationships and ‘development’ often portrays how ethnic relationships negatively influence development (Bates, 2000). Most of these studies are purely qualitative or quantitative in nature and are based on economic or demographic data, or interviews with community members of various localities. Most conclude that members of ethnic groups predominantly favour those from their own group, implying that the

prevalence of different ethnic groups is detrimental to development on a national-, regional- and local level.

From a pragmatic epistemological standpoint, and using a mixed-methods research design, this thesis will seek to gain a deeper understanding of how both inter- and intra-ethnic group

relationships affect developmental intervention(s) and distribution in terms of recipient selection and resource allocation from the perspective of development professionals themselves. This approach is not often used. Yet, approaching the topic from this angle appears to contribute to the understanding of this topic in this particular context. Also, it may serve as a tool for other development professionals to learn and improve their performances in Burkina Faso and elsewhere. This study, instead of

(12)

11 focusing on pre-existing datasets and accounts of those ‘enduring’ (under)development, focuses on the perceptions of the initiators and implementers of development projects through mixed research methods. Furthermore, the relevance of the present study is enhanced by the rather unique

circumstance that Burkina Faso, with a history of relative peace between ethnic groups, is at a point in time when (ethnic) tensions are on the rise and conflict has escalated in the northern and eastern regions of the country.

1.2. Problem statement

Development initiatives usually intend to improve the lives of society’s weakest. Yet overlooking, undervaluing, or in some cases even neglecting the importance of ethnic group relationships, lead to biases in the distribution of resources and selection of recipients for developmental interventions, away from these weakest members of society. Therefore, a deeper understanding and increased sensitivity to inter- and intra-ethnic group relationships might ensure a more just allocation of resources, to reach intended target audiences and increase the effectiveness of development interventions in general. In Burkina Faso’s north, a lack of development initiatives led to the

deprivation of the region and its inhabitants, thereby not only creating fertile ground for renewed and intensified ethnic tensions and conflict but also providing opportunity for radical ideology, be it through jihadists or militias, to gain footing among individuals and in communities.

In an era where international developmental aid and cooperation are still very present, those initiating developmental interventions should take ethnic group relationships into account to

counteract and avoid the exacerbation of inequalities existing between regions and/or specific (ethnic) groups. Failing to do so development interventionists, or their staff, run the risk of becoming

complicit to the further exacerbation of inequalities, tensions and even conflict.

1.3. Scope and focus

This study investigates the existence of links between ethnic group relationships and development in Burkina Faso. The focus of the study are the relevant perceptions of former and current development professionals in Burkina Faso. Attention will go out to if and how ethnic group relations are of effect and if and how development professionals cope(d) with them. The study was conducted in a six-month period during which 175 surveys were collected and 33 semi-structured interviews were conducted with development professionals.

1.4. Thesis outline

This thesis first discusses literature dealing with the concept of ethnicity and its connections to various forms of development interventions on various scales. Subsequently, the research design and

(13)

12 methodologies are outlined and ethical considerations and quality criteria discussed. This is followed by Burkina Faso's history, demographics and politics in brief. The subsequent three chapters analyse the data retrieved during the fieldwork. Chapter 5 profiles the research population in terms of age, nationality, ethnicity and their professional experiences in the country. Chapters 6 and 7 deal with inter- and intra-ethnic group relationships, and what they mean to development professionals with regard to their work. Chapter 8 contains perceptions of research participants regarding both inter- and intra-ethnic group relationships, with specific attention to the effect on resource allocation and recipient selection. The final chapter discusses the main findings of the study, reflects on the theory and methods applied and deals with its contributions to knowledge and implications. Suggestions for future research and policy recommendations conclude this study.

(14)

13

Chapter 2 - Literature Review

2.1. Ethnicity and Ethnic Groups: From Tribe to Ethnic Group

The concept of ethnicity is derived from the Greek word ethnos, used to indicate a small group of people spread over a region that could be bound by politics, custom and/or religion. The concepts of ethnicity and ethnic groups originate in the seventeenth and eighteenth century and have evolved from documenting structures and traits of ‘tribes’ to a focus on dynamics between ethnic groups (Jenkins, 2015, p.148). The notion of tribe, the precursor of ethnicity, came into existence as a result of the anthropologists’ ambitions to understand human diversity and was characterized by two interrelated principles. Firstly, human social life is based upon membership of bounded, distinct and culture-bearing ‘tribal’ corporate groups; and, secondly, the unity of humankind is understood as made up of patterns of cultural differentiation between various cultural groupings (Jenkins, 2008, p.17; Jenkins, 2015, p.148). Colonial governments valued the term because it distanced tribal society from ‘civilized society’, portrayed tribes as backward and primitive and thereby legitimized intervention by

‘civilized’ colonial governments (Jenkins, 2015, p.148; Gingrich, 2015, p.645).

Tribes were initially also conceptualized to be biologically self-perpetuating, its members sharing cultural values, the group being socially bound in communication- and interaction, and group members identifying themselves and being identified by others as belonging to a group (Jenkins, 2008, p.18). For identification of group membership, the boundaries of a group and differences between groups, ‘objective’ enumeration of cultural traits was the preferred method (Jenkins, 2008, p.54). By the 1960s the concept of tribe was largely replaced by the concept of ‘ethnic group’, although underlying assumptions about the boundedness of these groups and the ‘objective’ categorizability of traits of its members remained, until Frederik Barth initiated a paradigm shift in 1969 (Jenkins, 2008, p.18).

Barth, in his book titled Ethnic Groups and Boundaries. The Social Organization of Culture

Difference, criticised the rigid model of ethnicity and ethnic groups and showed that ethnicity and

membership of an ethnic group are, to varying degrees, strategically or tactically manipulable, capable of change and socially constructed (Jenkins, 2008, p.46). Barth (1969a) argues against the notion of ethnic groups as bounded-, stable entities and also against the notion that ethnic groups and their members could be identified by summing up cultural traits (Barth, 1969a, p.10-14).

Contrarily to anthropologists that formerly and, to a lesser extent, currently still delimit ethnicity and ethnic groups as based on corporate groups and “…the distribution of cultural and other ‘objective’ traits” (Eidheim, 1969, p.40) behavioural forms can also be a consequence of inter-ethnic dynamics and power relations (Eidheim, 1969, p.55-56). Similarly, Blom (1969) shows that

differences in cultural traits are not indicative of differences in ethnicity or responsible for the maintenance of boundaries between such groups. Rather, differences, such as in language, are

(15)

14 characteristics of social organization, possibly internally motivated and/or externally imposed,

between ethnic groups serving as a manner of alignment with specific group values (Blom, 1969, p.83-84).

Haaland (1969) furthermore held that an individual's ethnicity is not fixed and identity is malleable. According to Haaland (1969) change in ethnicity can be a consequence of a choice in lifestyle, nomadic or sedentary, and corresponding distribution of rights. He concludes that ethnicity, therefore, stems from what ethnicity members of both ethnic groups ascribe to those lifestyles

(Haaland, 1969, p.71) while also noting that those who changed ethnicity still possessed cultural traits from their former group making it hard to unambiguously classify them as part of one or the other (Haaland, 1969, p.68). Classification of a specific ethnic group, therefore, should take place based on self-ascription to categories by individuals (Haaland, 1969, p.70).

Summarizing the arguments by Barth (1969), Jenkins (2008) defines the social

constructionist model of ethnicity in four main characteristics, namely: “Ethnicity is a matter of

‘cultural’ differentiation (bearing in mind that identity is always a dialectic between similarity and difference); ethnicity is a matter of shared meanings – ‘culture’ – but it is also produced and

reproduced during interaction; ethnicity is no more fixed than the way of life of which it is part, or the situations in which it is produced and reproduced; and ethnicity is both collective and individual, externalized in social interaction and the categorization of others, and internalized in personal self-identification.” (Jenkins, 2008, p.169). It becomes apparent that, despite the socio-constructivist notion gaining popularity, studies concerned with ethnicity often still conceptualize ethnic groups as bounded groups based on ‘objective’, and quantifiable or measurable, traits such as culture, language or geographic origin.

2.2. Ethnic Favouritism, Ethnic Bias and Elite Capture

A large body of scientific literature describes ethnic diversity among a population in a territory as impairing development or counter-developmental, and is inspired by works of the economists Easterly and Alesina (Easterly, 1997; Alesina, Baqir & Easterly, 1999). Easterly and Levine (1997), based on a quantitative analysis of data related to African, educational and financial development and other economic indicators, research the influence of ethnic diversity on economic performance. Their underlying assumption is that ethnic diversity affected economic performance of African countries by encouraging adoption of growth-retarding policies, which foster rent-seeking behaviour, in turn hindering the forming of consensus on growth promoting public goods, public policies and political stability (Easterly & Levine, 1997, p.1205-1206). This assumption, following political economic models, states that the presence of various ethnic groups can lead ethnic groups to struggle among each other over (political-) power and resources which impedes the creation of political consensus for the adoption of (economic-)growth promoting policies.

(16)

15 Political clientelism can be understood as somewhat personalized relationships between actors or sets of actors possessing unequal levels of wealth, status and influence, based on conditional loyalties and involving mutually beneficial transactions (Lemarchand, 1972, p.69). In the context of various African countries, Lemarchand (1972) shows how political clientelism based on ethnicity can lead to populations electing or supporting political actors based on the ethnicity of this political actor in the expectation that this choice will reap various benefits. Franck and Rainer (2012), similarly, studied how the ethnicity of an African country’s leader affects primary education enrolment and infant mortality rates of different ethnic groups under their leadership. The authors show that “The effects of ethnic favouritism”, a phenomena whereby preferential treatment is given to a specific ethnic group, “…are quite large and widespread” (Franck & Rainer, 2012, p.294).

Ilorah (2009), in turn, shows how ethnic bias is a product of personalised ties between members of a specific group and, in some cases, between groups. Just like personalised ties, ethnic bias has features of favouritism. Ilorah (2009) explains how in societies that “…practice ethnic favouritism, the behaviour of the leadership towards others varies depending on whether those others are from the in-group, with whom affinities are shared and who therefore should be treated fairly, or outsiders, including people from other ethnic groups, who must be politically and economically marginalised” (Ilorah, 2009, p.3). The author argues that the root causes of ethnic bias and favouritism in Africa can be traced back to colonial days and stem from to the redrawing of boundaries and re-administering applied by colonists. The colonists were mostly interested in natural resources, engaged in predatory activities regardless of the indigenous peoples' religious beliefs, cultures and ethnicities which, in turn, led to the emergence of divided societies (Ilorah, 2009, p.689). De Luca, Holder, Raschky and Valsecchi (2016), conversely, showed that ethnic favouritism is not only an African phenomena, but “...a global axiom of politics that exists in both rich and poor countries” (De Luca et al., 2016, p.2).

A different dynamic involving ethnic groups and development initiatives can be found in the works by Platteau and Abraham (2012) and Arnall, Thomas, Twyman and Liverman (2013). Platteau and Abraham (2012) state that for participatory development initiatives to work, communities must first evolve in terms of the adoption of clear rules for decision-making and the application of majority voting, to avoid undue appropriation of resources by elites, or elite capture (Platteau & Abraham, 2012, p.104). The authors explain that ‘elites’ exist in various forms, such as village chiefs or

unemployed former politicians and state employees, and that elite capture takes place between ethnic groups but also within, for example, families. According to the authors it is due to a logic of

patronage, residing with elites, that resources are unduly appropriated and channelled through communities or groups towards elites (Platteau & Abraham, 2012, p.128).

Similarly, Arnall et al. (2013) distinguish between benevolent elite capture, whereby elites are motivated by a genuine interest in communitarian well-being, and malevolent elite capture, whereby resources are primarily channelled towards those in favour of this elite. They state that elites “…have

(17)

16 varied and complex motivations that cannot be reduced to the ‘timeless’, static image” (Arnall et al. 2013, p.326), meaning that they signal that communal norms and values and other social dynamics also influence if and how elite capture takes place. This begs the question: is culture completely irrelevant in trying to understand ethnic groups, their relationships and its effects on development interventions?

2.3. Culture and Equilibria

The socio-constructivist notion of ethnicity, as put forth by Barth (1969), does not perceive culture, such as religion, tradition and custom, as irrelevant but rather as an element of ethnicity (Jenkins, 2008, p.111). For example, Jenkins (2008) explains that it is possible that two groups seemingly share many or all cultural traits, yet consider themselves or are considered as different ethnic groups.

In presenting a model for decoding effects of cultural differences in international business, Meyer (2014) identifies eight cultural scales at play in inter-cultural interactions. She states that cultural patterns of behaviour and belief impact perceptions, cognitions and actions. Awareness of cultural differences in communicative mannerisms of evaluation, persuasion and disagreement improves effectiveness of inter-cultural cooperation (Meyer, 2014, p.13-16).

An example where these cultural dynamics are at play in relation to NGO development initiatives can be found in the article by Willems (2009), concerning the application of participatory methods in clientelist societies. The cultural meaning of ‘participation’ is questioned and the author states that “…discourses transposed into culturally different contexts are likely to lead to unexpected results due to differences in cultural logic” (Willems, 2009, p.9). Willems (2009) states that where patron-client relationships exist between various groups, even rigorously planned initiatives can inadvertently lead to elite capture due to miscommunications stemming from misunderstanding the cultural logics of the communities in question. But does this mean that the existence of ethnic groups and how they relate to each other always leads to negative outcomes in relation to development? Fearon and Laitin (1996) argue that although literature concerning ethnicity mainly highlights negative aspects, peaceful and cooperative relations between ethnic groups do often occur. They argue that in contexts where state authority is weak “…non-state institutional mechanisms may often arise to mitigate problems of opportunism in interactions between individuals from different ethnic groups” (Fearon & Laitin, 1996, p.716). The authors describe two mechanisms, relating to both intra- and inter-ethnic group relations, by which such cooperation comes into existence. Spiral equilibria, the first mechanism, stems from mutual fear that disputes between two individuals of different ethnic groups might lead to broader conflict between the two groups, in turn, leading to interethnic

cooperation. In-group policing equilibria develop from the expectancy of the members of one group that transgressions from a member of another group will be sanctioned by his/her fellow group members. The former group, therefore, ignores these transgressions due to equilibrium being

(18)

17 maintained by the other group (Fearon & Laitin, 1996, p.715).

2.4. Research Question

To sum up, a shift in conceptualizations of ethnicity and ethnic groups is discernible, away from the identification and categorisation of ‘objective’ traits of bounded and rigid ethnic groups, towards a focus on the dynamic nature and social constructedness of these groups. As will become clear from the present study, in Burkina Faso (and many other places) the identification of an individual’s ethnicity or the delineation of ethnic groups is not only based on shared cultural characteristics such as language or other cultural characteristics. Therefore, for the purpose of this study the social constructivist conceptualization of ethnicity was selected as the best theoretical underpinning to understand the role of inter- and intra-ethnic group relationships in development in Burkina Faso. Moreover, it became apparent that a large body of literature perceives ethnic diversity as causing poor economic development whereby underlying reasoning revolves around the assumption that individuals will generally favour in-group over others leading to favouritism, ethnic bias and elite capture. Next to that, culture, especially in communication, although conceptualized as part of

ethnicity, is a relevant aspect that can affect development initiatives and their outcomes. Furthermore, although a large body of literature perceives ethnic diversity as a burden on development (Fearon & Laitin, 1996), in many cases equilibrium and cooperation between various ethnic groups exist. While the above literature review brings greater insight into the concepts of ethnicity and the effects of inter-ethnic group interactions on development interventions, it also makes clear that little is known about intra-ethnic group dynamics and to what extent development practitioners take these dynamics into account. To fill this gap, the present research aims to answer the following question:

How do inter- and intra- ethnic group dynamics and relationships in Burkina Faso affect local and international NGO development intervention(s) and distribution in terms of recipient selection and resourceallocation?

2.5. Sub-questions

To answer the main Research Question (RQ), several sub-questions were formulated. Sub-question one will provide contextual data that will also guide the way in which sub-questions two and three are answered. The three sub-questions are formulated as follows:

• SQ1: What are the most prominently described inter- and intra-ethnic group relations and dynamics in Burkina Faso and how do these take shape according to development

(19)

18

• SQ2: How and to what extent are and were ethnic group relationships taken into account by development professionals formerly and/or currently active in Burkina Faso?

• SQ3: To what degree are dynamics of ethnic favouritism, elite capture or a combination of both of these phenomena stemming from ethnic group relationships described by former and current development professionals and how do they vary among them?

The operationalisation of the concepts of ethnicity, ethnic-group relations and distribution of benefits stemming from development interventions can be viewed in table A1 in Appendix A.

2.6. Operationalisation and conceptual model

The operationalisation scheme used in this study is visible in table 1. Following Barth (1969) in putting an emphasis on self-identification and Jenkins (2009) in his description of ethnicity as consisting of, among other things, common ancestry, kinship, economic way of life and lifestyle (Jenkins, 2009, p.10-11) ethnicity is conceptualized as consisting of four dimensions. The dimensions, visible in the operationalisation scheme in table 1 and and conceptual model in figure B1, consist of geographic area of ancestry, clan-/lifestyle affiliation, family affiliation and individual ethnic identification. Consequently, every dimension has one corresponding variable regarding the, for example, self-declared geographic area of ancestry or ethnic group affiliation. These four variables, in turn, will be measured through four corresponding questions which inquire into self-ascription to an ethnic group.

The concept of ethnic groups relations consists of intra-ethnic group relationships and

inter-ethnic group relationships which can both affect the distribution of benefits stemming from a

development intervention (Fearon & Laitin, 1996). Both intra- and inter-ethnic group relationships are operationalized as consisting of three variables: affectionate relationships, neutral relationships and hostile relationships. It has to be noted that these types of ethnic group relationships are not mutually exclusive nor fixed, but vary across time. For measurement of these variables research participants were asked to typify intra- and/or inter-ethnic group relationships when they came up during interviews.

Distribution of benefits stemming from development intervention(s) is operationalized in three dimensions: intended distribution, mixed distribution and unintended distribution. These

dimensions in turn consist of Ethnic Favouritism, Benevolent Elite Capture, Malevolent Elite Capture and no influence of ethnic groups. Similar to the variables concerning ethnic group relationships, the variables related to distribution of benefits stemming from developmental interventions are not mutually exclusive and, depending on the nature and goal of the development intervention, might change over time. As can be noticed in table A1, the variables for all three dimensions of distribution of resources from development interventions are the same, following the reasoning that if

(20)

19 (pre)knowledge about existing intra- and inter-ethnic group relationships exists, it is imaginable that employees of both local and international NGOs are able to anticipate on possible ethnic favouritism or elite capture taking place which, in turn, makes that these forms of distribution may fall in the category of intended distribution.

A conceptual model showing how the above operationalized concepts stand in relation to each other is shown in figure A3 in Appendix A. The model is intended to be read from left to right. Beneath the conceptual model clarifications of its components are provided.

2.7. Hypotheses

On the basis of the Research Question and three sub-questions the following hypotheses are formulated:

H1: In the context of Burkina Faso, dominant ethnic groups, in terms of demographic size,

economic wealth and social standing, appropriated and still appropriate a disproportionately large share of resources stemming from local and international development interventions than defensible and/or intended by the providers of these resources.

H2: Within ethnic groups, dominant figures or groups, in terms of wealth, political or

traditional power and status, influence allocation of resources and participation in developmental initiatives to the benefit of those close to them, in terms of family or kin relationships.

(21)

20

Chapter 3 - Methodology and Methods

This chapter starts with an illustration of the paradigm and the ontological and epistemological stance in which the research is nested, as well as the research design applied. This is followed by a

description of the quantitative and qualitative research methods chosen, a discussion of the ethical considerations, the researcher’s positionality and axiology and the implications of the chosen research design, methods and quality criteria.

3.1. Ontology and epistemology

As put forward already in the previous chapter on literature, the aim of this study is to gain greater understanding of ethnic group relationships in development, i.e., the unit of analysis, through investigation of the views of current and former development professionals involved with Burkina Faso, i.e., the unit of observation. This is done through inquiry into the perceptions of former and current development professionals active in or on Burkina Faso.

A fixed-convergent mixed methods design was adopted, focused on data collection of the same topic through both qualitative and quantitative methods, where the results are subsequently merged (Morgan, 2007, p.52, Creswell & Plano Clark, 2017, p.70-71). This mixed methods design also allows for a more in-depth understanding of perceptions contributed by research participants than would be possible if only one of the two methods was applied. The conducted research is therefore nested in the pragmatic paradigm, which ontologically frees the researcher from methodological constraints of positivism and constructivism (Feilzer, 2010, p.8). Emphasis was put on abductive reasoning, intersubjectivity and transferability. Abductive reasoning entails moving back and forth from induction to deduction, where observations are converted into theories that, through action, are then assessed (Morgan, 2007, p.71). The pragmatic emphasis of intersubjectivity relates to how the researcher is never fully objective or subjective. The emphasis is on shared understanding with the respondents and interviewees and captures duality (Morgan, 2007, p.72). The transferability in the pragmatic approach relates directly to how the finding of the research can be applied in a different context (Morgan, 2007, p.72). The set-up of the research enabled for the conversion of observations, through analysing overall survey responses and impressions gained during interviews, into theory or adaptation of prior theoretical ideas which, in turn, were added to interview questions.

3.2. Research methods and analysis

The applied methods in this research are of cross-sectional design with case study elements (Bryman, 2012, p.69). The applied qualitative methods are desk research from September 2019 to May 2020 and the conduction of semi-structured interviews during fieldwork in Burkina Faso, as well as upon

(22)

21 return in the Netherlands. These interviews were held between September 2019 and February 2020. The applied quantitative method was an online- and face-to-face survey from September 2019 to May 2020.

3.3. Quantitative research method

The quantitative online survey aimed at understanding how former and current development professionals perceived different types of ethnic group relationships in their work. It allowed for comparison of responses on the basis of characteristics such as age, sex, nationality and ethnicity and the developmental sector they were active in. It also gave insight into to what extent these relations affect development interventions, to what extent knowledge of these relations is deemed valuable and to what extent these relations were or are taken into account in their work.

The survey was disseminated as a self-completion online survey and through the conduction of surveys by the researcher(s) in person. Both surveys were created through the online survey tool Qualtrics, in both French and English. The online survey was printed for the in-person surveys. From the research population (N) a convenience sample (Bryman, 2012, p.202) was drawn since, although some of the contours of the sampling frame are clear (Bryman, 2012, p.187), an exhaustive listing of the research population does not exist. This has implications for the quality of the research, discussed later on. Respondents were identified through institutions where they are or were active, the

researcher’s personal network, LinkedIn and snowball sampling (Bryman, 2012, p.203). For

dissemination of the online survey, email addresses were acquired through similar means and through various national- and international NGO-confederations. Figure B1 in Appendix B shows a rendering of the survey.

3.4. Quantitative analysis

The methods applied for analysis of the quantitative data consist of a general review of retrieved data based on cross-tabulations and corresponding visualisations, the application of Correspondence Analysis (CA) and, if found, further analysis of found associations using the chi-square and Cramer’s V statistical measures.

Correspondence analysis, from here on referred to as CA, is an exploratory statistical method often used in marketing research, which enables researchers to investigate and visualize possible patterns and associations between categorical variables in cross-tabulations. With CA, and chi-square tests, the null hypothesis is that the two categorical variables under investigation are independent, while the alternative hypothesis says that the variables are dependent. CA visualisations use the chi-square statistic, a non-parametric test aimed at uncovering independence of or association between categorical variables, to create a ‘bi plot’. The bi plot contains points representing the categories of variables that are to be compared, where the distances between these points, based on the Euclidean

(23)

22 distance between chi-square scores, gives a visual representation of the extent to which categories of variables are independent from each other (Doey & Kurta, 2011, p.5-6, Greenacre, 2009,p.2-3). On the bi plot, points that are displayed close together have similar response profiles, while points further apart have more differing profiles (Doey & Kurta, 2011, p.12). In other words, the distance between the points indicates how similar or dissimilar categories are with respect to one another.

This method was chosen because the survey mainly consists of variables measured on an ordinal or nominal scale. Moreover, this research aims to explore and gain greater insights into if and how responses of former and current development professionals differ along the lines of the

categorical variables included in the survey. A benefit of using CA for this research is that it help to discern possibly existing patterns in data that would be hard to find through pairwise comparison of variables, such as in cross tables (Doey & Kurta, 2011, p.7). Moreover, CA does not require the data to meet pre-set assumptions, such as being normally distributed, except for data being discrete and that it fits in a rectangular data matrix without negative entries (Hoffman & Franke, 1986, p.213).

To test the significance of the models produced through CA a significance level of P < 0.05 is maintained whereby, keeping in mind the scope of this study, only significant models are discussed. Non-significant models are included for reference in a separate section of the appendix. In significant models some basic characteristics of the model are analysed through a discussion of the

corresponding bi plot. Lastly, while CA is a useful method for the visualisation of similarity and difference between various categories, as noted by Doey & Kurta (2011), the method does not allow for precise conclusions. However, due to the exploratory aim of this study and issues with regard to sample bias described later on in the chapter, this is not perceived as a problem. It is believed that a general discussion of retrieved responses and CA, combined with qualitative interviews, provides enough depth in the retrieved data. This discussion can be found in chapter 9.

3.5. Qualitative research methods and data analysis

The applied qualitative research methods consists of semi-structured interviews with former and current development professionals and, to a lesser extent, desk-research of textual documents acquired from the University of Amsterdam library, Google Scholar and other academic platforms as well as output from social media and printed and digital news platforms.

The selection of relevant actors for the semi-structured interviews was done by purposive sampling, consisting of sampling of context and sampling of participants (Bryman, 2012, p.416-417). The overall population (N), consists of current and former local and international NGO workers. No clear outline of the internal structure of this population was known to the researcher. Interviewees were approached through the organisation(s) where they are active, via email, telephone or in person and through the researcher’s personal network. Snowballing (Bryman, 2012, p.424) was applied to contact additional interviewees. The interviews followed a pre-assembled item-list, shown in figure

(24)

23 B2 in Appendix B, and were audio-recorded. After, all audio-recordings were transcribed,

anonymised and analysed.

The analysis is based on categorisation and subsequent refinement, using a method of open coding, re-coding, compilation and comparison, as described by Burnard (1991). These procedures were carried out using the software program Atlas.ti. To avoid research bias (Burnard, 1991, p.463), the coding and re-coding process was reviewed and discussed with various interviewees, peers and supervisors for respondent validation. The compilations of excerpts were combined with the earlier described survey data and used in the write-up of this study.

The desk-research is based on textual documents acquired from the University of Amsterdam library, Google Scholar, the CIA Factbook and social media and news websites. When searching for articles in scientific and policy documents the following search terms – and combinations thereof - were used: Ethnic groups, Ethnic group relationships, Ethnicity, Development, Kinship, groupe

ethnique, Inter-ethnic, Intra-ethnic, Resource allocation, Development intervention, Participant

selection, Burkina Faso, parenté plaisenterie. On social media, attention mostly went out to posts and pages dealing with ethnicity and ethnic group relations in the Burkinabé context. The aim of de desk-research was to better understand the current socio-political context in Burkina Faso and to gain a better overview of how the subject of ethnicity and development are perceived nationally and internationally by various institutions and actors.

3.6. Ethical considerations

Throughout this research, the Amsterdam Institute for Social Science Research (2017) ethical procedures were taken into account. Where it concerned the acquisition of research participants and subsequent data collection the researcher ensured that participants provided informed consent before they were included in the research (Bryman, 2012, p.138). On all occasions, the researcher clearly stated the purpose of the research and, if required, provided additional information.

Prior to the interview, it was explained verbally and in writing that all retrieved data would be anonymised to assure participants that they could speak freely without repercussions and to make sure participants were protected from possible unforeseen consequences. All audio-recordings were deleted after transcription and transcripts and all other data was saved on a secured hard disk. To ensure anonymity of participants, all retrieved data and findings were only discussed during respondent validation and shared with the thesis supervisor after full anonymisation.

3.7. Researcher positionality and axiology

Regarding the researcher’s positionality and axiology, several things need to be noted. Researcher positionality concerns the position a researcher chooses to adopt in a study (Holmes, 2014, p.2), as it might influence how research participants react to and formulate answers (Bryman, 2012, p.233). The

(25)

24 researcher, of Dutch nationality, lived in Burkina Faso for 9 years and has mixed Dutch-Burkinabé roots. These characteristics have multiple implications. Firstly, the researcher already possessed substantial knowledge about the workings and some of the effects of intra- and inter-ethnic group relationships in Burkinabé society. Due to having one Burkinabé and one Dutch parent, having in mind the earlier described flexibility of the notion ‘ethnicity’, the researcher could be perceived by others as a member of a local ethnic group. In turn, these characteristics might affect how

interviewees and, to a lesser extent, survey respondents react to and interact with the researcher, possibly leading to participant bias. To cope with this, the researcher, unless specifically asked, avoided revealing his ‘membership’ of an ethnic group. This procedure was deemed to be the best solution, because it is believed that being perceived as an outsider, at least from the present ethnic groups, reduces the likelihood of participants giving biased or socially desirable answers.

Concerning the axiology of the researcher the following considerations were made.

Working from a pragmatic approach paradigm one must be aware that one's values and political views are always part of our person and the actions we undertake (Morgan, 2007). It therefore seems right to highlight some of the values and political views the researcher holds. The researcher is distrustful as regards certain aims and practices of large, international NGOs present in Burkina Faso. This scepticism is based on the idea that these NGOs, in some cases, are guided by national agendas of their countries of origin which might lead to these NGOs, possibly intentionally, overlooking or neglecting certain aspects relevant to the context wherein they operate. Next to this scepticism on the international development sector, the researcher also has certain social perceptions on inter- and intra-ethnic group relationships in Burkina Faso. He can identify – based on the circumstances as described above – as Samo, which, in a way, has shaped his thinking and perception of ethnic group

relationships in Burkina Faso.

3.8. Implications of research design(s) and research methods

The outlined research design and applied methods have implications for various quality measures of the conducted research. To abide by quality criteria set by Creswell and Plano Clark (2018, p.70) regarding the conduction of mixed methods research, specific attention went to the ‘proper’ merging of the obtained qualitative and quantitative data for further interpretation. To ensure inference quality and the quality of conclusions made on the basis of research findings (Plano Clark & Ivankova, 2016), and the internal reliability and validity (Bryman, 2012, p.390), throughout the research several

respondent validation sessions were held in person, in written form and over Skype.

Subsequently, the selection of a cross-sectional research design has several implications concerning the replicability (or external reliability), and validity of the research. A high level of replicability was strived for through clear description of all research procedures. However, the level of replicability varies between the qualitative and quantitative research methods applied, as will be

(26)

25 discussed further on. The application of a cross-sectional design negatively impacts the overall levels of internal validity because, as explained by Bryman (2012), cross-sectional designs produce

associations instead of findings on which causal inferences can be made.

It can be stated that the transferability, or generalizability, of the overall research findings is low due to the specificity of the researched subject matter (Bryman, 2012, p.47). However, it is believed that the general conclusions are relevant in all contexts where ‘development’ is pursued. Specific parts of the findings and conclusions are also applicable to the Sahel region as a whole. The limitations in transferability or generalizability are to some extent accepted because the overall aim of this thesis was not only to inquire into how, in the views of development professionals, ethnic group relations do or do not affect development interventions but also to explore if and how perceptions on this matter vary between development professionals (formerly) involved in the country.

To increase the confirmability of the overall research and its findings (Bryman, 2012, p.421) the researcher refrained from corrupting the obtained data with personal presumptions or assumptions while also making sure not to influence the opinions of fellow research participants. As explained in the previous section, this was pursued by the researcher being aware of his own axiology and by only sharing personal views on the subject after an interview or survey was completed.

3.9. Quality criteria

Regarding quality criteria with reference to the qualitative and quantitative research methods, the following has to be noted. Firstly, the acquired data, qualitative as well as quantitative, was prone to sampling bias due to the sampling frame being unclear, which affected the sampling methods applied in this study and its overall external validity and generalizability. As noted before, no clear outline of the research population or accurate sampling frame thereof was accessible to the researcher. This substantially increased the chance of acquiring a non-representative sample and also made the use of a probability sample out of the question (Bryman, 2012, p.59-61, p.416-417). Consequently, due to the use of non-probability sampling methods consisting of purposive and convenience sampling,

combined with the use of snowballing for participant acquisition, it is possible that several obvious and possibly less obvious sources of sampling bias (might) have occurred in the form of

overrepresentation of certain groups in the survey-data and among the interviewees (Bryman, 2012, p.188). Overall, this negatively affects the external validity or generalizability of the acquired

quantitative data (Bryman, 2012, p.171-172) and weakens the external validity or transferability of the qualitative data (Bryman, 2012, p.390).

These issues have been dealt with in three ways. Firstly, on several occasions during the research participants were asked to describe who, in terms of nationalities, types of development organisations and corresponding development sectors, in their memory has been active in

(27)

26 sample bias. Consequently, the reader is made aware of the possibility of sample bias and how this bias takes shape. The possibility of and/or identified biases were kept in mind during the write-up of the findings and conclusions of the thesis and the reader is again alerted to them. In relation to the conducted semi-structured interviews and the related measures of internal reliability or dependability (Bryman, 2012, p.390) an auditing approach was adopted through extensively making notes during interviews. Interview transcripts and data-analysis decisions were kept and discussed repeatedly with peers and supervisor. The conduction of interviews continued for as long as possible to approach theoretical saturation, within the given time span (Bryman, 2012, p.421).

(28)

27

Chapter 4 - Research Context

This chapter first briefly outlines general characteristics of Burkina Faso, its geography, history and politics. The subsequent section provides characteristics of ethnic groups, with special attention to the Mossi, Gourounsi, Bobo and Peulh (Fulani). The chapter concludes with a review of developmental paradigms at work which shaped the ‘development’ context of Burkina Faso today in which the study was conducted.

4.1. Burkina Faso: geography, history and politics

Burkina Faso is a landlocked country in West Africa and a critical regional transit hub as it borders Mali, Niger, Benin, Togo, Ghana, and Côte d’Ivoire (G20 Compact with Africa, N.D.). The country has a tropical savanna climate, with dry winters and wet summers, and has a surface area of 274,200 km² of which about half is used for agriculture. The majority of its population is found in the southern parts (CIA, 2019).

Burkina Faso has a history of colonial rule by the French, dating back to 1896 when it was made part of French West Africa. Until 1984 Burkina Faso was known as Upper Volta. Upper Volta gained its independence from France in 1960. After its first president in 1966, young intellectuals started to perceive the army as an institution to discipline corruption, counterbalance the influence of traditional chiefs and push towards modernization (Harsch, 2014). In those years the young army leader Thomas Sankara gained support among student - and labour organizations and with their help came to power in August 1983. Then he embarked on a revolutionary course (Harsch, 2014). He changed the name of the country to Burkina Faso, meaning the land of the honourable men (or people) in which the citizens are called Burkinabé (‘people’ or ‘children of’). The name is a

combination of Mooré (language of the Mossi), Fulbé (language of the Peulh) and Dioula (language widely spoken in the western regions) and was meant to unify the Burkinabé as one nation. Despite his popularity, his revolution only lasted for three years. Sankara was assassinated in October 1987 and succeeded by an alleged complicit in the assassination, Blaise Compaoré, who ruled the country for 27 years. In October 2014, Compaoré was ousted in a popular uprising as a reaction to his attempt to change the constitution (Frère & Englebert, 2015, p.295-296). In the run up to the elections of November 2015, Compaoré’s former military right-hand and commander of his presidential guard, Gilbert Diendéré, attempted a coup d’etat (Zeilig, 2017, p.155-156). In reaction Burkinabé massively hit the streets and urged for democratic elections. On November 29th 2015, elections could be held and Roch Kaboré of the new MPP party took power (Dragstra, 2017, p.91). New elections are planned for November 2020.

Ethnicity matters: Development in Burkina Faso between Symbiosis, Favouritism and Exclusion