Horizontal Inequalities and Food Riots in Africa.

In recent years, protest events in a range of African countries have attracted widespread interest and public debate. The sharp increase in world food prices in 2007 and 2008, for example, was associated with the occurrence of a wave of food riots on the continent. In a number of countries, including Burkina Faso, Cameroon, Côte d’Ivoire, Egypt, Morocco, and Tunisia, protests erupted against high costs of living and frequently escalated into violence. These actions led to serious concern for political stability in several of these countries. The Arab Spring, which started at the end of 2010 in Tunisia was constituted by large-scale protest movements against authoritarian rulers. Multiple countries experienced opposition in the streets with events in Egypt, Libya, and Tunisia capturing most attention because of the violence which erupted as a result of protests and repression, and their far-ranging political consequences. In Sub-Saharan Africa as well, strong opposition protests recently emerged against political rulers. In Madagascar (2009) and Niger (2010) street demonstrations resulted in military interventions and the ousting of incumbent presidents. In Senegal (2012), protests prevented president Wade from changing electoral rules in his favour and reduced his chances of winning a third term in office. When long-time president Compaoré of Burkina Faso attempted to amend the constitution in 2014 in order to continue his rule, opposition revolt led him to eventually flee the country.

A key question that emerges when witnessing such protest movements concerns their causes or drivers. For many commentators, these African protests can be explained by economic discontent among the local population. During the increase in world food prices, for example, Jacques Diouf, then Director-General of the Food and Agricultural Organization, stated that “Naturally people won't be sitting dying of starvation, they will react.” (in: Pomeroy, 2008). Such arguments correspond strongly to neo-Malthusian views which argue that resource scarcity and resulting poverty fuel animosity and conflict (e.g. Kaplan, 1994). Indeed, reduced access to food, unemployment, and hunger are commonly argued to drive people into the streets and engage in protests and violence.

In recent years, the role of economic stressors in fuelling conflict has also led to an increasing body of academic research. The scientific debate on how resource scarcity can lead to conflict mainly goes back to the work of Homer-Dixon (1994, 1999), who held that different environmental stress factors such as land degradation, freshwater depletion, and population growth can lead to economic setbacks and increasing competition between groups. Grievances resulting from resource scarcity would in turn form important motivations for engaging in collective action. Several studies have focused on how different resource scarcity indicators are related to violent armed conflicts (e.g. Hauge and Ellingsen, 1998; Kahl, 2006).  Yet researchers have also turned their attention to how different low-scale forms of conflicts, or ‘social disorder’ events are driven by resource scarcity (e.g. Hendrix and Salehyan, 2012; Urdal and Hoelscher, 2012). Widespread concern with African food riots, for example, has led to a substantial research agenda on the effect of food prices on protest and riot occurrence (e.g. Weinberg and Bakker, 2015). Both in public debate and the resource scarcity literature, economic shocks and resulting grievances are hence seen as principal explanations for protest mobilization in Africa.

In this thesis, however, I engage critically with the view that resource scarcity is the main cause of protest mobilization and conflict in Africa. While drawing on the social movement literature , I argue that the theoretical and empirical focus on resource scarcity as a cause of conflict is too limiting to understand mobilization dynamics in Africa. The resource scarcity literature primarily relies on the idea that grievances caused by poverty and inequality explain conflict emergence. Yet while motivations to engage in protests form an important explanation for mobilization, the social movement literature demonstrates that motivations alone are often insufficient to explain participation in street demonstrations (e.g. McAdam, McCarthy, and Zald, 1999). Grievances often have to be actively mobilized and organized by group leaders for instance while government reactions to challengers, including repression, can inhibit protest actions. Although social and political contextual factors are recognized within the resource scarcity literature, they are rarely addressed in depth. By contrast, additional socio-political explanatory paradigms for mobilization besides motivations have received substantial attention in the social movement literature. In this thesis, I argue that these paradigms can also be applied to African contexts, and that they are indeed needed to provide crucial insights into the complexity of protest mobilization in Africa.

Furthermore, the thesis not only draws theoretically from social movement studies, but also methodologically. Within the social movement literature a number of methodological approaches have been developed and used to investigate a range of research questions pertaining to the study of protest movements. Some scholars focus on individuals and the characteristics that make them more likely to participate in protests, for example, while others focus on the political systems in which protest events are more likely to occur. Too little focus has been directed to methodological insights from social movement studies in the study of African protest movements, however. The first core critique that emerges is that many resource scarcity studies have little methodological focus on the individual level, and that this is highly paradoxical given their theoretical focus on individual motivations. The second critique is that many studies rely heavily on news reports to approach African protests, while neglecting misrepresentations and errors arising from these data. Both critiques are substantiated empirically in the thesis.

The thesis itself consists of a collection of articles, and consequently each following chapter can be read as a separate contribution. The red line is formed by the critical engagement with the resource scarcity literature and each chapter contributes to the theoretical and methodological arguments outlined above. Hence, the two main research questions guiding the thesis as a whole are the following:

What are the main drivers of African protest mobilizations, and to what extent can social movement theory be applied to explain these protests?

How can insights from social movement studies improve data reliability and validity in the study of African protests and mobilization dynamics? 

Theoretically, the thesis argues that resource scarcity theory as such does not sufficiently improve our understanding of African mobilizations, and that perspectives from the social movement literature are needed to improve our insights. Methodologically, the study of protests in Africa is argued to suffer from crucial shortcomings which jeopardize research findings. A number of these shortcomings have already been raised and addressed earlier in the social movement literature, and can guide further research in African settings. The combination of theoretical and methodological insights in particular demonstrates that there is too little interaction between research on protests in Africa and Western-focused social movement studies, which creates important lacunae in our understanding of protest mobilization on the continent. These arguments are addressed and empirically substantiated in the individual chapters of the thesis.

Chapter 2 ‘Food Price Rises and Political Instability: Problematizing a Complex Relationship’ focuses on the occurrence of food riots following the rise in international food prices in 2007. By means of a critical review of the literature on these particular events, the paper highlights major theoretical and methodological gaps in the literature on food price rises and conflict. Specifically, I argue that the food riot literature suffers from a number of crucial shortcomings, focusing on three issues in particular: a lack of precision in the use of concepts such as conflict or political instability, uncritical data collection based on media sources, and presupposed and understudied economic as well as political causal mechanisms. 

The paper first stresses the need for conceptual distinctions between different types of conflict discussed in the food riot literature. It highlights in particular the distinctions between peaceful protests, riots, and civil war, arguing that the underlying mechanisms leading to each form of conflict can be quite different. Moreover, the occurrence of each of these conflict types cannot simply be equated with political instability. Furthermore, the food riot concept in particular is scrutinized and argued to be a container term for both riots, in which protesters act violently, and (violently repressed) demonstrations. This argument has a crucial link with our empirical understanding of food riots as the term seemingly implies the rise of hungry and violent mobs, whereas in reality most violence seems to have been perpetrated by governments. Indeed, many researchers have taken both protests and riots together in the construction of food riot lists.

The second topic addressed in the paper regards the use of news reports to collect empirical data on conflict events. When reviewing data on food riots more closely, it emerges for instance that many researchers make use of different lists in which different events are either included or excluded. Furthermore, besides the use of violence in these events, the protest issue is unclear as well. It seems that many events have simply been framed as food riots by media accounts even when case descriptions do not include clear links to food concerns. This categorization has been uncritically adopted by several researchers.

The third issue highlighted in the paper consists of understudied and presupposed causal mechanisms. Much of the literature and commentaries on food riots argue that the link between food price rises and conflict runs through food insecurity, implying that the hungry poor are the ones participating in conflict. Such a claim requires more detailed, micro-level evidence, however, and cannot be verified by relating macro-level variables (e.g. food imports) to food riot occurrences. Furthermore, I argue that political and institutional factors play important roles as well in conflict processes, and cannot be replaced by purely economic accounts. Mobilization structures and political opportunities can be key in explaining protest mobilization.

Chapter 3 ‘Staging a “Revolution”: The 2011-2012 Electoral Protests in Senegal’ engages further with the theoretical limitations of resource scarcity accounts of conflict. The case of Senegal is particularly interesting because the protests, directed at then-President Abdoulaye Wade’s attempts to change electoral rules in his favour, have been predominantly explained from the viewpoint of grievances at the population level, in public debate as well as the academic literature. Economic grievances, in particular caused by rising food prices and youth unemployment, were seen as major drivers of mobilization. Some scholars also emphasized political grievances, stemming from democratic aspirations. The case of the Senegalese electoral protests could be regarded as a ‘positive’ case of the scarcity-conflict thesis, in which economic downturn and inflation run together with mobilization. Nonetheless, in this paper I argue that a grievance perspective limits our understanding of the protests and their political implications. Theoretically, I use a resource mobilization perspective to trace how the protest movement organized, gained success, and eventually demised during the electoral process, which ended with the victory of Macky Sall in the second round. The approach reveals key insights on the role of opposition parties and political interests in the mobilization, and hence empirically substantiates the value of using different theoretical lenses to look at protest mobilization in.

The study relies on qualitative interviews with key members of the main protest movement against President Wade, as well as internal movement documents. By tracing the evolution of the protest movement, the study reveals the importance of coalition-building between civil society organizations such as human rights and development NGOs, youth movements, and community-based organizations. These organizations also connected to political parties, which played major roles in the organization of the movement and mobilization processes. Opposition parties were major members of the protest movement, contributed the bulk of the resources (finances, vehicles etc.) for the organization of protests, and were often responsible for mobilizing individuals at the grassroots level. The analysis also reveals that protesters strategically made use of violence, in the form of vandalism or tyre-burning, as a resource to gain international media attention and interest for events ongoing in Senegal. This indeed led to support from major foreign powers for the movement.

Whereas Chapters 2 and 3 focus more on the theoretical shortcomings in the literature on protest mobilization in Africa, Chapters 4,5, and 6 focus predominantly on methodological shortcomings. Chapter 4 zooms in on the use of micro-level theory to underpin macro-level relations and makes use of available datasets on protest events and protest. Chapters 5 and 6 focus on the method of event analysis and the use of news reports to capture events. Both chapters focus on a wide range of events. Besides protests and riots, armed conflict events are also considered, for example. This allows for a discussion of the methodological implications of measuring different types of events occurring in Africa.

Chapter 4 ‘Between fallacy and feasibility?  Dealing with the risk of ecological fallacies in the quantitative study of protest mobilization and conflict’ reviews several recent quantitative scarcity-social disorder studies and demonstrates how these predominantly make use of micro-level grievance theory to substantiate empirical relations between macro-level variables, such as scarcity indicators and event counts. This form of empirical testing stems to an important extent from the econometric study of civil war onset for which a lack of reliable data on the individual level in violent settings led to the use of macro-level approaches. However, the reasons for individual mobilization in low-level conflicts, including protests and riots, can more easily be captured by means of opinion surveys. Hence it is possible to generate micro-level data, which could be used to investigate mobilization dynamics in social disorder events. Micro-data can also reveal possible ecological fallacies arising from macro-analyses.

The paper shows empirically that macro-analyses risk leading to flawed inferences to the individual level. We do this by making use of macro data on protest and riot events (cfr. ACLED) and micro-data on protest participation (cfr. Afrobarometer surveys). We investigate two theoretical models. The first model states that poorer countries are likely to witness more protest events, because poorer, and hence more aggrieved, people are more likely to engage in protests. The second model holds that less democratic regimes are likely to experience more protests because people who rate their country as not being a democracy are more likely to engage in protests in order to voice their discontent about the lack of political participation, transparency and accountability. We test these relations both on the macro and on the micro-level. Results show how different relations can be found on both levels of analysis, and hence one has to be careful when inferring micro-level relations from macro-level data.

The empirical approach in the resource scarcity literature can be compared to early motivational theories discussed in the social movement literature. These stated that structural strains or economic stressors engender certain psychological processes (i.e. frustration, grievance), which were argued to add up to a collective action event. The micro-level was empirically taken for granted, however. Further empirical engagement with the micro-level in the social movement literature engendered additional theoretical insights. The paper hence argues that a similar development could improve our understanding of mobilization in Africa and refers to several methodological approaches drawn from the social movement literature which can be used to this aim.

Chapter 5 ‘How Events Enter (Or Not) Datasets: The Pitfalls of Using Newspapers in the Study of Conflict’ focuses on the errors that can arise from the use of news reports to study conflict events and introduces an analytical framework to understand and investigate these errors. The chapter discusses the increasing research interest in the establishment of conflict event datasets that focus on events in developing settings. Yet it is noted that much of the methodological debate with respect to the use of media reports to construct event data is to be found in Western-focused social movement research. While critical assessments of conflict event data have emerged in recent years, it is argued that this field of study stands to benefit from further engagement with the social movement, as well as communications literatures. We also devote attention to the implications of the focus on a wide range of events in the conflict literature, including protests and armed conflict, while the social movement literature is mainly focused on protests.

In order to support cross-fertilization between both literatures with regard to the collection and use of event data, the chapter sets forth the Total Event Error framework. This framework draws on insights from the survey research literature and the Total Survey Error framework. It distinguishes between measurement errors and errors of representation and captures bias as well as unreliability. Several crucial errors arise from the media source. Selection bias refers to the systematic selection of some events into the news by media while neglecting others. Description bias refers to how media describe events and their systematic tendencies in doing so. These errors arise from the rationale of the media source, but errors also arise in the data collection process. We focus on coding rules, intercoder reliabilities, the accuracy of external sources such as police records and NGO reports, and data adjustments in analyses such as data weighting and missing data imputation. Some errors are addressed in the social movement and communication literatures, but are considered only to a lesser extent in conflict event studies.

The chapter brings together insights from conflict studies, and the social movement and communication literatures. A such, it captures the current state of the art on the use of news reports to study conflict events. Drawing from these insights, we formulate guidelines for researchers developing and using event data. However, we also note that more methodological research is needed on events in developing settings, as well as on how errors can diverge for different types of violent and nonviolent events currently studied in the conflict literature. The Total Event Error Framework can be used to identify and situate errors and investigate them further.

Chapter 6 ‘Fit For Purpose? Selection Bias in African-Focused Conflict Event Datasets’ is predominantly concerned with media selection effects   and how datasets drawn from different news sources reveal differing patterns in the absolute and relative occurrence of conflict events, as well as their spread across Nigerian territory. The paper compares the coverage of conflict events of two existing cross-national datasets, SCAD and ACLED, with a self-composed dataset for the case of Nigeria. Whereas SCAD is based on Associated Press and Agence France Press news reports extracted from Lexis-Nexis, ACLED is based on online international and national news sources as well as NGO reports. The Nigerian dataset  (for which the acronym GDN is used from the newspapers on which the dataset is based) draws from the hard copy versions of The Guardian, This Day, and The Nation, three Nigerian newspapers with national coverage. The period covered by the dataset is April 2014 to March 2015, corresponding to a full year before the presidential elections. Archival research for the project was carried out from July to September 2015 in Lagos, Nigeria. The case of Nigeria was chosen because the country witnesses a large range of conflict events: Boko Haram violence and terror attacks, pastoralist-herder conflicts, electoral violence, riots, as well as protests. The dataset covered all these types of events, which also allows us to compare news errors for different forms of conflicts. All datasets cover the same types of events, ranging from protests and riots, to armed conflict events and terrorist attacks.

Firstly, the absolute counts of events indicate that SCAD, which is solely based on international sources, covers the least events, while GDN covers most events. The latter is interesting as ACLED draws from various Nigerian newspapers, but only online versions which could be the source for the underreporting. The findings hold for all conflict types. Secondly, the relative occurrence of events is different with more armed conflict events than protests and riots counted in SCAD, while the reverse is true for ACLED and GDN. Looking at the specific events identified in each dataset, and in line with the previous finding, it emerges that the overlap between datasets (i.e. the extent to which they identify the same events), is dependent on conflict type, with terrorist attacks, Boko Haram-related violence, and lethal violence being more likely to be included in all datasets. The overlap is low for low-level conflict events such as protests and riots. Finally, when looking at the sub-national distribution of protest events —the type of conflicts for which the datasets are most variable— it emerges that SCAD shows strongly divergent pattern from both ACLED and GDN.

The results demonstrate that it can be misleading to investigate protest and riot trends in Africa by relying on international or online news sources alone. Although the overlap between the datasets for more violent conflict events is far from perfect, results are all the more extreme for low-level violence —exactly those types of conflict which have received increasing research attention in recent years. Indeed, in the social movement literature, protest events are predominantly investigated by making use of national and local or provincial newspapers. The paper concludes that much remains to be learned on conflict events in Africa and that national news media proof an interesting empirical information source to this goal.

Chapter 7 ‘General Conclusion’ summarizes the lessons learned in the thesis and lays out an empirical research agenda for future studies. This research agenda flows logically from the findings in the thesis and is designed to address the theoretical and methodological shortcomings in the study of protest mobilization in Africa.