CHOOSING BETWEEN THE CHIEF AND LOCAL GOVERNMENT IN SENEGAL: HOW PEOPLE USE FORMAL AND INFORMAL INSTITUTIONS IN TIMES OF NEED By Christina Scheller A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Political Science—Doctor of Philosophy 2024 ABSTRACT This dissertation explores the ways in which people use formal and informal local government institutions in Africa, primarily in Senegal. I look specifically at people’s choices in contacting the local council versus the village chief. I expect to find variation along three dimensions: identity, issue area, and institutional structure. More specifically, I expect to find variation across gender and local/non-local status in the community in the patterns by which people use either the local council or their village chief when they need help with an issue. I argue that the customary sector is more likely to discriminate against women and non-locals, which would push them towards the formal sector compared to men and locals. I also expect variation across issue type, meaning that those with certain types of issues will be more likely to go to the chief, while other issue types will bring people to the council instead. I explore these two dimensions of variation specifically in Senegal using both Afrobarometer data and novel data from a conjoint experiment conducted in southern Senegal. Lastly, I explore cross-country variation in institutional structure in Senegal, Ghana, and Malawi using Afrobarometer data, as well. I argue that when chiefs are highly integrated into the formal system, people will favor the chiefs when they have an issue to solve. Likewise, when chiefs are not well integrated at all, people will favor the councilors when they have an issue to solve. The findings support my theory that identity type will affect the type of institution people will contact, though less evidence is found to support variation across gender. Land issues are found to still be the purview of the chief, while for more administrative issues, people favor the council. Lastly, while people favor the chief over the councilor in highly integrated systems and the councilor over the chief in less integrated systems, people still contact the chief quite a bit in less integrated systems, suggesting that institutional variation is important, but that chiefs continue to be resilient even in systems where they are sidelined. ACKNOWLEDGEMENTS They say it takes a village to raise a child, but I hadn’t realized it would take a village to raise a dissertation. This project would undoubtably not have been possible without the immense help and support from both my family and the faculty of MSU’s Department of Political Science. I would like to especially thank my dissertation committee, Jeff Conroy-Krutz, Michael Wahman, Ani Sarkissian, and Eric Chang, for their continued support and encouragement over the years. Ani’s thoughtful advice, Michael’s willingness to push my ideas, and Eric’s humor all helped me through this process. Also, thank you to Sarah Reckhow for working hard to make sure my second fieldwork trip was possible, as well as ensuring I had the extra time needed to finish after all the delays. I’m grateful, as well, for Sarah Krause for always finding answers to my questions and keeping track of the administrative details behind the scenes. I especially owe a debt of gratitude to Jeff for his enthusiasm for this project, his optimism, his patience, and his kind and nonjudgemental mentorship. From the early stages of brainstorming ideas to troubleshooting how to conduct fieldwork during a pandemic, to coaching me through each and every chapter, Jeff has provided a clarity to my muddled thoughts and ideas, without which, none of this would have been possible. I would also like to thank my top-notch research team in Senegal: Falaye Danfaga, Samba Kande, Omar Danfaga, Ibrahima Biagui, Rene Pierre Sambou, Bocar Balde, Aisha Balde, and Seydou Sy. Their hard work and dedication to the project kept everything running smoothly during the survey implementation and their friendship kept me company during my fieldwork and beyond. Samba especially worked tirelessly to coordinate the team and keep everything on track and for that I am grateful. This project would also have not been possible without the financial support of Dr. Michael Bratton and Ms. Anne Schneller through their Travel Fellowship, as well as the College of Social Science Research Scholars Award, and research funding from MSU’s Department of Political Science. Thank you for believing in this project. I am also forever grateful for my parents, Jeanne Scheller, Jack Wheeler, and Gwynne McClure for their support and for always believing in my abilities, even when I doubted them myself. For years, they’ve listened to triumphs, doubts, and evolving ideas on endless phone calls, talking me down and talking me up, as needed. iii Lastly, thank you to my family Kris, Hazel, Jackson, Lulu, and now Elliott. Kris in particular has weathered all the ups and downs of this extended process. She’s been by my side since Week 1, always armed with a joke to brighten my spirits and keep things in perspective. She kept me going through covid delays and watched Hazel during two fieldwork trips, despite the massive disruption to the household that cat integration turned out to be. I couldn’t have done this without her. Thank you also to Hazel, Jackson, and Lulu for providing distraction and cuddles when I needed it the most. And finally, for Elliott, who has just joined us at the tail end of this journey: you’re just what I needed to get this project over the finish line. iv TABLE OF CONTENTS CHAPTER 1: INTRODUCTION…………………………………………………………………1 CHAPTER 2: A THEORY OF WHO USES FORMAL AND INFORMAL LOCAL INSITUTIONS…………………………………………………………………………………...10 CHAPTER 3: IDENTITY AND USING FORMAL AND INFORMAL LOCAL INSTITUTIONS…………………………………………………………………………………28 CHAPTER 4: ISSUE AREA AND USING FORMAL AND INFORMAL LOCAL INSTITUTIONS…………………………………………………………………………………55 CHAPTER 5: INSTITUTIONAL STRUCTURE AND USING FORMAL AND INFORMAL INSTITUTIONS…………………………………………………………………………………99 CHAPTER 6: CONCLUSION…………………………………………………………………124 BIBLIOGRAPHY………………………………………………………………………………130 v CHAPTER 1: INTRODUCTION 1.1 Introduction In the period after most African countries gained independence, many observers predicted the chieftaincy would fade away. In 1970, Pierre Alexandre wrote, “The problem of chieftaincy in Africa would seem today…to be outmoded, a thing of the past relegated to the background by the more pressing questions of political, social, and economic development which are more in tune with the modern world” (1970: 24). Many saw chiefs as “representatives of the old order [and] obstacle[s] to progress and modernization” (van Rouveroy van Nieuwaal 1987: 3). Chiefs often worked as local-level functionaries in colonial administrations, which tarnished their reputations and decreased their legitimacy (Boone 2003). Post-colonial leaders, in their efforts to supplant colonial-era institutions, were expected to replace the chieftaincy with bureaucrats and civil servants (Baldwin 2016). These predictions were not entirely wrong: there were changes to the chieftaincy after independence. Indeed, many post-independence governments limited or even abolished the chieftaincy (Wilfahrt 2022). Certain chiefly responsibilities, such as tax collection, customary land tenure, and conflict resolution were diminished in the years after independence (Baldwin 2016). Even as late as 1996, Mamdani famously called traditional authorities “decentralized despots” and argued that they were inhibiting African countries from democratizing. As such, the more democratization took hold on the continent, the more bureaucrats, civil servants, and elected officials would replace the chiefs. Yet, ultimately, the chieftaincy did not dissolve. As time went on, predictions of its downfall shifted to surprise in its resiliency (Geschiere 1993, Buur and Kyed 2007, Logan 2009). Instead of fading away as the state gained in strength, in many countries, chiefs held onto power and in the wave of democratization that swept across Africa in the 1990’s, their powers even began to grow. Several countries have institutionalized the chieftaincy, defining its role in the constitution or through other legal means (Henn 2022). Some countries, including Ghana and Zambia, have even created a House of Chiefs with responsibility over all matters pertaining to the chieftaincy. Chiefs also experience high levels of popular legitimacy and public trust. In fact, 1 a majority of Afrobarometer respondents would like to see chiefs’ influence expand beyond what it is today (Logan 2013). Still, not everyone saw this resilience as a boon for democracy. In the 1990s, in the wake of democratization, there was a push by international development agencies to decentralize power from the central state to elected local governments (Ribot 2003). While many countries went on to create these formal systems of local government, they often failed to devolve significant power to these institutions. Instead, leaders sidestepped the decentralization push by devolving power to unelected chiefs in addition to the new set of formal officials. Ribot (2003) argues that central states devolved power to customary authorities as a “backlash against local democratization” (56). In other words, many states have implemented democratic institutions of local governance, but purposefully kept those local institutions sidelined. Instead of working through these formal institutions at the local level, many have maintained the chieftaincy as a parallel system through which to work, as chiefs might be more accountable to the center, compared to locally elected officials, who would theoretically be more accountable to the populations that select (and fire) them. Thus, according to Ribot, the central state’s continuation of the chieftaincy is just a way to keep the ruling party’s adversaries out of power at the local level. For Ribot, this is a problem. By not fully decentralizing power to democratically elected local officials, states are not providing citizens with any alternatives to chiefs for local governance, even if people desire such alternatives. In other words, he argues, people would not choose the chief if they did not have to. 1.2 Theory Today, many of these parallel systems of local authority are still operating in countries across Africa (Holzinger, Kern, and Kromrey 2016). It is not unusual for local councils to have power over many aspects of local government, while chiefs maintain local authority, as well. Often, there is a degree of overlap between the powers of the formal and informal sectors of local government. When such overlap exists, it gives people a choice on whom to see when they have an issue that needs solving. Ribot would suggest that people will choose the formal system over the chief in such instances. However, we know little about how people actually use local government institutions when authority overlaps between the formal and informal sectors. Do 2 people favor one type of official over another? Do some people prefer to use the formal system, while others prefer the informal system? If so, which types of people prefer each system? On the one hand, Ribot could be right. As local government institutions professionalize, it’s reasonable to think that people would prefer to use the system that represents the rational, bureaucratic state over the customary system (Weber 1978). The formal system may be more likely to provide fair and equitable services and may be easier to hold accountable through local elections compared to a customary system that largely serves for life (Ribot 2003). On the other hand, the fact that the chieftaincy has not faded away as countries democratize suggests that maybe people still prefer the informal sector of local government. Chiefs maintain high levels of trust and legitimacy (Logan 2013, Baldwin 2016). People may also have ways of holding chiefs accountable, despite the fact that they are not popularly elected like local councilors or mayors are. Baldwin and Holzinger (2019) argue, for instance, that chiefs benefit from alternative mechanisms of accountability, such as using consensual decision-making and having to publicly justify decisions. Thus, it’s also reasonable to think that people will continue to use the chief for assistance, even after democratically elected officials are available at the local level. I argue that the answer is likely much more nuanced. There is likely some variation in how people make these decisions. First, I expect that one’s identity may affect how someone interacts with formal and informal institutions. For example, women may show a different pattern of interacting with local institutions than men. This could particularly be true if women suffer from different rates of discrimination across the two sectors. In addition to gender, I look at those with local and non-local status in the community. Non-locals are generally outsiders, such as migrants, and they may also face discrimination in the customary sector, causing them to seek assistance from the formal sector more than locals with stronger ties to the community. There’s reason to think that discrimination may be more of an issue in the informal sector, as it’s less likely to be held to rational bureaucratic norms of the state (Weber 1978, Joireman 2008). Therefore, identity is one dimension along which people’s choices may vary. The type of issue that someone is having is another dimension that may influence who people seek out for assistance. While variation along identity types means that different people will make different choices depending on their identity, variation along issue area means that the 3 same person might make different choices in whom to contact, depending on what issue they’re having. For instance, much of what the local government does is provide administrative services for obtaining certificates and documentation, such as birth, marriage, and death certificates, titles, and other documentation. People may be more likely to use the formal sector for these services because they are the type of services that were created by the administrative state and were likely nonexistent under customary systems prior to colonialism (Scott 1999). Other issues may be the purview of the informal sector. For instance, chiefs have commonly been in charge of conflict resolution and, therefore, people may prefer using the chiefs for these types of issues compared to the formal sector (Logan 2013, Wilfahrt and Letsa 2023). Lastly, there may be variation in how people choose between these two sectors depending on the structure of the institutions themselves. There is variation across countries in the way formal and informal local institutions are designed and the degree to which their powers overlap. For instance, in some countries, chiefs are quite integrated into the local council. In such cases, chiefs may serve on the local council and are generally much closer to local decision-making than in less-integrated systems. Other countries may have a chieftaincy that is less integrated into the local formal system, where chiefs may not serve as members, but are consulted as necessary. Lastly, in some countries, chiefs’ powers are kept quite separate from the powers of the local formal system with very little overlap in responsibilities. Across these cases, we will likely find different patterns of interaction between the formal and informal sector. 1.3 Empirical Strategy I test these theories focusing on Senegal as my main case. I then add Ghana and Malawi as contrasting cases when I look at variation in institutional structure. Senegal is a good case for studying this type of institutional overlap for a few reasons. First, Senegal has a long history of decentralization1, meaning some power has devolved to local officials for longer than most comparable countries. However, while the formal system has evolved over time, village chiefs have also maintained their place in village governance, as well. Chiefs have been partially incorporated into the state in Senegal, serving as the representative of the president at the village level (Wilfahrt 2022). Thus, chiefs have some responsibilities that overlap with those of the local 1 The first Act of Decentralization was passed in 1974. 4 council. This makes Senegal a particularly good case at studying how people choose which sector they go to with their problems. I use Afrobarometer data in my initial analyses and examine variations in how often people contact chiefs and local councilors in Senegal. I then use novel data from a conjoint experiment with a vignette with four attributes, which are varied at random to further explore these questions. The survey was conducted in the fall of 2021 in southern Senegal. The conjoint experiment allows me to vary several attributes at once, as well as ask all types of people—even those who have not themselves contacted a local official—what they think someone should do if they need assistance. Overall, I find some variation in the patterns with which people from different identity groups seek assistance from local authorities. I find that both men and women still overwhelmingly choose to go to the chief for assistance, though women are generally less likely to seek assistance than men. Those with non-local status are less likely to contact the chief than locals but show no difference in contacting the councilor. This suggests that non-local status is not keeping non-locals from using the formal sector, but outsiders may feel less inclined to use the informal sector when they have a problem to solve. I also find that issue type contributes to people’s decision-making, with more people likely to go to the mayor for a birth certificate than an issue involving land. Finally, institutional structure does seem to play a role in how people use local institutions, with people in a system in which the chief is more integrated into the formal sector more likely to see the chief than those in less-integrated systems. 1.4 Overview of Chapters Chapter 2 presents my theoretical expectations for the first two empirical chapters. Chapters 3 and 4 delve into my empirical analyses focusing on Senegal. In Chapter 3, I look at the effect of identity on the way people use formal and informal local institutions, particularly focusing on gender and local/non-local status. In Chapter 4, I expand on this analysis, continuing to look at identity, and adding issue area, as well. Chapter 5 broadens the analysis beyond Senegal to look at country-level variation in institutional structure. Finally, Chapter 6 concludes the dissertation. I further discuss each chapter below. 5 1.4.1 Chapter 2 Chapter 2 situates my research into the broader context and then explores my theoretical argument along many dimensions, including identity and issue area. I start by discussing the roles of formal and informal actors at the local level, particularly chiefs and elected local officials. These are the local officials that people are more likely to be interacting with and seeking out in times of need, so understanding the relationship between the two types of local officials is important for understanding how people use formal and informal institutions at the local level. I then present my theory of how people use local institutions and how it may vary by gender, local/non-local status, and issue area. I expect that women will be more likely to face discrimination in the customary sector, and thus will be more likely to go to the councilor compared to the chief, relative to men. I expect to find a similar relationship between those with local and non-local status, with non-locals more likely to go to the councilor compared to the chief, relative to locals. I also expect to find that some types of issues are more the purview of the formal sector, while for other types of issues, people are more likely to use the customary sector. 1.4.2 Chapter 3 As the first empirical chapter, Chapter 3 uses Afrobarometer data from Senegal to explore how identity affects the way people use formal and informal local institutions. More specifically, I look at how gender and local/non-local status affect people’s choices of which institution to use. This chapter begins with a discussion of the decentralized governing system in Senegal and how it has evolved since it was introduced in 1974. This provides context for my theory and explains the roles of local councilors, mayors, and village chiefs, among other officials. Most importantly, this section underlines the overlap in responsibilities between local councils and village chiefs. I then present my theoretical expectations, particularly that while there may be an overall inclination for people to use the chief over the local councilor, this tendency will depend on one’s identity, as women and those with non-local status may feel less welcomed in the customary sector than men and those with local status. If the customary sector is more likely to discriminate 6 against some groups over others, then I expect those groups to be more likely to seek out other avenues for solving their problems, mainly in the formal sector. I use a difference-in-differences test to study the gaps between different identity groups’ likelihood of going to the chief versus the councilor. Contrary to my expectations, women’s tendency to go to the chief over the councilor is actually the same size as men’s, though men are more likely to seek assistance overall (i.e. men go to both sets of actors more than women do). This suggests that, despite the likelihood of discrimination by customary institutions, women do not show any proclivity toward formal institutions, and they show the same degree of preference for informal institutions vis-à-vis formal institutions as men (i.e. the difference-in-differences is not significant). This might be because women, like men, see informal institutions as inherently more legitimate than formal ones, despite traditional discrimination. It might also be because they see more discrimination from formal institutions and/or they do not know how to or have the means to access formal institutions. The findings on the local/non-local dimension are different, however, and are more in line with my theoretical expectations. I find that non-locals are less likely to contact the chief than locals but show no difference from locals in contacting the councilor. This suggests that non-locals see potential discrimination from chiefs that they do not see from formal institutions. 1.4.3 Chapter 4 Chapter 4 continues my exploration of how people use formal and informal local institutions. I use novel survey data from a conjoint experiment to explore this question further. The conjoint experiment uses a vignette to vary four attributes about a hypothetical individual facing a hypothetical situation at random: the individual’s gender, their local/non-local status, and the issue they are facing. I also vary the time in which the individual is facing the issue, to try to capture whether institutional changes in Senegal over the last decade have resulted in over- time changes. Here, I seek to determine whether variation in these attributes affects the respondent’s assessment of the likelihood that the hypothetical individual will seek out help from the informal and formal sector. I find that neither the gender nor local/non-local status of the hypothetical individual in the vignette affects respondents’ assessments of whom that individual will see for help. I also find that the timing of the vignette does not seem to matter. However, I 7 find that respondents rate the individuals as more likely to see the chief if they’d like to buy land or have a land conflict, compared to if they need to obtain a birth certificate for a child. This suggests that issue is indeed a factor in how people decide whom to approach for assistance. 1.4.4 Chapter 5 In my final empirical chapter, I expand my analysis outside of Senegal to explore how country-level variation in institutional structure affects the way people use local institutions. More specifically, I look at the degree to which chiefs are integrated into the local council and its effect on whom people choose to contact when they have a problem. At one end of the spectrum, I look at Ghana, where chiefs have historically been sidelined at the local level and are not incorporated into the council at all. I compare that case to Senegal, where chiefs are not formal members of the council, but still meet with councils as needed. Finally, at the other end of the spectrum sits Malawi, where chiefs are full members of the council and are thus much closer to the decision-making than chiefs in either Senegal or Ghana. These three cases highlight a range of institutional structures. I expect that in Malawi, where the chief is highly integrated, people will go to the chief more than the councilor, there will be a larger gap between those contacting the chief and those contacting the councilor compared to Senegal, and that the gap will favor the chief. In Ghana, I expect the that people will go to the councilor more than the chief, that there will also be a larger gap than in Senegal, but for the gap to favor the councilor. Overall, I find support for my theory with the Malawi case with people preferring the chief and a larger gap between those contacting the chief compared to the councilor. In Ghana, I do find that people prefer the councilor over the chief, but I actually find a smaller gap compared to Senegal. This suggests both that institutional structure matters and that chiefs are still an important venue for people to bring their issues, even if they are not formally integrated into the local formal government. 1.4.5 Chapter 6 In the final chapter, I conclude my discussion of how people use formal and informal local governing institutions. I discuss the overall findings in the dissertation and address some 8 limitations of the analysis. I go on to discuss some new questions raised by my findings and suggest a few avenues for further research. 9 CHAPTER 2: A THEORY OF WHO USES FORMAL AND INFORMAL LOCAL INSITUTIONS 2.1 Introduction In much of Africa, there are parallel formal and informal institutions that provide services and assistance to the public. In the formal sector, most countries have some form of local governance that consists of a mayor or local elected official with some executive power, as well as a local council, which is either elected or partially elected, that determines the outcomes of local policy issues. On the informal side, most African countries have local traditional authorities of some kind, most commonly some sort of chief. While there is a lot of variation in the design of both the formal and informal sectors across Africa, there is commonly some degree of overlap in the duties assigned to the two types of local governance. When this happens, citizens often have a choice of whom to go to for assistance when they have a problem. Yet, little is known about how people use these overlapping local institutions. Do some people choose one type of local actor over another? Do some people prefer to visit formal leaders, while others prefer informal? If so, which types of people? It's possible that as the formal sector of local governance develops, one might predict that people will be more inclined to use the formal sector over the informal one. As part of a more rational, bureaucratic state, people may find the formal sector to be better at providing services in a fair and equitable manner, at least compared to the customary sector (Weber 1978). Chiefs, on the other hand, have been seen as “obstacle[s] to progress and modernization” (van Rouveroy van Nieuwaal 1987: 3). In 1996, Mamdani referred to traditional authorities as “decentralized despots” and argued that they were preventing African countries from democratizing. Ribot also (2003) argues that “Empowering authorities that are not held downwardly accountable to local populations…can imperil democracy by taking resources away from emerging democratic structures while strengthening and helping to entrench the very non-democratic institutions that democratic reforms aim to replace” (56). Thus, in this view, when given a choice, many people will choose the democratic and accountable local actors over the customary ones when seeking assistance. Others have noted the resiliency of the chieftaincy, even as countries democratize. Chiefs have continued to maintain popular legitimacy, are well trusted, and are often given higher marks 10 on performance than their counterparts in the formal state (Logan 2013, Baldwin and Holzinger 2019). Thus, it’s also possible that when given the choice, people will prefer the informal sector instead. I argue that both of these theories are too simplistic. There is likely to be some variation in how people make the decision of which local institution to use when they have a problem. I expect there to be variation along two dimensions: identity and issue area. One’s identity may affect whom they go to see because people have different relationships to these institutions, different cultural expectations, and different resources. For instance, there may be differences between how women and men make these decisions, particularly if levels of discrimination discourage women from seeking help from one sector over another. A similar dynamic may exist among those who are not local to a community. Outsiders may feel less comfortable seeking assistance from the customary sector and choose the formal sector instead when given a choice. As Joireman (2008) writes, “Traditional authorities can practice the politics of exclusion, excluding resources from groups with less political power, such as divorced women and migrants, who are easily labeled and denied access to land communally held” (1237). Thus, identity may affect people’s choices on whom to see for assistance with a problem. The second dimension is issue area. The same person may choose to go to different institutions for different types of issues. Some issues we may expect to be the purview of a given sector. For instance, the local formal state likely maintains responsibility over official documentation, such as birth, death, and marriage certificates. Because these types of documentation were designed by the administrative state (Scott 1999), it is likely that someone in need of such documentation would opt to go to the formal sector to meet that need. On the other hand, tasks such as local conflict resolution have long been the purview of the customary sector. Logan (2013) finds that when chiefs play a large role in resolving conflict, it is one of the strongest predictors of local support. Thus, people in need of solving local conflict may be more likely to seek help from the customary authority over the formal one. However, I argue that these divides are not as clear as one might think. Take, for instance, land issues. In many places, chiefs’ control of land has been seen as a large source of their power (Ntsebeza 2005). Yet, the formal state has taken over authority over land in much of Africa. In Logan’s (2013) analysis, people reported that the local government had more of a role than 11 traditional leaders in allocating land in 11 out of the 18 countries included in the analysis. Thus, it’s unclear if people will select the customary authority due to its high levels of popular support and historical ties to the issue, or if they will prefer the formal sector due to its legal control over the issue. It is important to understand more of the nuance in how people use local institutions because these local actors are the closest arm of the state to the population. There has been a push by development organizations for decades to increase the power given to local officials through decentralization initiatives so that local government can provide solutions tailored to the local population (Ribot 2003, Wibbels 2019). At the same time, chiefs maintain power to varying degrees in many states across Africa (Baldwin 2016). Yet, little is known about how people navigate these local systems when they need assistance. In this chapter, I first discuss the roles of formal and informal actors at the local government level. I then look at local governance more broadly in Africa, particularly focusing on the push towards decentralization. I also examine issues around identity, including gender and local/non-local status, and what we know about how these identities affect interaction with government at the local level. Finally, I look at issue areas that may affect who one seeks for help, including land ownership and conflict at the local level and administrative tasks. 2.2 Formal and Informal Local Officials In this chapter, I discuss a theory on how people use local institutions when overlap in authority blurs who the “proper” authority is. Use of local institutions is under-studied and important for understanding how the structures and characteristics of local institutions are actually affecting the everyday use of these institutions by the people for whom they were created. First, though, to understand this relationship, I will discuss two common types of local actors with overlapping authority: traditional chiefs and local elected officials. 12 2.2.1 Chiefs Chiefs are often important actors at the local level. Baldwin defines traditional leaders2 as “rulers who have power by virtue of their association with the customary mode of governing a place-based community” (2016, 21). This definition is helpful because it allows for variation across chieftaincy institutions, while stressing the importance of the customary nature of their legitimacy and power. Chieftaincies vary in their historical ties: while many chiefs can trace their lineage to pre-colonial times, some chieftaincies were created whole cloth by colonial administrators, though this was rare (Mamdani 1996, Baldwin 2016). Some chieftaincies are hierarchical, with different types of chiefs operating at different levels, as in parts of Ghana, Malawi, and Zambia. In other systems, such as Senegal and Mali, there are village-level chiefs without anyone presiding over them within the customary system. In hierarchical systems, the top of the hierarchy will sit the paramount chief, with senior chiefs, chiefs, and sub-chiefs below them. The chiefs at the bottom rung of the hierarchy are generally called village headmen (Baldwin 2016). My theory focuses mainly on this bottom level because in hierarchical systems, it is the closest level to the villagers. Thus, when people seek assistance, they are likely seeking a village headman. Chieftaincies also vary in the extent to which they are incorporated into the state. On one end of the spectrum, chiefs can be highly incorporated into the state. For instance, chiefs can be incorporated into the constitution, which Baldwin (2016) finds to be the case in 41% of African constitutions in 2010. Some countries, including Ghana and Zambia, have even created a House of Chiefs with responsibility over all matters pertaining to the chieftaincy. At the other end of the spectrum, the chieftaincy remains a true parallel institution, in which chiefs operate at the local level parallel to the state, but not formally incorporated into it (Henn 2022). Many countries are somewhere in between, where chiefs hold some degree of official power at the local level but may not be so institutionalized as to be incorporated into the constitution or organized nationally. The roles and responsibilities of chiefs also vary across different institutional settings. Historically, chiefs often played a large role in control of land and managing conflict. In many settings, they maintain some responsibilities over these issues today (Logan 2013). Other 2 Traditional leader is a term used to broadly encompass many types of local actors, some of whom may not specifically be chiefs. I use it here interchangeably with chief. 13 responsibilities may include tax collection, local goods provision, and facilitating community projects (Baldwin and Raffler 2019. Chiefs also play a cultural role in maintaining traditions and customs. The majority of chiefs across Africa are still largely unelected, with positions passed down within families to male heirs who rule for life (Baldwin 2016). Some have argued that the unelected nature of the chieftaincy would put chiefs at odds with democratic governance. Mamdani (1996) famously critiqued the chieftaincy for being captured by first the colonial state and later the post-colonial state. He argues that turning traditional leaders into operatives of the central state caused them to lose credibility and legitimacy in the eyes of the populace. Ribot (2003) makes a similar argument: that the chieftaincy exists only at the behest of the state, as a way for weak states to decentralize governance to local actors without empowering democratic actors at the local level. And yet, the chieftaincy survived democratization. Much has been written on the resiliency of the chieftaincy. Logan (2013) finds that chiefs are quite highly regarded across Africa, with high levels of trust and support. Some argue that due to chiefs’ high regard, they make good vote brokers (Koter 2016, de Kadt and Larraguy 2018). In other words, chiefs are able to use their power and authority over local populations to rally voters to a specific candidate. However, with the secret ballot, it’s difficult to know how powerful chiefs really are at garnering support. Baldwin (2016) argues instead that chiefs serve as development brokers, working with national politicians to secure investments in their communities. In return, voters vote for the national politicians that their chief has strong ties to because it will likely lead to higher local investment in the future. However, focusing on the broker relationship between chiefs and national politicians ignores the relationship between chiefs and the other state actors they engage with at a local level all the time. I’ll discuss local elected officials next. 2.2.2 Local elected officials Just as in much of the rest of the world, local government in the formal sector in Africa typically consists of locally elected officials who serve on a council and some sort of executive, such as a mayor. In many instances, the executive may serve as the head of the local council, as well. Bratton (2012) defines local government as, “The set of formal institutions legally 14 established to deliver a range of specified services to relatively small geographic jurisdictions” (517). This definition is useful in that it clearly specifies that local government is legally established as part of the formal institutions of state, while allowing for variations in size of sub- national unit and responsibilities devolved to them. In the hierarchy of decentralized government, this is the bottom rung of state officials. As such, elected local government officials tend to be members of the local community itself and are less likely to be outsiders or central government elites parachuting in to make local policy. Local governments have varying administrative and fiscal responsibilities; however, most local governments provide some level of service provision (Rodden and Wibbels 2019). This could include services such as education, health care, natural resource provision, infrastructure, and land allocation, among others (Ribot 2003, Bratton 2012, Wibbles 2019). Local governments are also often involved in administrative record-keeping, such as providing birth, death, and marriage certificates, as well as various titles and permits. While many local governments have some level of fiscal responsibility to collect taxes, local governments in Africa often rely more on central state transfers and international donor funding (Rodden and Wibbels 2019). The push to create local governments in Africa occurred largely in conjunction with the push for democratization in the early 1990s. As states wrote new constitutions and adopted democratic ideals, they incorporated ideas around decentralization, or the transfer of power away from the central state to varying levels of subnational units (Dickovick and Wunsch 2014). Decentralization was often encouraged by donors and international organizations, such as the World Bank and the IMF (Caldeira et al. 2015). The idea behind decentralization was simple: devolve power closer to the people, and citizens can get policies and services tailored to their local context. Indeed, Bratton (2012) finds that people tend to find local councilors to be more responsive than central government officials. Erk (2018) writes, “In all cases, decentralization was presented as a way to bring in better democracy, better public policy formulation and delivery, economic development and growth” (537). Unfortunately, the reality has not always been so ideal (Rodden and Wibbels 2019). The hierarchy of decentralized structures can easily overlap with clientelist networks, which can limit the accountability that decentralization promised (Leon and Wantchekon 2019). Likewise, overlapping responsibilities between sub-national units makes it difficult for people to know 15 whom to hold accountable for underperformance (Wibbles 2019). Another issue with decentralization initiatives has been that funding mechanisms are often not enough to support successful local governance, and many local governments have been found to be ineffective. Bratton (2012) writes, “Local governments in Africa tend to have acquired few powers, attained limited technical competencies, and often been subordinated—politically, organizationally, and fiscally—to central states” (517). Despite the lack of means, local governments continue to operate and are often the first or even the only part of the state that citizens commonly interact with. This means that the relationship between local elected officials and citizens is important for understanding citizens’ relationship with government as a whole. In addition, local elected officials make up the formal sector of local governance, but they are not the only actors providing services on the local levels. The next section examines the relationship between local elected officials and traditional chiefs. 2.2.3 Relationship between Local Elected Officials and Traditional Chiefs Much of the discussion on both chiefs and local government examines the relationship between the central state and local actors. While this relationship is important in a broad sense, it doesn’t tell the story on the ground where these actors operate. What is the relationship between chiefs and other local officials? Local council members and other officials such as mayors operate in the same sphere as the chief. These are the arms of the state that regular citizens have the highest likelihood of interacting with, so this relationship is worth exploring. In this vein, Clayton, Noveck, and Levi (2015) find that competition between chiefs and local elected officials in Sierra Leone yields increased public goods provision, while collusion between these actors leads to the provision of fewer public goods. Essentially, the more contact between chiefs and local councilors, the lower the public goods provision will be in a community, while the more conflict reported between chiefs and local councilors, the higher the public goods provision will be. Myers and Fridy (2017) examine how Ghanaians navigate formal and informal local institutions. They find that when people evaluate chiefs’ performance highly, they are also likely to evaluate district assemblymen’s performance highly. They find a similar relationship between chiefs’ performance and central government officials, such as members of parliament. They 16 suggest that this indicates that chiefs and government officials have a cooperative and collaborative relationship. Wilfahrt (2018a) finds the relationship between chiefs and local councilmembers in Senegal to be cohesive and local government to be stronger in areas that were politically centralized in the precolonial period, while areas that were historically acephalous are more likely to suffer from competition and division between elites. This suggests subnational variation in the relationship between chiefs and local elected officials, based on ties to historical lineages. Henn (2022) also finds variation in the relationship between local officials and chiefs, depending on whether the chieftaincy is institutionalized in the constitution and the level of state capacity. When chiefs are institutionalized in states with high state capacity, chiefs have less influence compared to their counterparts in states without an institutionalized chieftaincy and low state capacity. In other words, when the state is strong and has more control over the chieftaincy, local officials can supplement local chiefs, while weak states without control over the chieftaincy result in a chieftaincy that complements other local officials. In short, there is strong evidence that the relationship between local officials and chiefs is important for local outcomes. Yet, while these studies start to offer a start at understanding the dynamics of governance at the local level, they fail to incorporate local citizens into the story. We still know little about how the community interacts with either chiefs or local elected officials. By looking solely at elites, we fail to understand how the relationship between elites affects people’s ability to obtain goods and services or solve their day-to-day problems through local actors. More specifically, how do people navigate using a local governance system with multiple types of local actors, both from the formal and informal sectors? And how does one’s identity or issue type affect whom they seek for assistance? Understanding this dimension will allow a more holistic understanding of how local governance operates in areas where elites have overlapping responsibilities. In the next section I will discuss more about how people use these local institutions. 2.3 How Do People Use Local Institutions? Given that people have a choice, whom are they more likely to go to for help? Do they select the formal sector for assistance, given the expansion of decentralization policies that have 17 brought the state to the people at the local level? Or do they select the informal sector because of their familiarity with and high levels of trust in the chief? On the one hand, people may choose the formal sector over the informal one. The formal sector consists of officials who maintain legitimacy through the decentralized powers of the central state. In some states, these officials are popularly elected; in others they are appointed by the central government. In still other systems, such as Uganda, parallel sets of officials serve at the local level, some elected and some appointed. In theory, these officials are beholden to the legal authority invested in them by the state and thus are to offer constituents consistent and impartial solutions to local problems. As Weber (1978) describes, the rational-legal state gets its authority from the laws enacted by the state. Therefore, people should come to know what to expect when going to local state officials for help with their problems because the solutions should be similar, regardless of who is in need of help. Of course, this description is likely somewhat idealistic, since officials are people and can still have biases even when working within the law. However, I argue that while individual officials or bureaucrats may have room to discriminate within the system of laws, the level of discrimination by state actors will be much lower than by customary actors, who are beholden primarily to custom and tradition. Likewise, chiefs tend to be unelected local officials who serve for life. They also are likely to hold much less legal authority to solve local problems or provide services than officials in the formal sector who have received decentralized powers from the central state. Ribot (2003) and Ntsebeza (2005) both argue that chiefs lack accountability and that, if given a real choice, people would use the formal sector instead. Indeed, even Weber describes (1978) traditional authority as an outdated form of governance, where a ruler’s legitimacy is derived from tradition instead of through legal authority. Again, because chiefs lack accountability, there is a chance of unchecked bias towards certain groups if they use the informal sector. More specifically, long- rooted customs around inheritance and land issues may exclude groups such as women or outsiders (Winters and Conroy-Krutz 2021, Boone 2014). There’s a sense that people go to chiefs because they lack better options (Ntsebeza 2005). On the other hand, people may choose the informal sector over the formal one. Traditional leaders maintain high levels of trust, lower levels of corruption, and are seen as better listeners than local councilors (Logan 2013). Chiefs may also be more convenient for people to 18 access. For example, village headmen—the bottom rung of chieftaincy in societies with hierarchical chiefdoms—are a feature of every village, while often in rural areas, local councilors or other local elected officials may represent a larger jurisdiction and thus have more constituents overall. Chiefs may also be more efficient, as they are not beholden to the sort of complex bureaucracy that might inhibit state actors. While both of these options seem plausible, I argue first that people will likely seek assistance first from the chief, all else equal. Chiefs maintain a lot of legitimacy from their long history of customary rule. In fact, they often have more legitimacy than elected officials at both the local and national level (Chasukwa et al. 2019). Chiefs also usually serve for life, meaning they have longer time horizons than local elected officials (Baldwin 2016). This means that people have time to get to know the chief, whereas local officials may change every term. People are likely more familiar with what the chief can do for them than they are with the roles and responsibilities of local elected officials. While chiefs’ power has not been static and is ultimately up to the state (Boone 2003), popularly elected local governments have only existed in most African countries since the late 1980s or early 1990s with the push for decentralization reforms. Often times, their power has been reformed multiple times since then. Thus, it may be hard for citizens to keep up with what the local government can do for them when compared to what the chief can. That all said, I believe the assumption that people will generally seek the chief over the formal system is overall too simplistic. People vary along several dimensions and some of these dimensions will likely affect their choices in which type of local actor to seek for assistance. More specifically, I argue that there is likely to be variation along two dimensions: identity and issue area. I discuss two types of identity that may affect the type of local actor that someone would go see: gender and local/non-local status. Women, for instance, may be more likely to seek assistance from the formal sector because the bureaucratic state may be more likely to treat them equally under the law than leaders beholden to long-standing customs. Those with non- local status are likely to make decisions using a similar logic. I also expect to see variation along different issue areas. The same person may choose to go to different actors depending on the type of issue they’re having. I look at two types of issues: administrative tasks and land issues. For instance, people may be more likely to use the formal 19 system for administrative tasks, such as getting a birth or marriage certificate, because these are tasks that were created by the formal system. In contrast, land issues have long been the purview of customary authority, suggesting people may feel more comfortable going to chiefs for such issues. In the next section, I discuss further how I expect gender to affect who someone sees for assistance when they need help. I will then discuss local/non-local status, followed by two types of issue areas: administrative tasks and land issues. 2.4 Gender We know that men and women differ in a lot of areas of political participation. For instance, across Africa, men are more likely than women to vote and more likely to hold elected office (Amoateng et al. 2014). The gap in participation between men and women has been linked to women’s lack of resources, such as education and labor force participation (Coffe and Bolzendahl 2011, Isaksson 2014, Logan and Bratton 2006). Yet, these explanations often fail to account for all the variation the African context. For instance, rural voters and poor voters may be expected to participate less due to less access to resources than their urban or well-off counterparts. However, both groups have actually been found to participate more (Logan and Bratton 2006). Similarly, Isaksson, Kotsadam, and Nerman (2013) look at various factors that explain the gender gap in the Western context, such as resources, employment, and religion, and study their effects across 20 African countries. They find these factors explain only a small portion of the gender gap and suggest instead that male-dominated clientelist networks may be an important factor. Logan and Bratton (2006) do find some support that variation in education level explains some of the gender gap in political participation. However, attempts at increasing access to information for women have often backfired. Gottlieb (2016) looks at how increased knowledge affects civic engagement among men and women in Mali. In her field experiment, she finds that male participants increased civic participation after taking a civic education course, while women decreased civic participation after taking the same course. She suggests that this perverse effect occurred because women’s participation in the course defied the entrenched gender norms in Mali. Thus, because women wouldn’t normally attend such an event, instead of empowering 20 them, it limited their future participation in the civic sphere. This suggests that increased education or knowledge can only help decrease the gender gap in participation as much as the local cultural norms will allow. Another large literature explores how cultural norms can influence the gender gap in political participation. As Robinson and Gottlieb (2019) write, “Societies promote different beliefs about which individuals are welcome in the public sphere, and women are frequently thought to be unwelcome participants. This helps explain why decreasing the material costs of participation alone can often fail to improve outcomes for women” (2). Logan and Bratton (2006) find that women profess less confidence and less interest in the political realm, which they suggest is evidence that, “…in Africa, politics is a realm of action reserved for men” (26). The importance of culture in explaining political participation is furthered by Robinson and Gottlieb (2019), who compare matrilineal societies in Africa with patrilineal ones. They find that matrilineal societies have a smaller gender gap in political participation and attribute it to the expectations that matrilineal societies have around women’s place in the community. Thus, cultural norms seem to affect women’s level of comfort in participating in politics. There is also variation in women’s political participation across contexts. For instance, women are more likely to contact officials when more women hold office at the national level, usually through gender quotas (Barnes and Burchard 2013). Yet, this, too, may face backlash: Clayton (2015) finds that in Lesotho, gender quotas actually reduce women’s political participation. Most of what we know, however, only looks at participation in the formal sector of governance: acts like voting or contacting an elected official. Less is known, however, about how men and women differ in their interactions with the informal sector. While not specifically looking at political participation, Logan (2013) looks at whether people support increasing chiefs’ power. She finds no difference between men’s and women’s support. She asserts that it is often assumed that chiefs are “inherently bad for women,” but uses this finding to say, “women themselves don’t appear to see it that way” (Logan 2013, 370). However, while there may not be differences in hypothetical support, we don’t know if there are differences in actual political participation and use of local customary institutions. 21 In this study I look at how the gender gap in participation differs between the formal and informal sectors of local governance. While we can expect that men and women will show different rates of participation in the formal and informal sectors, very little of the literature on participation in the formal sector can inform our expectations for how women will interact with the customary sector. I argue that women may experience discrimination when using the customary sector, which may affect which sector they choose when they need assistance at the local level. Writing on South Africa, Beall, Mkhize, and Vawda (2005) write that, “Women are particularly vulnerable under the traditional system, where they have curtailed rights, no access to communal resources outside their relationship with their father or husband and limited representation on traditional councils” (764). Thus, in situations where women have a choice in whom to seek for assistance, they may be less inclined to see the chief if they feel more vulnerable in that system or expect to be treated poorly compared to their expectations for the formal system. In many African countries, chiefs maintain some authority over land, which can be a large source of gender discrimination at the local level, as women are denied access according to the customary law administered by chiefs (Rangan and Gilmartin 2002, Tripp 2004). Inheritance is another area where women may face discrimination in the customary sector (Fonjong 2012). I argue that this type of gender discrimination by chiefs may discourage women from choosing the customary sector for assistance when the formal sector, beholden to the law, may treat them better when they have a problem. 2.5 Local/Non-Local Status In addition to gender, I argue that variation in people’s membership in their communities will affect whom they seek for assistance. Some community members have more status and power than other community members. While there is tremendous nuance in the dynamics across communities and cultures, one common distinction between community members is the degree to which someone “belongs” in the village. Often, those with “first-comer” claims hold more privilege in a village than those who descend from those who arrived later (Lentz 2013). In other words, anyone that can claim they descend from the founder of the village, “belongs” in the village more than those who came later. For instance, in Senegal, Wilfahrt (2018b) explains, "Local elites, such as village chiefs and notables, traditionally derive their authority from village 22 social hierarchies, for example, the village chief almost always descends from the village’s founder” (251). Thus, village chiefs use their claim of belonging to legitimize their rule today. Newcomers to a village tend to hold the least amount of status and power, with those coming from farther away holding less than those coming from villages nearby (Boone 2014). The main exception to this is marriage. From my time in Malian villages3, I observed that when women marry into a village, they are not considered outsiders because they are incorporated into existing households. Sometimes this divide falls along ethnic lines, but not always. For instance, people of the same ethnicity as the dominant ethnicity in the village may hold more power than those of minority ethnicities. However, in-migrants may be of the same ethnicity as the dominant ethnicity in the village and still be considered “outsiders” compared to those who have lived there longer. Thus, it’s important to not solely delineate these divides as ethnic, though this can be a useful proxy when other data are lacking. When newcomers move into a village, it creates a distinction that goes by many names in the literature. This can be referred to as an insider/outsider dynamic, a stranger/native dynamic, or a local/non-local dynamic (Honig 2017). In francophone Africa, the French words “allochtone/autochtone” are often used to express this same concept (Wilfahrt 2018a). Boone (2014) uses the terms ethnic insider and ethnic outsider. I follow the lead of Honig and use local/non-local to describe this type of status. By using local/non-local status instead of ethnic insider/outsider, I include in my theory those who may have moved to a village but who may share an ethnicity with the majority of the village. While local/non-local divides may often fall along ethnic lines, there are instances of migration that do not fall along ethnic lines and I wanted to use a term that included those scenarios, as well. Regardless of the exact words used, the idea that some community members hold more of a claim to a place than others can affect how different community members experience life in a given place. As Lentz (2013) writes, “Currently, invoking shared national citizenship, for instance, is often not enough to guarantee effective access to, and durable rights over, landed resources outside one’s “home” community” (166). Similarly, Honig (2017) finds that kin of customary leaders in Senegal and Zambia were more likely to use the customary land rights 3 I spent 2011-2012 and 2014 living in a small village in Mali, first as a Peace Corps volunteer and later on my own. 23 system, while those without shared kinship with the customary authority were more likely to hold state titles for their land. Thus, in terms of land use approaches, Honig finds differences between those with some of the deepest ties to a place compared to those who are less rooted or connected to the chief. Sometimes the families with the most privilege in the village will also have increased access to resources such as subsidized farming inputs and food aid, as well, as I observed in rural Mali. For this reason, I look at this dynamic and expect that non-locals will make different choices than locals when they need help with an issue. Those with more “belonging” to a village will be more likely to go to the chief for help, as the chief may bias their assistance towards those with more privilege in the village. As with gender, this could be through outright discrimination against non-locals by the chief or could be more of a perception among non-locals that they’re going to be treated worse going to the chief for assistance than to other local actors. Thus, I argue that non-locals with less local privilege will be more likely to opt for help from the formal system by going to a local elected official when they need assistance. 2.6 Issue Area Aside from identity, I expect that people will also vary in how they use local institutions depending on the type of issue they are having. Here, the same person might use a different local institution depending on the type of issue they are having. Obviously, people will want to go to the official who can best help them with their problem, assuming they don’t have any barriers to that person. Therefore, it may be more clear whom someone would go to for some issues over others. However, when responsibilities overlap between the formal and informal sector, it becomes less clear how these choices are made. People can choose their preferred venue or they can shop their issue between venues to see which garners them the best results. Hern (2022), for instance, describes the ways that women were able to take advantage of the ambiguities in systems of legal pluralism in colonial Senegal and Gold Coast to gain rights in inheritance and divorce. This shows that not only do people strongly consider which venue to use, but also that they can take advantage of systems with ambiguity. Below, I outline my theory around how issue type affects how people use local institutions. I describe three different types of issues that people commonly have at the local level in Africa. These include administrative tasks, conflict resolution, and land issues. 24 Administrative tasks are one type of issue someone may have that a local official could help with. For instance, when a family welcomes a new baby, they need to use local officials to get a birth certificate and register the birth with the state. Similar tasks include acquiring marriage and death certificates, titles, and other types of official documentation. We might expect that these types of issues would clearly be the purview of the formal sector of local government because they are tasks invented by the administrative state. Prior to the creation of the legal- rational state in West Africa, there was little need for such documentation. Even today, many children do not get registered with the state and face difficulties accessing health care and education as a result (Cappa et al. 2013). As Scott (1999) explains, “If you wish to have any standing in law, you must have a document that officials accept as evidence of citizenship, be that document a birth certificate, passport, or identity card. The categories used by state agents are not merely means to make their environment legible; they are an authoritative tune to which most of the population must dance” (83). Therefore, because these types of issues derive from the state, I argue that people will think first to go to local state officials, such as members of the local government, for help in acquiring these documents. On the other hand, conflict resolution has long been a responsibility of customary leaders and remains so today. According to Baldwin (2016), “Chiefs’ most important contemporary power is resolving disputes” (33). Winters and Conroy-Krutz (2021) conduct a survey experiment in rural Mali to see if people are more likely to prefer formal or customary judicial institutions to solve land dispute issues. They find that Malians report that justice provided by chiefs was expected to be fairer, quicker, and cheaper than using the formal system. This suggests that there are still strong preferences for using customary authorities for dispute resolution. Therefore, I argue that when given a choice, people will prefer to take their disputes to customary leaders over elected officials. Lastly, land issues are prominent at the local level across Africa. In many areas, the state holds formal authority over land tenure, including the titling, sale, and purchasing of land. However, chiefs often maintain a level of customary authority over land and access to resources, as well. Boone (2014) describes two types of land tenure regimes that are prominent across 25 Africa: statist land tenure regimes and neocustomary4 land tenure regimes. Statist regimes are those in which the central state manages access to and use of land, while neocustomary regimes allow the central state to devolve power over land access and use to traditional leaders (Boone 2014). In some places, these regime types are somewhat clearly delineated, as in Ghana, where traditional chiefs are given control over stool lands, while the state controls land elsewhere (Ubink 2008). In other places, such as Mali, the two regimes can overlap, where the state maintains official authority over land in economically productive areas like the southern cotton belt and the rice-growing regions along the Niger River, but chiefs also enjoy customary authority throughout those same regions (Camara 2013). Where the two regimes overlap, there can be multiple types of local officials with some power over land. This can give citizens a choice as to whom to go to for assistance with their land issues. However, it is a priori unclear in these situations which type of local leader people will favor. Honig (2017) looks at this specifically in regard to obtaining titles for land. She finds that those with closer ties to the chief in Senegal and Zambia were less likely to have titles and were more likely to rely on the customary system instead. This suggests that those who are less close to the chief will be less likely to use the customary sector for their land issues and may instead be more likely to use the formal sector. Thus, I argue that land issues are less clear cut than the other two issues. People may choose to use the formal sector due to their legal authority over the issue, or they may choose the informal sector due to the long-standing power over land. 2.7 Conclusion In this chapter, I developed my theory of how people use local institutions, especially when they have a choice between using the formal and informal sector. I outlined the continued importance of customary chiefs across much of Africa, as well as the introduction of decentralized local government officials since the third wave of democratization in the 1990s. These local institutions have coexisted now for decades, but our knowledge about how people make choices about which set of institutions to use, and for what purposes, remains incomplete. I argue that whether people use the formal or informal system will vary depending on their identity 4 Boone uses the term neocustomary to indicate that these informal land regimes are not direct legacies of pre- colonial land tenure regimes, but rather have been manipulated by the colonial and postcolonial state into what they are today. I use the term interchangeably with customary. 26 and the type of issue they have. For instance, gender minorities and those with non-local status may be more likely to avoid poor treatment in the customary system and may instead choose the formal system. I also expect the type of issue to matter, with administrative tasks to be more likely the purview of the formal sector, while conflict resolution issues will be more likely the responsibility of the chief. Land issues, on the other hand, are a good example of an issue with overlapping responsibilities by the formal and informal sectors, and thus a less clear distinction of how people will choose. In the next chapter, I will begin to test this theory empirically using Afrobarometer data from Senegal. Senegal is a good case to test this theory because of the overlap in responsibilities between the formal and informal sectors. I first detail Senegal’s local governance system, describing both the formal and informal sectors of local governance. I then test my theory looking specifically at variation in identity, examining both gender and local/non-local status. This initial test of my theory allows us to better understand how identity affects people’s choices when using local government in a country with significant overlap of responsibilities at the local level. 27 CHAPTER 3: IDENTITY AND USING FORMAL AND INFORMAL LOCAL INSTITUTIONS 3.1 Introduction In this first empirical chapter, I test my theory using Afrobarometer data from Senegal. The main focus of this chapter is to explore how different groups interact with the formal and informal systems of local governance. Specifically, I compare the frequency with which two different identity groups seek assistance from the chief and the local councilor. I focus on the differences between men and women, and between local and non-local individuals. Local individuals are those with long-standing ties to the community—often families that have lived in a given village or area for generations. Non-locals are more recent transplants to an area, often migrants, with more of an outsider status in the village. I expect women and individuals who are not local to the community to have a harder time getting assistance from the informal system than from the formal system, leading them to seek assistance from the formal system at higher numbers than they would if they could get the assistance they need from their customary leaders. This is important because it offers insight for leaders on one of the costs or benefits of incorporating customary systems into local governance. If there is increased discrimination among informal leaders than formal ones, then this is an area that can be addressed by improved decentralization policy. In this chapter, I will first delve into the history of customary leadership and decentralization in Senegal. This is necessary for understanding the system of local governance—both formal and informal—available to citizens for help with local issues. The customary system in Senegal, in which village chiefs have been integrated into the formal system, is particularly helpful in exploring this question. Instead of two separate silos of local governance that citizens must navigate, the village chiefs in Senegal have some overlapping responsibilities with local elected officials, which gives citizens a choice of who to contact for help. After discussing the Senegal case, I next discuss my theory and the hypotheses I will test, followed by a discussion of the data and methods I will use in my analysis. I then discuss my results, followed by a discussion of their implications. 28 3.2 Local Governance and Decentralization in Senegal Senegal is a good case to look at the ways in which people use formal and informal local institutions. Senegal has a long history of decentralization, dating back to the first Act of Decentralization of 1974, when the first local councils were created under President Senghor. Two subsequent rounds of decentralization reform further shifted power to these local councils in 1996 and, most recently, in 2014 (Beck 2008, Wilfahrt 2022). The three acts of decentralization outline the powers of officials at the local, departmental, and regional levels of government. The hierarchy of multilevel government in Senegal contains two parallel systems: the deconcentrated system and the decentralized system. Vengroff and Johnston (1989) write, “[Senegal’s] administrative system might be described as a hybrid of the colonial administration that predated it and the various reforms undertaken by the government to modify the colonial system since independence” (2). This was especially true at the time Vengroff and Johnston were writing this statement but remains largely true today. French colonies, including Senegal, were governed through highly centralized administrations, modeled largely on the French system itself (Wilfahrt 2022). The deconcentrated system transfers authority over some functions away from the central state to lower levels of government. However, the central government maintains strong oversight and control over these lower-level administrators (Vengroff and Johnston 1989). In Senegal, the deconcentrated system consists of officials mainly appointed by the president, including the regional governor, the prefet, the sous-prefet, and the village chief. These officials, with the exception of the village chief, often serve in areas away from their home region, which serves as a way to tie them closer to the central government than to the local population (Interview with secretary to the mayor, Kolda Region, 8/2/2019). The decentralized system, on the other hand, is a newer component to local government structure. It contains directly and indirectly elected officials, including members of the department council, the mayor, and members of the commune council. These officials are theoretically more accountable to citizens on the ground than to the central state, though this isn’t always true in practice. That said, elected local officials are not entirely autonomous. Instead, they operate within a system called tutelle, in which centrally appointed officials from the deconcentrated system oversee elected officials in the decentralized system (Ribot 2003). These 29 dynamics have evolved over time. As noted by a candidate for mayor in the Kolda region, “Mayors used to just be respondents. The sous-prefet gave the instructions to the mayor. Now there is free administration at the local level. The sous-prefet can’t tell the mayor what to do anymore. He can just review what the mayor has planned and verify that it’s legal. Before, it was the sous-prefet and the prefet that decided everything” (Interview with mayoral candidate, Kolda Region, 11/21/2021). The state delegates certain decisions to the various levels of the decentralized system. For example, communes, at the bottom of the decentralized system, have the power to make some decisions pertaining to local issues such as primary education, health, or agriculture. However, elected officials in the decentralized system must seek approval from their superiors in the deconcentrated system on most major decisions (Wilfahrt 2022). According to a local councilor, “The system is a line…The sous-prefet does bureaucratic paperwork and operates at the direction of the central government. We [the local council] are the representatives of the population. But we can’t do anything without telling the sous-prefet. If you go against the sous-prefet, you go against order” (Interview with local councilor, Tambacounda Region, 7/29/2019). On paper, this ensures that the sous-prefet double checks all the decisions made by the local council to ensure every decision is in line with the law (Diop 2006). That said, this also allows the state to keep tabs on officials elected to the various decentralized bodies. Tutelle is also a way for a state to broadcast its authority throughout the country. Table 3.1. Deconcentrated and Decentralized Systems of Governance in Senegal Level of Administrative Unit National National Region Department Arrondissement Commune Commune Village Deconcentrated System President Minister Governor Prefet Sous-prefet Mayor Local councilor Village chief Regional council Departmental council Decentralized System President In this chapter, I look specifically at the lowest rungs of the decentralized and deconcentrated systems: local councilors and village chiefs. However, to understand the political context in which this lowest level operates, I’ll briefly explain the full system of multilevel government in Senegal. 30 The national government stands at the top of both systems, with the president and then the National Assembly in charge overall. Then, positions in both systems are broken down into ever-smaller geographical areas. The largest geographical unit is the region, of which there are fourteen in Senegal. At the regional level, the deconcentrated system is run by a governor, who is appointed by the president. This is the president’s representative at the regional level. The regional level currently has no representation in the decentralized system.5 Each region has three-to-five departments, and each department contains a prefet, or prefect, in the deconcentrated system and a department council in the decentralized system. The department council is indirectly elected by local councilors, while the prefet is appointed by the president. The department council is charged with making decisions in several policy areas, including secondary education, health care, and economic development within the department. Within each department are arrondissements, which have no representation in the decentralized system, but are governed by the sous-prefet, or sub-prefect, in the deconcentrated system. Like governors and prefets, sous-prefets are appointed by the president. Sous-prefets provide some administrative services to citizens, though fewer than in the past. Sous-prefets approve the budget and other initiatives voted on by local councils. A secretary to the sous-prefet told me, “A lot of our work is writing reports. We have weekly, monthly, and quarterly reports, which we give to the prefet” (Interview with a secretary to the sous-prefet, Tambacounda Region, 7/29/2019). Sous-prefets also supervise village chiefs. In practice, this supervision can vary quite a bit. Some chiefs reported frequent contact with the sous-prefet, while others had very little. One chief told me, “The sous-prefet never shows up. We hear if there’s a meeting, but he never comes here. We don’t know the role of the sous-prefet. We barely go to the sous-prefet if we can’t fix a problem. We go first the mayor or the gendarme” (Interview with village chief, Kaffrine Region, 5 The decentralized system had a regional council until Act III of Decentralization was enacted in 2014, at which point the duties of the regional councils were transferred to the department level. Act III of Decentralization sought to bring government decisions closer to the people, and the shift from a regional council to a departmental council was one of the main ways that was achieved (Interview with official at the Agence Regionale de Developpement, Tambacounda Region, 11/17/2021). 31 8/7/2019). Another chief said, “Most of the sous-prefets, they spend all their service not stepping into the villages” (Interview with village chief, Kolda Region, 7/31/2019). Within each arrondissement are two-to-four communes. In the decentralized system, each commune is governed by a local council comprised of elected councilors. Until 2022, councilors then voted among themselves on which of them would serve as mayor. Starting in the local elections of January 2022, the mayor is now directly elected as the top of the party list who receives the most votes (Interview with Director of the Office of the Mayor, Kolda Region, 11/25/2021). The mayor and the local council make up the bottom of the decentralized system. There is no representation in the deconcentrated system on the commune level. In rural areas, each commune is made up of several villages, each led by a village chief. The number of villages per commune varies substantially across the country. The village chief is the bottom of the deconcentrated track and is supervised by the sous-prefet above them. Since there are several villages in each arrondissement, each sous-prefet oversees several village chiefs.6 While the two systems have separate responsibilities, those in the deconcentrated system often oversee and approve of the work done at the comparable level in the decentralized system. For instance, while the mayor and the local council draft and vote on their own budget, it is then approved by the sous-prefet before being passed up the chain of the deconcentrated system for approval at all levels. In this way, presidential appointees have some capacity to oversee elected officials at all levels. Mayors and their councils are clearly part of the formal system, while village chiefs hold a more ambiguous role. Since precolonial times, most Senegalese villages have had village chiefs. Usually, chiefs are descendants of the founder of the village (Wilfahrt 2022). Under French colonial rule, even villages that were traditionally egalitarian were assigned chiefs 6 In urban areas, the commune generally makes up the whole or the majority of the urban area. Urban areas also have chiefs, though they are called neighborhood chiefs, not village chiefs. These neighborhood chiefs are generally appointed by mayors and do not hold the customary authority or legitimacy of their rural counterparts (Interview with village chief, Kolda Region, 8/3/2019). 32 (Boone 2003). The role of village chief was institutionalized under the French and continues in a similar capacity today. Chiefs serve many functions at the village level. As one chief described their role, “Chiefs are the liaison between the government and the population” (Interview with village chief, Tambacounda Region, 7/27/2019). In other words, chiefs are advocates for their village at the mayor’s and sous-prefet’s offices, while also keeping the village informed when government officials have information to pass on to the village. They respond to all village-level conflict, attempting to solve all problems before the police or the gendarmes become involved. A village chief said, “Chiefs are the ones awakened in the night, bothered every time there’s a little problem” (Interview with village chief, Tambacounda Region, 7/28/2019). Chiefs also collect annual taxes and bring them to the Treasury, host visitors to the village, and attend meetings called by the mayor or the sous-prefet. There is no position above the village chief in the system of chieftaincy (Interview with former local counilor, Kolda Region, 8/1/2019). Mayors and their councils collect certain local taxes, prepare annual budgets, maintain public records and certificates of births, deaths, and marriages, and provide services and programming in fourteen competency areas.7 They also work with the sous-prefets, village leaders (including chiefs), as well as development agents, NGOs, and other civil service agents to implement the policies within these domains (Interview with secretary to the mayor, Kolda Region, 11/26/2021). The duties and responsibilities of chiefs and mayors overlap. For instance, it is the responsibility of the mayor’s office to issue birth certificates for infants and children. This means that villagers can go to the mayor’s office after the birth of a child and receive a birth certificate. However, chiefs may also gather the information on births in the village over a given period of time and go to the mayor’s office to get the birth certificates on behalf of villagers with new children. This means that often times, villagers can choose whether they want to go to the mayor’s office themselves to get a birth certificate or if they want to have the chief do it on their 7 The fourteen competencies include: education; health; environment and natural resources; youth, sports, and leisure; culture; housing and urban development; agriculture and livestock production; fishing; tourism; water resources; land management; sanitation; professional development; and planning (Wilfahrt 2022). 33 behalf. There is overlap of responsibilities in other areas, as well, such as land tenure and usage, tax collection, conflict resolution, and service provision. Due to this overlap in responsibilities between the formal and informal sectors, people are presented with a choice between local government structures when seeking help with an issue. How do they decide whom to go to? Do some groups prefer informal leaders, while others prefer formal? While much work examines the relationship between chiefs and elected local government officials, little is known about how such a system affects local residents’ engagement with local governance, particularly among groups such as women and minority ethnicities. In the next section, I will outline my theory and expectations, followed by a discussion of my methodology and results. 3.3 Theory and Hypotheses In Senegal, there is overlap between the formal and informal actors people can go to for assistance with local issues. For instance, both chiefs and mayors’ offices register births. In an interview with a teacher and local activist, I was told, “Chiefs are implicated in governing. Chiefs write the birth certificates and that helps the mayor” (Interview with teacher and local activist, Tambacounda region, 11/17/2021). A village chief mentioned, however, that, “some people go straight to the mayor to get birth certificates instead of coming [to me] … The chief is supposed to sign, but if I don’t, that’s ok, because it lightens my load” (Interview with village chief, Kolda Region, 8/4/2019). Thus, for certain issues, people can choose whom to go to for help. How these decisions are made and how they vary across types of people, though, is largely unknown. Do some groups of people choose the chief more often, while others opt for local elected officials instead? Does gender or one’s local or non-local status affect the type of local actor they seek? Answers to these questions will help us understand how people operate within these hybrid formal/informal local governing systems. As decentralization efforts evolve over time, this knowledge can be useful in shaping reforms to increase the access to and effectiveness of local governance, especially in rural West Africa. 34 While the mayor and local councilors have more legal power to sort out local issues, village chiefs still garner high levels of trust and legitimacy among the population (Logan 2009). Yet, customary institutions operate on customary norms and are not necessarily held to the same standards as formal systems regarding equity (Lentz 2013, Fonjong et al 2013). Women can face discrimination in the customary system, especially regarding land issues, inheritance, and conflict resolution (Tripp 2004, Winters and Conroy-Krutz 2021). Chiefs may be explicitly or implicitly favoring men. This could be simply because the chieftaincy is a patriarchal institution in Senegal, in which chieftaincy is traditionally passed from father to son, from uncle to nephew, or among brothers. Because chiefs are men, they may hold some implicit bias favoring other men who come seeking their assistance. Men have also traditionally held responsibilities over managing issues that arise in the household, meaning there may remain more expectations for men to seek out traditional authorities for assistance than women. Women might also feel more comfortable using the formal sector because current law requires gender quotas for all elected institutions (Law 2010-11 Act of May 28, 2010). This means that there should be an equal number of male and female local councilors, and female villagers may feel more comfortable bringing their issues to female councilors in the formal system than using the customary system. Similarly, in many areas of West Africa, there are insider/outsider divides at the local level. These divides center around who has the most claim to belonging in the community. Generally, those who have lived (or whose families have lived) in the community the longest time are considered to belong there the most. This is the same idea around the descendants of the founder of the village holding the most claim to the chieftaincy. This idea is sometimes discussed as the autochtone/allochton divide (Geschiere 2009), or more simply the local/non-local divide (Honig 2017). Because the chieftaincy is a system rooted in belonging, I expect there to be some favoritism among chiefs towards other local residents: those whose families have long-standing ties to the village. Those who have come later (non-locals) may get less favorable treatment from the chief, leading those families to seek help elsewhere. In addition, even in the absence of outright discrimination, if women or non-locals perceive favoritism or inequality of access among chiefs, they may modify their behavior when they have an issue. In other words, a given village may have a chief who seeks to treat everyone equally, but if non-locals expect the chief to treat them differently, then they may operate as if the chief is biased, even in the absence of 35 outright discrimination in that village. Thus, I expect women and non-locals to have different patterns of visiting formal and informal institutions compared to men and locals. Examining how different groups interact with the local government system in Senegal is important because it gives insight into the effectiveness of this hybrid governance system. As mayors and local councils continue to gain power, it’s vital to know who is able to access these services. Moreover, if local formal officials are providing more equitable services to the community, this is a boon for the success of liberal democracy at the local level. On the other hand, as chiefs continue to maintain their power at the village level, it’s important to know the extent to which they’re serving the community. If the customary system is discriminating against certain groups, it’s possible that local governing systems can be reformed to minimize this type of discrimination at the local level. To explore this issue, I make one important assumption at the outset: people generally will prefer to seek assistance from the chief compared to other local government actors in the formal system. Village chiefs have been the village-level authority since the pre-colonial era. While the institution has evolved over time, chiefs still maintain strong roots in their communities. People are accustomed to seeking out their chief for assistance. One chief said, “When there’s a conflict, the chief can come talk to the mayor, or they can call the councilor and they’ll try and solve the problem together. Sometimes when there’s a problem people go straight to the mayor, but they should go to the village chief first” (Interview with village chief, Kolda Region, 8/2/2019). Chiefs are also the local official that is closest to the population because each village has a chief, while formal officials serve at the commune level, which covers several villages. Likewise, because chiefs generally serve for life, they are replaced less frequently than local councilors or mayors, who serve five-year terms and may or may not be reelected. This gives chiefs more time to develop strong relationships across the community (Baldwin 2016). In other words, chiefs maintain high levels of legitimacy in their villages (Logan 2013). Chiefs may also be favored due to shortcomings in the formal sector. In Senegal, the formal system of local governance has been in place since the 1970s with the first Act of Decentralization, but the powers given to this sector have often been limited by centralized control and have changed and evolved over time (Poteete and Ribot 2011). Therefore, people 36 may be less clear about what can get done through the mayor compared to the chief. Additionally, some mayors are absent or not seen as embedded community members. For instance, a village chief in the Kaffrine region described the local mayor by saying, “He is Dakarois and has a different mentality than the people here” (Interview with village chief, Kaffrine region, 8/7/2019). A secretary to the mayor in Kolda region told me that the adjoint mayor was in charge because the elected mayor worked at the consulate in Guinea Bissau (Interview with secretary to the mayor, Kolda Region, 8/2/2019). Another mayor in Kolda region was a professor at a university on the other side of Senegal. He made a point of coming to the mayor’s office a few times a year (Interview with mayor, Kolda Region, 8/3/2019). When mayors are absent or seen as higher class than local villagers, people may feel more comfortable going to the local village chief, all else equal. For all these reasons, I expect all groups to be inclined to seek help from their chief before they ask other local officials for assistance. However, if people don’t expect to get the help they need from their chief, of if they feel that their chances for success might actually be higher with another set of actors, they may be more willing to ask for help elsewhere, particularly within the formal system. This type of venue shopping is common in situations of legal pluralism, in which people can choose whether to take their case to the courts or have it solved using customary law (Hern 2022). I argue that at the local level in Senegal, people can choose to bring their issue to the chief or to the formal system. People may also choose to contact both types of officials until they get the help they are looking for. I further argue that the way people go about solving their issue between the formal and informal sectors may be a function of their identity. For instance, if non-locals don’t expect to get the help they need from the customary sector, they may be more inclined to use the formal system than others. Thus, identity types will likely affect the decisions people make on who to seek for assistance. While gender may affect the type of actor one chooses to contact, women will tend to seek assistance from any local actors (formal or informal) less often than men. Culturally, men are more likely to take care of administrative household issues—the types of issues one might visit a local official for—than women. These issues could include seeking official documents like birth, marriage, and death certificates; land titles; or local funding for a community project, among other things. Women also tend to have higher workloads in the household and fewer 37 resources, all of which might make it harder for her to contact any type of official when she has a problem. That all said, while men may be more likely to seek assistance from local officials, there are still plenty of reasons a woman may visit these officials. The main difference I expect to find between men and women is that the gap for women between seeking help from formal and informal actors will differ from the gap for men. All else equal, I expect men and women to both prefer seeking assistance from the local chief over local officials in the formal sector, for the reasons previously discussed. To be clear, I do not expect that one group will prefer the formal while the other will prefer the informal. However, if the customary sector discriminates on the basis of gender more easily or frequently than the formal sector and if the formal sector is seen as more equitable, then women may be more likely to seek out help from the formal sector than they otherwise would be if the customary authorities were welcoming to them. Therefore, my first hypothesis states: H1: The gap for women between visiting the village chief and the local councilor will be smaller than the gap for men. I expect a somewhat similar dynamic to operate for those who are not local to the village in which they live, though with some differences. I do not believe that non-locals will be less likely than locals to contact any official. The dynamic between locals and non-locals does not likely yield the same cultural, time, and resource constraints that keep women from contacting officials. However, those who are native-born members of the village will likely have an easier time obtaining assistance from the village chief compared to those who are newer to the village. Those who are not native to the village will still likely seek help from the chief, but may need to seek help from other sources if they face difficulties in the customary sector. Thus, my second hypothesis states: H2: The gap for non-local villagers between visiting the village chief and the local councilor will be smaller than the gap for local villagers. I discuss the variables and methods I use to test these hypotheses below. 38 3.4 Methods and Data I use Afrobarometer data to explore the extent to which different groups contact the chief, the councilor, or both. I use Senegal rounds 3, 4, 6, 7, and 8 for this analysis. Round 1 was not conducted in Senegal, while rounds 2 and 5 did not include the contact variable either for traditional leaders or local councilors or both. Afrobarometer asks subjects, “Over the past year, have you contacted any of the following persons about some important problem or to give them your views?” They ask this question for a series of political actors, from Members of Parliament to political party officials to traditional and religious leaders. The possible responses include “never,” “only once,” “a few times,” and “often.” For this analysis, I use the question from this series about the frequency of contacting local councilors and the one about the frequency of contacting traditional leaders. These variables serve as my dependent variables. My independent variables look at two attributes that may affect who goes to see the chief and who sees the local councilor. The first is gender; thus, I create a binary variable for female and male. The second attribute attempts to capture whether someone is a local or non-local member of the community. The idea here is that outsiders are less likely to feel comfortable using traditional institutions to solve their problems compared to local formal institutions. Without ample questions in the Afrobarometer questionnaire to measure whether someone is local or non- local to their community, I use the Ethnic Power Relations (EPR) dataset (Wucherpfennig et al. 2011) to code each geographic area with one or more dominant ethnicities. The dominant ethnicities included in the EPR data for Senegal include Wolof, Pulaar, Serer, Mande, and Diola. In general, a given geographic area is coded as having one dominant ethnicity; however, in some areas, there are two. I then use the geocoded Afrobarometer data to place each observation within the EPR geographic areas and code based on the subject’s reported ethnicity whether or not they are a majority or minority ethnicity in their area. There are some disadvantages to this approach, namely that the EPR data present a very broad measure, lacking some of the nuance on the ground. From this measure, we cannot tell if a given respondent who is coded as a minority ethnicity lives in a village made up of members of 39 the majority ethnicity of the region (making them a minority in their village), or if they live in a village consisting primarily of others of their same ethnicity (thus, a member of the village majority, but a minority in the wider area). Either could be true across Senegal, where minority ethnicities commonly but not always live in villages with those who share their ethnicity. For example, Pulaar people who live in a predominantly Wolof area may live in a primarily Pulaar village. However, there could also be some Pulaar families in an otherwise Wolof village. We also don’t know from this measure if the respondent is the same ethnicity as their village chief. Such instances may make it more likely for someone to feel comfortable going to the chief, even if they’re not part of a dominant ethnicity in the broader area. However, with the available data, the minority ethnicity measure is the best possible available indicator of how local/non-local status may matter in how people choose an institution with which to solve their problems. The following control variables are included in the analysis, as well. Whether the subject is in a rural or urban area is expected to have a big impact on whether they go to the chief for assistance or the local councilor. In urban areas, there is not just one chief of the whole city, but rather several neighborhood chiefs who control smaller areas and lack the legitimacy of rural village chiefs (Interview with secretary of the mayor, Kolda Region, 8/2/19). Thus, I include the binary urban rural variable as a control. Trust in political leaders and parties is also expected to impact who people might go to see when they have a problem. As the local formal institutions are filled with elected officials who affiliate with a political party, community members who support other parties may be less likely to use the formal system for assistance compared to going to local chiefs, who are not allowed to officially affiliate with a political party. Chiefs are largely seen as not being partisan actors. In the 2020 Afrobarometer survey, 75% of respondents in Senegal agreed with the statement, “Traditional leaders should stay out of politics and leave people to make their own decisions about how to vote.” Across many interviews in 2019, I was told both by chiefs and other local officials that chiefs’ roles are administrative and apolitical; therefore, they are not members of political parties. While not always adhered to—one chief told me, “This doesn’t apply to me. I do politics” (Interview with village chief, Kolda Region, 8/4/2019)—this sense that chiefs are less political than elected officials tends to hold. 40 To control for this disparity between chiefs and local officials, I control for trust in political parties.8 Afrobarometer asks a series of questions about trust in various actors, including the President, the Parliament, ruling and opposition parties, police, and other various leaders and institutions. The question asks, “How much do you trust the following, or haven’t you heard enough about them to say?” I use the questions specifically on trust in the ruling party and trust in the opposition party to create a new variable for overall trust in parties. The response options for these two questions include, “not at all,” “just a little,” “somewhat,” and “a lot.” Those who responded for both the question on the ruling party and the opposition parties that they trusted them “not at all” or “just a little” were coded as trusting neither; those who responded that they trusted the ruling party “somewhat” or “a lot,” but trusted the opposition parties “not at all” or “just a little” were coded as trusting the ruling party; those who responded that they trusted the opposition parties “somewhat” or “a lot,” but trusted the ruling party “not at all” or “just a little” were coded as trusting the opposition parties; and finally, those who responded that they trusted both the ruling party and the opposition parties “somewhat” or “a lot” were coded as trusting both. Because the majority of local councilors align with the ruling party, I expect voters who have low trust in the ruling party to be less likely to go to the councilor for assistance. While this variable doesn’t capture whether a given person’s party affiliation aligns with the affiliation of their local council, it does come close because of the ruling party dominance among local councils in Senegal. In opposition strongholds, particularly in Casamance, this variable will not capture party alignment quite as well. I also control for education, lived poverty,9 and age. To examine whether women or minority ethnicities have different patterns of contacting formal and informal officials, I use a difference-in-differences approach to examine the gaps between the frequency of contacting the chief and contacting the councilor for women and men. I 8 I use trust in ruling party and opposition instead of the variable that asks about party affiliation because Senegal’s ruling party is often in coalition with several smaller parties. This coalition changes frequently and would be quite difficult to code accurately. 9 The lived poverty variable is constructed by creating an index of five questions on the Afrobarometer survey. These questions ask, “Over the past year, how often, if ever, have you or anyone in your family: Gone without [enough food to eat/enough clean water for home use/medicines or medical treatment/enough fuel to cook your food/a cash income]?” The possible responses include, “Never,” “Just once or twice,” “Several times,” “Many times,” and “Always,” which I coded to range from 0 to 4. To create the index variable, I added the responses from each of the five variables together. The possible values for the index variable range from 0 (a response of “Never” on all five questions) to 20 (a response of “Always” on all five questions). 41 repeat this for minority vs. majority ethnicities. I include the control variables listed above and use region and round fixed effects in my analysis. The difference-in-differences approach is necessary because simply comparing the rates at which women and men contact different leaders, for example, will not show the nuance of whether women are showing a different pattern in whom they seek out for help. This is because women will likely seek out both the chief or the councilor at much lower rates than men. However, by using difference in differences to compare the differences between the rates that men and women seek formal or informal actors for help, we can see whether women compensate for discrimination in the customary sector by seeking help from formal actors instead. I do not expect women to contact their local councilors more than men do; men are likely to contact any official at higher rates than women. Neither do I expect women to contact their local councilors at higher rates than they contact chiefs; again, I expect everyone’s first preference to be contacting the chief. However, I expect the gap for women between contacting the chief and contacting the councilor to be smaller than the gap between men. I use the same approach for looking at the relationship between having a majority or minority ethnic identity and seeking help from formal or informal actors. 3.5 Results Most people in Senegal contact neither the chief nor the local councilor. In fact, only 23% report having contacted their local councilor at least once in the past year, while only 31% report having contacted their village chief. It’s likely that most people do not have issues in a given year that local authorities are required for solving. It’s also possible that some people go to other authorities for assistance, such as a religious leader or a sous-prefet. Afrobarometer does not ask specifically about contacting sous-prefets, but fewer than ten percent report contacting an official of a government agency or their member of the National Assembly. People report contacting a religious leader in much higher percentages than any of these other actors, with 44% reporting contact in the past year. However, it’s unclear if the issues brought to religious leaders are comparable to those they would take to a local government official such as a councilor or a chief. Some people who contact their religious leader may bring issues that could also be addressed by other local officials, but because religious leaders do not have any specific power to handle local administrative problems, it’s unlikely that this is the bulk of what is brought to them. 42 When looking specifically at women, even fewer have contacted a local councilor (18%) or a village chief (26%), compared to men contacting a councilor (28%) or a chief (36%). There are fewer differences overall when looking at minority and majority ethnic groups. For minority ethnic groups, 21% have contacted a local councilor and 27% have contacted a chief, compared to majority ethnic groups, of whom 23% have contacted a local councilor and 33% have contacted a chief. Figure 3.1.a. Frequency of contacting the councilor and contacting the chief, by gender 9.3 11.3 5.7 9.0 11.5 7.3 73.7 72.2 4.4 7.3 5.9 82.4 15.9 15.0 5.6 63.6 100% 80% 60% 40% 20% 0% Male Female Male Female Contacting chief Contacting councilor Never Only once A few times Often While seeking assistance from local actors is not necessarily a common occurrence, there are still some important differences in the frequency with which people do contact their local officials. Again, the variables ask subjects how often they have visited a local councilor or a chief in the past year, with the options of never, only once, a few times, and often. These values range from 0 for “never” to 3 for “often.” The following two figures show the frequencies with which people contact their chief and their councilor, by identity group. For contacting the chief, Figure 3.1.a shows that men are more likely to contact the chief than women. Among men, 64% never contact the chief, while 74% of women never do. Six percent of both men and women report contacting the chief only once, while 15% of men contact the chief a few times, compared to only 11% of women. Sixteen percent of men contact the chief often, compared to only nine percent of women. 43 The difference in contacting the councilor is similar to that of the chief. Figure 3.1.a shows that men are also more likely than women to contact the councilor. For men, 72% never contacted the councilor, compared to 82% for women. Seven percent of men contacted the councilor only once, while 6% of women did the same. The differences grow, however, with increased contact. Twelve percent of men contacted the councilor a few times, compared to only seven percent of women, and nine percent of men contacted the councilor often, compared to only four percent of women. Figure 3.1.b. Frequency of contacting the councilor and contacting the chief, by ethic status 14.3 12.7 5.7 67.3 11.5 10.4 5.2 73.0 7.5 8.4 6.7 6.0 9.2 5.8 77.4 79.0 100% 80% 60% 40% 20% 0% Majority ethnicity Minority ethnicity Majority ethnicity Minority ethnicity Contacting chief Contacting councilor Never Only once A few times Often The differences in frequency of contacting the chief between majority and minority ethnicities are narrower than the differences between men and women. As Figure 3.1.b shows, for majority ethnicities, 67% have never contacted the chief, compared to 73% for minority ethnicities. Six percent of majority ethnicities have contacted the chief only once, nearly the same as the five percent of minority ethnicities that have done so. Thirteen percent of majority ethnicities have contacted the chief a few times, compared to ten percent of minority ethnicities, while 14% of majority ethnicities contacted the chief often, compared to only 11% of minority ethnicities. Again, while there is some variation between majority and minority ethnicities, it is smaller than that seen between men and women. The differences between majority and minority ethnicities in contacting the councilor are even smaller than they were for contacting the chief. For majority ethnicities, 77% report never 44 contacting the councilor, compared to 79% of minority ethnicities. Seven percent of majority ethnicities contacted the councilor only once, compared to six percent of minority ethnicities. In fact, slightly more minority ethnicities contacted the councilor a few times at nine percent, compared to eight percent for majority ethnicities. Finally, seven percent of majority ethnicities contacted the councilor often, while six percent of minority ethnicities did the same. The next obvious question is are these differences significant? The following tables show the means of these two variables for men and women and for majority and minority ethnic groups. Figure 3.2.a looks specifically at contacting the chief. Men are indeed significantly more likely than women to contact the chief, as are majority ethnicities. Figure 3.2.a. Mean values for contacting the chief (0-4 scale), by gender and majority/minority ethnicity group (with 95% confidence intervals) 0.9 0.8 0.7 0.6 0.5 0.4 Male Female Majority ethnicity Minority ethnicity Figure 3.2.b shows that men are also significantly more likely than women to see the councilor. However, there is no difference between majority and minority ethnic groups contacting the councilor. This suggests that there is some degree of leveling happening in the formal system compared to the informal system for ethnic minorities. In other words, while minority ethnicities seem to be just as comfortable as majority ethnicities to use the formal system, they are less comfortable with the informal system. It’s likely that some people from minority ethnicities who would have gone to the chief have instead chosen to go to the councilor. 45 Figure 3.2.b. Mean values for contacting the councilor (0-4 scale), by gender and majority/minority ethnicity group (with 95% confidence intervals) 0.7 0.6 0.5 0.4 0.3 0.2 Male Female Majority ethnicity Minority ethnicity These results, however, are limited. Without controlling for other potential confounders, it’s hard to say if these differences hold up. To further explore this relationship, I next run an ordinal logit for each dependent variable: contacting chief and contacting councilor. I include dummy variables for female and minority ethnicity for my independent variables, along with the following control variables: urban or rural location, trust in the ruling party and opposition, education level, lived poverty, and age. I also include round and region fixed effects. This will show whether the previous results still hold after controlling for other possible causes. As seen in Table 3.2, the first model looks at the frequency of contacting the chief. I find that women are significantly less likely to contact the chief as frequently as men. Similarly, minority ethnicities are significantly less likely to contact the chief as frequently as majority ethnicities. This is in line with my expectations. The second model looks at the frequency of contacting the councilor. The variables in this model are all the same as the first model except for the dependent variable. Similar to the previous model, women are significantly less likely to contact the councilor than men. However, there is no difference in the frequency of contacting the councilor for minority and majority ethnicities. This is informative in that it tells us that even with controls, there is a difference between majority and minority ethnic groups and their decisions to contact the chief or councilor. 46 However, this still doesn’t address my main question, which looks at whether there is a difference between how different groups interact with the formal and informal system. Just knowing that women are less likely than men to frequent either system, for instance, doesn’t tell us if the proportion of people going to the councilor compared to the chief is higher for women than men. Table 3.2. Effect of identity type on contacting chiefs and councilors (1) Contacting chief (2) Contacting councilor Female Minority ethnicity Rural Trust Trust opposition Trust ruling party Trust both Education Primary Secondary Post-secondary Lived poverty Age Observations -0.569*** (0.062) -0.147** (0.066) 0.647*** (0.081) 0.196** (0.092) 0.442*** (0.086) 0.220*** (0.080) -0.016 (0.084) -0.161* (0.086) 0.099 (0.118) 0.024*** (0.007) 0.001** (0.001) 5,245 -0.523*** (0.068) -0.064 (0.070) 0.364*** (0.088) 0.188* (0.099) 0.382*** (0.093) 0.230*** (0.088) -0.007 (0.095) 0.221** (0.093) 0.834*** (0.119) 0.024*** (0.007) 0.002*** (0.001) 5,338 Note: Standard errors in parentheses. Includes region and round fixed effects. For the trust variable, the reference category is “trust neither;” for the education variable, the reference category is “no formal education.” *** p<0.01, ** p<0.05, * p<0.1. If the customary system discriminates against women or minorities in ways that the formal system doesn’t, then women and minorities might be more likely to use the formal system than if the customary system was more welcoming to them. To explore this and address my two hypotheses, I use a difference-in-differences test using OLS to see whether the gap for men in 47 seeing the chief or councilor is different from the gap for women, the results of which are shown in Table 3.3, Model 1. Table 3.3. Difference-in-differences estimates, by gender and minority ethnicity Chief or Councilor Female Chief or councilor x Female Minority ethnicity Chief or councilor x Minority ethnicity Rural Trust Trust opposition Trust ruling party Trust both Education Primary Secondary Post-secondary Lived poverty Age Constant Observations R-squared (1) Gender 0.260*** (0.027) -0.212*** (0.027) -0.049 (0.039) -0.042** (0.021) 0.237*** (0.025) 0.077*** (0.029) 0.187*** (0.028) 0.091*** (0.025) 0.014 (0.027) 0.008 (0.027) 0.182*** (0.037) 0.010*** (0.002) 0.001*** (0.000) 0.148*** (0.046) 10,583 0.079 (2) Minority ethnicity 0.281*** (0.026) -0.237*** (0.020) 0.010 (0.028) -0.105*** (0.039) 0.237*** (0.025) 0.077*** (0.029) 0.187*** (0.028) 0.091*** (0.025) 0.014 (0.027) 0.008 (0.027) 0.182*** (0.037) 0.010*** (0.002) 0.001*** (0.000) 0.137*** (0.046) 10,583 0.079 Note: Standard errors in parentheses. Includes region and round fixed effects. For the trust variable, the reference category is “trust neither;” for the education variable, the reference category is “no formal education.” *** p<0.01, ** p<0.05, * p<0.1. 48 I then use the same approach for minority and majority ethnicities, as shown in Table 3.3, Model 2. The difference-in-differences estimates are shown by interacting the frequency of contacting either the chief or the councilor with either gender or minority ethnicity. The advantage to this approach is that, for example, it lets us see if, despite the lower frequency of contact for women, they are more likely to seek assistance from the councilor instead of the chief, compared to men. All the aforementioned controls are also included in these tests. For H1, I find no difference for men and women in their respective gaps between going to the chief compared to the councilor. While men are much more likely than women to seek assistance from either the chief or the councilor, the gap for men in seeing chiefs as opposed to councilors is statistically indistinguishable from the gap for women in seeing chiefs versus councilors. This does not align with my expectations in H1, in which I expected the gap for women to be smaller than for men. This suggests that women are not in fact using the formal system to avoid using the informal system. This is an important finding because gender discrimination in the customary system is commonly discussed, but it does not seem to affect women’s choices of who to seek when they need assistance. Women do not appear to be substituting the councilor for their chief when trying to get their needs met. This doesn’t mean that there are no differences across genders, but rather that the customary sector doesn’t seem to be uniquely less-accessed by women, compared to the formal sector. For H2, I do find a difference for minority and majority ethnic groups in their respective gaps between seeing the chief and seeing the councilor. While both majority and minority ethnic groups seek assistance from the chief more frequently than the councilor, the gap in frequency of seeing the chief and seeing the councilor is smaller for minority ethnic groups than majority ones. This means that minority ethnic groups use the formal system proportionally more than majority ethnicities do. This lends support for H2, in that minority ethnicities are going to the councilor in slightly higher proportions than majority ethnicities are. Increased discrimination against outsiders/non-natives in the customary system might explain this difference. 3.6 Discussion The results of my analysis are mixed. While women seek assistance from both formal and informal leaders at much lower rates than men, there is no indication that when they do seek 49 assistance, their preference for one institution over another is significantly different from men’s. Why would this be? One possible answer is that there’s not actually discrimination in the customary system. In fact, according to Afrobarometer data, in 2018, 92% of Senegalese women reported never having experienced discrimination or harassment based on their gender in the past year. With a result this one-sided, one might think this clearly explains the similar choices among men and women in choosing the chief or the councilor. However, survey questions about discrimination can be affected by social desirability bias, in which subjects give the answer they think the interviewer wants to hear. In fact, social desirability bias has been found to be stronger in in- person interviews (such as Afrobarometer) compared to web interviews (Liu 2017). Thus, it’s important to explore other explanations here, as well. Another possibility is that there is a similar level of discrimination in both the formal and informal sectors of local governance. This would explain why women are less likely to see either the chief or the councilor than men, but don’t appear to go to the councilor at higher rates. I don’t have a direct way to test this. However, according to Afrobarometer, women are less likely than men to say that the government is doing a good job promoting equal rights and opportunities for women: 70% of women say the government is doing fairly well or well, compared to 77% for men. When comparing the means of this question, the rate for women (2.682) is statistically lower than that for men (2.857) (t=3.937, p = 0.000). This suggests that women do believe there is room for improvement on the government’s end at working towards women’s equality. It’s possible that councilors do discriminate less than chiefs, but that women don’t have the same resources to get to the councilor, who may reside in a different village. I do attempt to address this issue by controlling for lived poverty, but that measure is limited. The lived poverty measure looks at cash income, access to food, water, cooking fuel, and medical treatment. However, women may be hampered by more specific resource imbalances, such as lack of access to transportation, or biases towards men using the family moto, so I explore this issue in more detail here. In Afrobarometer rounds conducted in 2018 and 2020, women are less likely to report owning a motorcycle or motor vehicle than men. While owning a vehicle is still fairly rare (12% of men own their own vehicle, compared to 6% of women), men are still twice as likely as women to own their own vehicle. When comparing means for this question, the rate for women 50 (0.218) is significantly lower than the rate for men (0.335) (t = 7.336, p = 0.000). This means that women are more likely to have to ask a family member or friend to borrow a vehicle if they want to venture out of the village on an errand. While public transportation is available in some villages, there are a lot of villages that do not have that as an option. Even when it is available, it can be unreliable, expensive, or unsafe. Thus, even if they wanted to go to the councilor over the chief, they may not have access. Women also face time constraints, which may limit their ability to seek assistance from the councilor. Household chores, particularly in rural Senegal, can take an abundance of time for women (Hyde et al. 2020, Bardasi and Wodon 2010). Cooking, cleaning, gathering water, childcare, and income-generating activities are difficult activities to manage on their own; errands to the mayor’s office may be quite difficult for women to squeeze into their schedules. Much more research is needed to understand the gender differences in participation in local governance in both the formal and informal sectors. While the analysis here does not find stronger gender discrimination in the customary sector, more research is needed to understand the scope of gender discrimination in local governance. Focus groups or interviews may provide more insight than surveys for this type of question. Now let’s look at the results for minority ethnicities. Minority ethnicities are less likely to seek assistance from the chief than majority ethnicities are. Minority ethnicities also go to the councilor in higher proportion than majority ethnicities. What explains this finding? The explanation may be in line with my hypothesis that chiefs discriminate more against minority ethnicities because they are not native members of the community. This could lead minorities to seek assistance from the formal sector over the customary sector when possible. This assumes that minority ethnicities are minorities in their community and are likely living in a village with a chief from a majority ethnicity. Again, this is very difficult to test, as there is very little reliable data on levels of discrimination or on community ethnic make-up. Similar to the Afrobarometer question on gender discrimination, in 2018, minority ethnicities overwhelmingly report not having experienced discrimination on account of ethnicity in the past year. A whopping 94% report never experiencing discrimination, compared to 96% of majority ethnicities. The rate for minority ethnicities (0.118) is statistically equivalent to that of majority ethnicities (0.083) (t=-1.300, p=0.194). While it is possible there is very little ethnic 51 discrimination in Senegal, it is also quite possible that these questions suffer from social desirability bias. In fact, Adida et al. (2016) find that the ethnicity of the interviewer in relation to the respondent does affect the respondents’ answers on Afrobarometer surveys. Therefore, this question on its own cannot rule out ethnic discrimination. However, there are other indicators that may suggest discrimination is occurring in the customary sector. If there is discrimination, ethnic minorities might report less trust in the chief than majority ethnicities do. Minority ethnicities do report less trust in the chief than majority ethnicities do, although only slightly so: 81% of minority ethnicities report that they trust the chief somewhat or a lot, compared to 84% of majority ethnicities. The mean for minority ethnicities (2.325) is statistically lower than the mean for majority ethnicities (2.449) (t=3.931, p=0.000), but substantively, this is still a very small difference. Similarly, for the reported levels of trust in councilors, there is only a slight difference between minority or majority ethnicities: 53% of minority ethnicities report that they trust the councilor somewhat or a lot, compared to 55% of majority ethnicities. The mean for minority ethnicities (1.577) is statistically lower than the mean for majority ethnicities (1.664) (t=2.653, p=0.008), but again, substantively, this difference is quite small and likely isn’t the best indicator of discrimination. An alternative explanation is that discrimination is not causing this difference. Instead, it’s possible that many minority ethnicities in Senegal are not in fact minorities in their village, but instead that they live in a village made up of their ethnicity but happen to live amongst villages dominated by a majority ethnicity for that region. For example, Mande are a minority ethnicity in the Kolda region. The Kolda region is dominated by Pulaar people. However, some villages in Kolda are Mande villages, whose inhabitants are mainly Mande and whose chief is also Mande. These villages may be nestled amongst several Pulaar villages with Pulaar chiefs. In this case, there is no reason to think that a Mande chief would discriminate against their Mande villages, regardless of the minority ethnicity status in the broader community. However, in these instances, it’s possible that Mande chiefs are less connected to the broader community network in the Pulaar-dominant area, compared to Pulaar chiefs. This would make the Mande chief less effective than the Pulaar chiefs nearby and may encourage minority ethnicities to get assistance from the formal sector instead. 52 If this were the case, one would expect minority ethnicities to report lower rates of chief performance compared to majority ethnicities, which is not the case. There is no difference in chief performance. However, this could also be an instance of social desirability bias, that minority ethnicities respect their chiefs and don’t want to report negative opinions in the survey. To properly explore this question, we would need much more information on what type of community ethnic minorities live in. Are they living mainly among coethnics or are they mainly living among majority ethnicities? In addition, we would want to know if someone shares an ethnicity with their chief to better understand if discrimination is the mechanism driving these results. 3.7 Conclusion In this chapter, I used Afrobarometer data to look at variation across groups in their decisions around contacting local government actors for assistance, both in the formal and informal sector. By using Afrobarometer data, I am able to use a very large sample with a wide variety of possible indicators. This initial empirical chapter is important because it shows that across Senegal, despite conventional wisdom around gender discrimination, there is not a difference in the gaps in contacting the councilor or the chief for men versus women. This means that the formal sector is not any more or less accessible to women than the informal sector. There is, however, a difference between majority and minority ethnic groups. While minority ethnic groups are less likely overall to see the chief compared to majority, there is no difference between minority and majority ethnicities in their likelihood to contact the councilor. Similarly, the gap for minority ethnic groups in contacting the chief and the councilor is smaller than that of majority ethnic groups. This means that while minority ethnicities are not more likely overall to contact the councilor, they do not favor one over the chief to the same degree that majority ethnicities do. This suggests that there is more of a willingness of minority ethnic groups to use the formal system, even though they still use the customary system quite a bit. This supports my hypothesis and suggests that there are in fact some differences in how different ethnic group dynamics affect people’s choices in how to solve their local problems. In Chapter 4, I continue looking at how different groups make choices around using local government. I focus on a conjoint experiment I conducted in Senegal in 2021, which also looks 53 at how gender and native/non-native status affect the ways people contact formal and informal local government actors. There are a few benefits to the empirical design in Chapter 4 compared to Chapter 3. For instance, I develop a better measure of native/non-native status. I also compare (in a hypothetical vignette) those who contact the chief with those who contact the mayor instead of the local councilor. This is a better comparison, as the mayor has much more power than the local councilor. Lastly, I introduce two new aspects that may influence people’s choices in seeking assistance: the type of issue they’re having and whether their choices may have changed over time with changes in the distribution of local power. This all allows us to dig deeper into the question of how different identity groups interact with local government, particularly in a setting where formal and informal sectors overlap. 54 CHAPTER 4: ISSUE AREA AND USING FORMAL AND INFORMAL LOCAL INSTITUTIONS 4.1 Introduction In this chapter, I build on the analysis on Chapter 3. In Chapter 3, I used Afrobarometer data to examine how gender and local/non-local status affect the types of local actors that people choose to see for help in Senegal. I find that women’s tendency to go to the chief over the councilor is the same size as men’s. However, men are just as likely overall to seek assistance than women are. Chapter 3 also explores local/non-local status using a minority ethnicity variable. I find that non-locals are less likely to contact the chief than locals but are not significantly different from locals in contacting the local councilor. I also find that the gap between contacting the chief and contacting the councilor for non-locals is smaller than the gap for locals. These findings suggest that there may be fewer differences between men and women than I initially thought, but that local/non-local status could affect the way people choose who to seek for assistance. Chapter 3, however, had some limitations. First, the measure for local/non-local status is a rough proxy. I used people’s self-reported ethnicity to code whether they were a majority or minority ethnicity in the area in which their interview was conducted, based on the coding of majority ethnicities in the Ethnic Power Relations (EPR) dataset. The EPR does not code ethnicities in a very granular way, so this measure could only broadly tell us if someone is a minority ethnicity in a broad region; it could not necessarily indicate status in one’s village or commune, which, given my focus on local governance, is the more relevant consideration. The Afrobarometer also has limitations in the language they use and the types of local actors they ask about. Instead of asking about the village chief, they ask about the “traditional leader,” of which village chiefs are one type, but not the only type in Senegal. They also do not ask about the mayor, but rather the local councilor, which is a position with less power at the local level. This means people may not be contacting the local councilor, but still choosing the formal system at the local level by going directly to the mayor. This chapter explores the same ideas around the variation in which types of groups seek help from which types of local government actors. However, by using novel data from a conjoint 55 experiment conducted in Senegal in 2021, I am better able to target my questions regarding village chiefs and mayors to get data that better reflect the situation in rural Senegal. As in Chapter 3, I examine variation in gender and local/non-local status. Instead of using ethnic minority as a measure, I use whether someone was born in the village or has lived there several years. I am also able to add two additional dimensions to the conjoint analysis. I examine the type of issue a person is having, suggesting there may be variation in whether formal or informal actors are sought for assistance depending on what someone needs help with. I also add a time element to this analysis, exploring whether behavior has changed over time. This is important because there have been changes to Senegal’s decentralization laws that could affect people’s behavior in choosing whom to seek for assistance. These additional components allow us to gain an even broader understanding of how people decide who to seek for assistance at the local level in Senegal. The survey covered three regions of rural Senegal. Subjects heard a vignette, which varied four attributes to explore how gender, local/non-local status, issue area, and time affect people’s choices around seeking assistance from a local official. Respondents were then asked if the person in the vignette would be likely to go to the chief for assistance, if they would be likely to go to the mayor for assistance, and finally which local actor they would be most likely to go to for assistance. The findings from the conjoint experiment show that people still overwhelmingly say people should go to the village chief for assistance. Regardless of whether the vignette had a male or female or a local or non-local character, there was no difference in who respondents thought they would be most likely to go to for assistance: the chief. Also, the timing of the incident in the vignette did not shift opinion on whom to see, either. However, the type of issue the person in the vignette was having did have an effect: when the vignette included a land issue (buying land or land conflict), people were more likely to say that person would go to the chief than the mayor, compared to those vignettes in which the character needed a birth certificate. This suggests that the type of issue one is having does matter for which local actor they choose to see for help. Overall, for vignettes with a female character, people were less likely to say they would go to either the chief or the mayor, suggesting that there is some difference between men and women. Lastly, ethnicity does seem to affect people’s choices: Mande respondents were 56 more likely than Pulaar respondents to recommend going to the chief and to recommend going to the mayor. This suggests that while local/non-local status may be less important, there are still differences between majority and minority ethnicities in this area of Senegal. These results show that while there is some variation in how people perceive these choices are made, the chief is still the overwhelming favorite in whom to go to for assistance. This finding suggests that even when there is a choice, people still prefer the customary sector when they have an issue (at least among those I tested). This has strong implications for decentralization policy, as more and more responsibilities are shifted away from the village chiefs and to the mayor and local council. In this chapter, I first outline the theoretical expectations of each aspect of the conjoint experiment. I then discuss my data and research design, followed by a discussion more specifically of the conjoint experiment. Lastly, I outline the results of the experiment and discuss their implications. 4.2 Local Officials In this chapter, I continue to examine the formal and informal sectors of local governance in Senegal. In Chapter 3, I focused on traditional leaders and local councilors as the main representatives of local governance, as these are the officials Afrobarometer consistently asks about at the local level. In this chapter, I am able to ask more specifically about village chiefs, as this term reduces ambiguity and allows me to speak directly to people’s likelihood of visiting the chef de village, a position within every village or neighborhood in Senegal. This specificity is quite helpful, as “traditional leader” could refer to a few different types of local actors in Senegal. One such set of traditional leaders are marabouts, who are religious leaders affiliated with the various Sufi brotherhoods common across Senegal (Koter 2016, Villalon 1995). Marabouts maintain a strong relationship with their disciples; however, their presence and significance across Senegal is varied. Marabouts hold more power in some areas of Senegal, particularly Mouride marabouts in Central Senegal in the groundnut basin (Beck 2008). These leaders can offer goods and services to their disciples, but not everyone at the local level has access to them: if someone is not a follower of a given marabout, then they will not receive the benefits of that relationship (Villalon 1995). Thus, village chiefs are much more universal in 57 Senegal, as each village and/or neighborhood has a village chief, despite the prominence of marabouts in Senegalese politics. I also ask about the mayor. Mayors in Senegal are the leader of the local council. As discussed in Chapter 3, local councils serve at the commune level (the lowest level of decentralized governance in Senegal) and are comprised of elected councilors. Councilors are elected using a mixed proportionality-plurality system (Wilfahrt 2022)10. Local councils are in charge of providing goods and services across several sectors, including primary education, local health centers, agriculture and livestock production, and water resources, among other areas. These councils are important because they are the closest part of the formal system of governance to the local population. Until 2022, mayors were chosen indirectly by the local councilors. Now, they are elected popularly by being placed at the top of the winning electoral list. While many aspects of the mayors’ duties are voted on by the council, mayors have much more power to direct the budget and the agenda compared to an individual councilor. Thus, it’s likely that many people skip visiting the councilor entirely, opting to see the mayor for assistance due to the belief that the mayor can be of more use. Therefore, by focusing in this chapter on the mayor instead of local councilors, as I did in Chapter 3, we can gain a better understanding of whom people are likely to see if they do have an issue they believe a local official can help solve. Thus, the analysis in this chapter allows for some increased specificity of language to get a better understanding of how people interact with officials at the local level. Similar to Chapter 3, I explore different attributes that may affect how people choose which type of local actor to see. However, by taking a slightly different methodological approach via a conjoint experiment, I am able to explore other areas that may affect people’s choices as well. In the next section, I discuss the choices that people make in getting local assistance across four main attributes. These include gender, local/non-local status, issue type, and time period. I will discuss each attribute in turn. 10 Under this system, 55% of the seats are allocated from the majority list of the winning party. The remaining seats are divided proportionally from the proportional lists of participating parties (Interview with election official, Kolda Region, 11/24/2021). 58 4.3 Gender Like in Chapter 3, I examine the effect of an individual’s gender on the general likelihood they will seek help from a local official, as well as which type of local official they would go to for help with an issue. Generally, I expect that gender will affect the likelihood that someone seeks help from any type of official. Men will be more likely to see either the chief or the mayor for several reasons. First, men are more likely to be responsible for dealing with household issues that would require the assistance of a local governing authority, be it formal or informal. Men may also feel more comfortable seeking assistance from leaders who are more likely to share their gender: the chieftaincy in Senegal is inherently a male institution that is passed primarily from father to son, older brother to younger brother, or uncle to nephew. The mayor’s office is also still largely male-dominated in Senegal, as female candidates have trouble gaining funding and support for the top spot on the local council. A female coordinator for an opposition party told me, “It’s very hard for women to be candidates for mayor. It’s very rare…I wanted to be a candidate, but I didn’t have the means. Women can do anything, but it's hard” (Interview with opposition party coordinator, Kolda Region, 11/26/2021). Thus, overall, I expect women to be less likely than men to seek assistance in general, regardless of the type of officials available. That said, when women do decide to seek assistance, I expect them to choose the formal system over the informal. This stems largely from the idea that inequities such as gender discrimination will be found in the informal sector more so than the formal sector. Such discrimination can influence not only the probability that one seeks help from an official, but which sector one chooses to use for assistance. While the occurrence of gender discrimination in formal and informal local governance has not been extensively examined empirically, theoretical work in this area often expects customary authority to lack accountability and even limit the rights of citizens compared to the legal rational authority of the state (Mamdani 1996, Ntsebeza 2005). These ideas around traditional governance harken back all the way to Weber’s three types of authority, in which traditional governance is deemed legitimate through custom, tradition, and inheritance, all of which can more or less describe customary authority in Senegal (Weber 1978). Indeed, when looking at gender at the local level, there is evidence across much of Africa that women often experience discrimination at the hands of customary leaders, particularly regarding land and inheritance issues (Tripp 2004, Rangan and Gilmartin 2002). 59 On the formal side, however, it’s expected that as a state develops, the democratically elected arms of the state—even at the local level—will be held to a bureaucratic legal standard that provides both accountability and uniformity of policy across local government outposts (Weber 1978, Baldwin and Holzinger 2019). Formal governance of this nature is expected to limit discrimination against minority populations, including women, because its authority is derived from the law and the law is applied equally to all citizens. From this theory, I expect discrimination to be higher among the informal sector than the formal one. Regarding gender, if women are experiencing or expecting discrimination in one sector, they may choose to send someone else—a man—on their behalf to get better outcomes on their problem. They may also choose not to visit the customary authority and go straight to the local mayor for assistance. Thus, I have two main expectations regarding gender and local officials:11 H1a: People will be more likely to think that men will seek help from either the mayor or the chief than women. H1b: People will be more likely to think that women will be more likely to go to the mayor than the chief. 4.4 Local/Non-Local Status Local and non-local status might also affect how someone interacts with the formal and informal sectors of local government. For instance, local insiders may feel more comfortable than outsiders do going to the chief for with their problems or may expect favorable treatment from the chief. Much of the evidence towards discrimination against those with non-local status is in the area of land rights. Boone (2014) notes, “Where customary authorities at the local level (chiefs or lineage leaders) control land allocation and adjudication, ethnic outsider or in-migrants are in a position of socioeconomic and political subordination” (100). For outsiders, if they don’t have confidence that the chief will help solve their issue, they may be more likely to go directly to the mayor. I also expect the opposite to also be true: those who are local to an area will have more privilege with the village chief and will be more likely to go there than to the mayor for 11 These hypotheses explore what people think others will do, rather than what they will do themselves or have done in the past. This decision is explained further when discussing the conjoint experiment in the methods section of the chapter. 60 assistance. Just like with gender, as part of the legal bureaucratic state, the mayor is likely to have less power to discriminate directly just because someone is not local to the area (Weber 1976). While of course bias and discrimination can still exist within the formal structures of the state, as these institutions are staffed by humans who may hold the same beliefs and expectations as those in customary roles, the level of expected or actual discrimination against women or non- locals will likely be lower in the formal sector than the informal one. Therefore, I expect: H2a: People will be more likely to think that those who do live in the village in which they were born will be more likely to go to the village chief than the mayor. H2b: People will be more likely to think that those who do not live in the village in which they were born will be more likely to go to the mayor than the village chief. H2c: People will be more likely to think that those who do live in the village in which they were born will be more likely than those who do not live in the village in which they were born to go to the village chief. H2d: People will be more likely to think that those who do not live in the village in which they were born will be more likely than those who do live in the village in which they were born to go to the mayor. 4.5 Issue The analysis in Chapter 3 also does not give any indication of why someone may seek assistance from the chief or the local councilor. In this chapter, I include three hypothetical issues to the conjoint experiment. This adds to the analysis because it gives insight into why someone might be going to the chief over the mayor or vice versa. People will rationally want to seek help from the person that is most likely to solve their issue, barring other barriers. Thus, when it is clear which official will deal with which type of issue, people will go directly to that official for help. However, in many instances, there is ambiguity and overlap around whether the formal or informal sector is responsible for a given issue. This is especially seen in situations of legal pluralism: people can choose whether to take their issues to the formal courts or use the system of customary law (Hern 2022). However, the same principle applies here: people can choose to 61 take their issue to the formal or informal sector for help, or to engage in venue shopping by trying multiple officials until their issue is solved. The first issue I look at, obtaining a birth certificate, is a good example of this overlap in responsibility. After the birth of a child, most children are registered with the state, so this is an experience that the majority of Senegalese would be familiar with. It is also an issue that chiefs and mayors can both help with (Interview with Commune Secretary, Tambacounda Region, 11/15/2021). When a child is born, the family can give the child’s information to the village chief. The chief will wait to collect the information from several births and register those births monthly at the mayor’s office. Families also have the option of going directly to the mayor’s office and registering the birth themselves. Thus, people have a choice on whether they go to the chief for this task or to the mayor. The second issue is obtaining land. This is an important issue in much of rural Senegal, as obtaining land can be difficult, complicated, and onerous for people (Interview with Political Candidate, Kolda Region, 11/26/2021). Traditionally, land was customary and controlled by village chiefs. To obtain a piece of land, one went to the chief and if the chief consented, the land would be given to that family or individual to either build on or to farm. While there was no formal ownership of such land, ownership was presumed unless someone stopped using their land. Chiefs no longer have formal control over land, though do retain informal control (Honig 2017). Through decentralization laws, land titling power has been given to the mayor, meaning those who want to acquire land start the process there (Interview with Political Candidate, Kolda Region, 11/30/2021). Mayors can still engage chiefs in the process. For example, a former councilor told me, “When someone buys land, we make sure the councilors and the chief are called. They ask each [the chief and the councilors] if they approve of the sale. If each doesn’t agree, the land isn’t sold. So, they both work together, but each has his or her role” (Interview with former councilor, Kolda Region, 8/1/2019). This isn’t always the case, though. Another chief said, “Some mayors just ignore the chiefs and just give away land themselves. That’s disrespectful. It doesn’t happen here, but I know of other places where it does. If we all follow the rules and norms, there should be no problem. It’s only a problem if they don’t respect the chiefs” (Interview with village chief, Tambacounda Region, 7/30/2019). The issue of obtaining 62 land is also part of the purview of both the formal and informal sectors, though has been changing over time. The third issue is land conflict. Because much of the land in Senegal is not formally titled, and available land is becoming increasingly scarce, there is lots of potential for land conflict (Kaag et al. 2011). Village chiefs have long been responsible for conflict resolution, especially for land conflict, since land was traditionally the purview of the chief. Even as certain responsibilities over land have shifted to the mayor, chiefs still play a large role in conflict resolution (Dzivenu 2008). For instance, one chief said, “When there’s a land conflict, I fix the problem if I can before the council finds out, because the council can make it worse. But I’ll call the council if I have to” (Interview with village chief, Kaffrine Region, 8/7/2019). Another chief said, “When there is a land issue between community members, we try to fix the problem among us” (Interview with village chief, Kaffrine Region, 8/10/2019). Thus, I expect people who are struggling with a conflict over land to first seek assistance from the chief. These three issues lead to the following expectations: H3a: People will be more likely to think that people will go to the chief for issues involving land disputes than for birth certificates or buying land. H3b: People will be more likely to think that people will go to the mayor for buying land than for land disputes. 4.6 Time Lastly, time is an important component of this analysis. In 2013, President Macky Sall signed Act III of Decentralization, which began to be implemented the following year. This new law expanded the power of mayors and local councils at a time when chiefs’ powers were declining (Interview with local development agent, Kolda Region, 11/21/2021). First, it removed many of the distinctions between urban and rural councils by giving rural councils the same terminology as urban ones. Rural communities became communes, and presidents of rural communities became mayors (Wilfahrt 2022). Act III transferred more administrative power to the mayors, lessening the load of the sous-prefets (Interview with secretary of the sous- prefecture, Tambacounda Region, 7/29/2019). As one local development agent described, “Act 63 III was implemented to give an increase in power at the community level, but a decrease in power in Dakar. And give equality. To give more fiscal autonomy. More possibility” (Interview with local development agent, 12/1/2021). It also gave mayors the power to subdivide villages and reappoint chefs de quartier when village populations start to increase. This effectively gave mayors the power to replace village chiefs with their own appointees if they so choose. This was seen by many chiefs as a threat to their authority (Interview with village chief, Kolda Region, 07/31/2019). Given this change in power dynamics between chiefs and mayors, particularly the expansion of administrative power given to mayors’ offices, I expect to also see a shift in how people choose whom to approach for assistance. This is important because any change based on Act III would indicate that people are responsive to the structure of the decentralized local governance and, therefore, changes to the structure can shift the type of local actors people seek assistance from. As the local government professionalizes—a major goal of Act III—this could provide people with improved access to services at the local level. However, if changes in structure such as Act III, which increases the power of formal-sector local actors, do not encourage increased use of the formal sector, then citizens are perhaps not benefiting from the changes or are otherwise resistant to them. My expectations around changes over time are as follows: H4a. People will think people will be more likely to go to the mayor recently than they would have ten years ago. 4.7 Data and Research Design To test these hypotheses, I use data from an original survey conducted in November and December of 2021 in three departments of Senegal: Kolda, Tambacounda, and Velingara. Within each department, I selected five communes,12 from which I selected all the villages in the sample. By varying the selected villages across several communes, instead of selecting several villages from only a few communes, the survey included opinions from subjects under several different mayors and local councils. All in all, 695 subjects were surveyed across 80 villages. 12 The communes were selected for both diversity of respondents, as well as ease of access for the enumerators. 64 Within each village, households were chosen using a random-walk approach. Within each household, survey enumerators used convenience sampling to select subjects. Enumerators alternated between men and women in selecting subjects to ensure both genders were well represented in the survey. Due to the high rates of illiteracy among the survey population, subjects were verbally made aware of the risks of participation, and each subject verbally consented to participate prior to the start of the survey. Names and other identifying information of survey participants were not collected to ensure anonymity. Privacy was maintained during the interview by offering the subject a private space to conduct the survey. Surveys were conducted in Pulaar, Mandinka/Bambara/Djakhonke, Wolof, French, and Sarakhole. The demographics of the survey do not entirely align with Senegal as a whole. As described in Table 4.1, this survey leaned more heavily male (55%) than the Afrobarometer surveys (50%), which are nationally representative. The average ages are similar between the surveys, with an average of 42 years old in this survey, compared to 44 in Afrobarometer. The ethnicities represented in this survey, however, are not nationally representative, as they are in Afrobarometer. For instance, 74% of respondents in this survey identify as Pulaar, compared to only 29% in Afrobarometer. In contrast, only 4% of respondents identify as Wolof, which is the largest ethnic group nationally at 39%. The differences in ethnic distribution across surveys is due to the geographic focus of this survey in southeast Senegal. Because ethnic groups still largely live in geographic areas correlated to their traditional homelands, the makeup of the departments of Tambacounda, Kolda, and Velingara is very different than Senegal as a whole. Lastly, this survey also has a higher rate of lived poverty than the Afrobarometer surveys.13 For instance, a full 67% of respondents on this survey reported having gone without food at least several times in the past year, compared to only 40% for Afrobarometer. 13 Because the lived poverty variable constructed in this survey is not identical to the one used in Afrobarometer, I report just one indicator that makes up the index in both surveys. This question asks subjects, “Over the past year, how often, if ever, have you or anyone in your family gone without enough food to eat?” 65 Table 4.1. Demographic differences between the original survey and Afrobarometer Indicator Gender (%) Male Female Original Survey Afrobarometer 55 45 50 50 Age (mean) Ethnicity (%) Wolof Pulaar Serer Mandinka/Bambara Diola Other Gone without food (%) Never Just once or twice Several times Many times Always 42 44 74 0 16 0 6 10 23 41 24 3 44 39 29 15 7 7 4 53 7 21 14 5 Again, the survey was conducted in three departments: Tambacounda, Velingara, and Kolda. Tambacounda is part of a region with the same name, while Velingara and Kolda are both part of the Kolda region. These three departments were selected due to the rural nature of many of the areas surrounding the capital cities of each of these departments. By selecting a largely rural sample, I can ensure that selected villages contain chiefs with some level of historical legitimacy and all the responsibilities that come with that. Urban areas have “chefs de quartier,” or neighborhood chiefs, which operate similarly to village chiefs; however, these urban positions were only recently created and are appointed by the mayor instead of inherited (Interview with village chief, Kolda Region, 8/3/2019). It was important in this study to focus primarily on customary village chiefs instead of neighborhood chiefs primarily due to the higher legitimacy found among the village chiefs. While this legitimacy isn’t directly tested, Afrobarometer data can show us some major differences between urban and rural respondents and their perspectives on village chiefs. For instance, rural respondents are more likely to contact the chief than their urban counterparts are: 39% of rural respondents, compared to 22% of urban respondents, 66 contacted their chief at least once.14 Rural respondents also show higher trust in their chief than urban respondents do: 87% of rural respondents trust the chief somewhat or a lot, compared to 77% of urban respondents.15 Lastly, rural respondents are more likely to report that chiefs listen to what people like them have to say: 52% of rural respondents agree that chiefs listen often or always, compared to 37% of urban respondents.16 Neighborhood chiefs thus lack the customary legacy of village chiefs. While selecting a sample that prioritized the type of chief was very important in the design of the study, it does lead to some major differences between the sample population and Senegal as a whole. The survey population is much more rural than Senegal more generally. Westen Senegal in particular is rapidly urbanizing. For instance, according to Senegal’s National Agency for Statistics and Demographics, the region of Tambacounda was 26% urban and the region of Kolda was 30% urban, compared to 48% urban for Senegal as a whole (ANSD 2022). Because the sample was not randomly selected, these results are not generalizable to the whole of Senegal. The rural nature of the sample cannot generalize to urban areas; however, there are many areas outside of Tambacounda, Kolda, and Velingara that are similarly rural, and I expect these results to generalize to those areas, as well. While the ethnic makeup of this survey also does not reflect Senegal as a whole, I expect that a more representative survey would show chiefs functioning similarly at the local level. In 2019, I conducted interviews with village chiefs in the region of Kaffrine, which is located in central Senegal. There were no major distinctions between how subjects in Kaffrine described the role of the village chief or issues regarding local governance compared to interviews with subjects in Tambacounda or Kolda. The one exception to this may be in religious centers such as Touba or Medina Gounass, where marabouts have outsized local power. 14 The rate for rural respondents contacting the chief (0.900) is statistically higher than that for urban respondents (0.471) (t = -16.487, p = 0.000, 0-3 scale). 15 The rate for rural respondents trusting the chief (2.540) is statistically higher than that for urban respondents (2.245) (t = -9.494, p = 0.000, 0-3 scale). 16 The rate for rural respondents reporting that the chief listens (1.421) is statistically higher than that for urban respondents (1.051) (t = -5.795, p = 0.000, 0-3 scale). 67 4.8 Conjoint Embedded in my survey, I included a conjoint experiment to better understand how the use of formal and informal local actors varies across the local population. The conjoint experiment contains a vignette in which four attributes are varied at random. The text of the vignette is as follows: [Awa ; Mamadou] lives in a village near here. S/he has lived in this village [since s/he was born ; for several years]. [Recently ; Ten years ago], s/he had a problem that s/he wanted to solve. [S/he had a newborn and wanted to obtain a birth certificate for her/his child ; S/he wanted to buy some land ; S/he had a conflict over land with her/his neighbor.] Due to the nature of the vignette, I do not randomize the order of the sentences, only the attributes within each sentence. The main reason for the lack of randomization in the vignette is that the population of Senegal still has quite high rates of illiteracy. According to the World Bank, in 2022, only 58% of the population over 15 was literate. Thus, it was essential to design a survey that could be given entirely orally. This limited my ability to include a conjoint table in the survey to show the respondents the attributes in the vignette, as is common practice with conjoint experiments (Bansak et al. 2021). Thus, I instead designed a vignette that could be easily understood when read aloud. However, in order for the vignette to make sense as read, I was limited in how I could order the sentences. Using a vignette instead of a table introduces some limitations, including the increased chance of respondent fatigue and a reduction in the total number of tasks subjects can complete. In this case, subjects were given four tasks to complete. This means that each subject was given four randomized versions of the vignette, each followed by three questions on what the person in the vignette should do. As many as 30 tasks have been shown to not cause issues in response quality when using conjoint tables, so the four tasks used here are likely not an issue, either (Bansak et al. 2018). Table 4.2 outlines all four attributes and their levels. The first of the four attributes varies the gender of the hypothetical person in the vignette. This is done by varying the name of the 68 hypothetical person in the vignette: Awa is a common female name in Senegal, while Mamadou is a common male name. This continues the inquiry from Chapter 3, examining how gender affects formal and informal local government usage. Table 4.2. Attributes and levels Attributes Gender Time of village residency Recent or not recent issue Issue type Levels Male Female Born in the village Lived in the village for several years Issue happened recently Issue happened ten years ago Obtaining a birth certificate Purchasing land Land conflict with neighbor The second attribute varies how long the person has lived in the village: either they were born in the village or have lived there for several years. This is designed to get at whether the person is a local or non-local to the village. In Chapter 3, I used a variable for minority ethnicity to approximate a respondent’s local or non-local status in the village. I used this approach because Afrobarometer doesn’t ask any questions specifically about being local or non-local to an area. However, we know that this type of insider/outsider status in a village can matter (Honig 2017). While the analysis in Chapter 3 provided some insight—that minority ethnicities seek assistance from the local councilor more than from the chief, compared to majority ethnicities— it did not indicate directly if someone were local to the area or not. In this chapter, subjects are asked about a hypothetical person who is either born in the village or not. This gets closer to approximating local/non-local status than was possible in Chapter 3. I expect that if someone moved into a village, they may have less access to the village chief or may feel less entitled to using the customary system than those who were born in the village. It is still not a perfect measure, as someone might be born outside the village and still considered an insider, and someone might be born in the village and still considered an outsider. For example, if someone from the village moves to the city for work, their children would be born in the city and not in the village. However, if that family moved back to the village, those children would be considered insiders because of their familial ties to the village. In contrast, a family that moves to the village with no previous ties to the village could have children in the 69 village, but those children would still be considered outsiders due to their family’s status, despite the place of their birth. Despite these possibilities, this measure does get closer to the concept than the ethnic minority measure that was used in Chapter 3. The third attribute varies the timing that the issue the person is having takes place. Either this is a recent issue, or it is one that happened ten years ago. This will identify if there has been a change over time in how people interact with the formal and informal local institutions. Due to the passage of Act III of Decentralization in 2014, this would also show whether the changes in law affect who people go to for help. While looking at variation over time is important to understanding the dynamics of how people use local government officials, there are also limitations to this approach. For example, people may not be good at remembering how things were different or what they would have done ten years ago. Any changes from Act III of Decentralization may also have been too subtle for people to notice if they don’t interact with local governance regularly. Or people may not remember when Act III of Decentralization took effect or connect that that’s what the question is getting at. However, asking directly about Act III of Decentralization would have limitations as well: people may not be familiar with the law, but may have noticed changes in governance, etc. While all of these limitations are important to consider, asking about change over time is still important, because institutional changes can have a big effect on how people interact with local government. Lastly, the fourth attribute varies the type of issue the person is having. The three issue types include: getting a birth certificate for a child, buying land, and having a conflict over land with a neighbor. This will help us understand if people are selecting the person they seek for assistance based on the type of issue they are having. After the vignette was read to the subject, interviewers asked a series of three questions about the hypothetical person in the vignette: 1. How likely is it that s/he would ask for assistance from the chief for this problem? 2. How likely is it that s/he would ask for assistance from the mayor for this problem? 3. Who is s/he most likely to go to for help solving this problem? 70 The first two questions get at the likelihood that someone would go to the chief/mayor for help with their problem. Response options included very likely, a little likely, not very likely, and not at all likely. These questions are just asking generally how likely it would be that the character in the vignette would go to a given local official for assistance, given the context of the vignette. Asking about the village chief and the mayor in separate questions allows for the possibility that someone may go to one or both sets of officials for each of these issues. The subject doesn’t need to choose between going to the chief or the mayor because it’s possible they believe the person in the vignette would go to both the chief and the mayor or neither official— neither of which would not be captured in the data is they were given a forced-choice option only. The third question, however, is intended to capture whom the hypothetical person is most likely to seek for help, including local actors besides the chief and the mayor. This was an open- ended question, designed to capture who the appropriate official is for each type of problem if it’s neither the chief nor the mayor. For instance, if instead of going to the chief for land issues, respondents actually sought out the marabout, the third question would give them a chance to report that. It also serves to show whether one type of official is a better fit for a given context than the other types. Despite being an open-ended question, 96% of responses said the person should go to either the chief (71%) or the mayor (25%). Two percent said the local councilor, and less than one percent recommended another entity, such as the marabout, the sous-prefet, a friend or neighbor, the health center, or the president of the women’s group. Thus, in this analysis I collapse the variable for this outcome into a binary variable for either chief or mayor, since those two were the vast majority of responses. (Other responses were dropped from the analyses.) This suggests that indeed village chiefs and mayors are the appropriate entities for solving the issues in the vignette. In addition to these four attributes, I include a series of control variables in my analysis. These include the gender, ethnicity, education level, lived poverty,17 and age of the subject. These controls largely align with those used in Chapter 3. 17 The lived poverty variable is constructed using two questions that align with the Afrobarometer questions on the same topic. The two questions ask, “Over the past year, how often, if ever, have you or anyone in your family: Gone without [enough food to eat/a cash income]?” The possible responses include, “Never,” “Just once or twice,” “Several times,” “Many times,” and “Always.” These are coded to range from 0 to 4. To construct the index variable here, I added the responses from both questions together. Thus, the possible responses range from 0 (a response of “Never” on both questions) to 8 (a response of “Always” on both questions). 71 I used Lukac and Stefanelli (2020) to conduct a power analysis to determine the target sample size for the conjoint experiment. The attribute with the highest number of levels (Issue) has three levels. If each subject completes four tasks, a sample size of N = 500 should be enough to achieve at least 80% power, given an effect size of 0.05. This type of research design has several advantages. First, by asking subjects about hypothetical situations, I can vary several attributes at once and see how people think someone of that profile would respond. This allows me to keep the variables I am most interested in from being overly obvious and limits social desirability bias from the respondents (Teele et al. 2018). The alternative of conducting a survey with a large enough sample size of all the targeted attributes would have to be significantly larger in order draw conclusions from people in each combination of attributes, as well. As was seen in Chapter 3, most people have not had a reason to see the chief or councilor in the past year. By using a conjoint experiment, I am able to ask all kinds of people, even those who have not specifically had an issue recently, what they think someone should do if they do have such a problem. In addition, this type of research design allows subjects to be asked about potentially sensitive identities or life circumstances without offense, as they are asked in regards to a hypothetical situation instead of a personal one (Bansak et al. 2021). The main downside to this approach is that respondents are asked about how they think a hypothetical character would respond to a hypothetical situation. Thus, in this case, men are sometimes providing answers for what they think a woman would do in a given circumstance and vice versa. This means that some responses may be based on stereotypes instead of actual experience. Basically, it might capture, for instance, what people think women would do instead of what they actually do. Another possibility is that people respond to the question based on how they themselves would respond, as opposed to how they think the character in the vignette would respond. This would mean that answers for what Awa would do may not be actually capturing, for example, what a woman who has lived in the village all her life would have done about buying land ten years ago, but rather what the respondent—regardless of their demographics— would do now. In an attempt to address this issue, I control for individual demographic features. Lastly, people may be responding based on what they think people should do, rather than what they would actually do. This could mean that some responses capture, for instance, that a woman 72 should go to the chief for a birth certificate, regardless of whether she would be likely to do so in reality. That said, with enough respondents, it’s likely that most people have a general sense of how people are using local institutions when they have an issue. In my analysis, I run a separate OLS regression model for each of the three outcome questions above. The four main attributes are the independent variables in each model, plus I include the controls mentioned above and cluster the standard errors by subject. While it was the third question asked in the survey, I focus my analysis below on the outcome question asking who the character in the vignette is most likely to see. I then present results for the other two questions: whether the character would go to the chief for this issue and whether they would go to the mayor. I also provide some additional exploratory analyses to better understand the results of the conjoint experiment, particularly using interactions to look more closely at gender and ethnicity. This exploratory analysis will provide additional insight into the results by looking at subgroups in the data. It’s possible that some results mask variation between, for example, male and female respondents, or along ethnic lines. By exploring the results in this way, we will gain a better understanding of the data and explore avenues for further research. 4.9 Results and Discussion 4.9.1 Descriptive Results If people perceive that factors such as gender, local or non-local status, issue type, or time affect how people choose to interact with local officials, we should expect to see differences in people’s expectations between who men and women and local and non-local populations are likely to choose for help with a problem, as well as differences across issue areas, and over time. In this section I discuss the results from the conjoint experiment, first discussing some descriptive results for all four attributes, then looking at regression results for the three outcome questions for the conjoint experiment. I start with the question that asks whom the person in the vignette is most likely to see for assistance and then examine whether they’re likely to go to the chief for help and whether they’re likely to go to the mayor for help. Lastly, I discuss some exploratory analyses, particularly looking at how the respondent’s gender and ethnicity affects their responses. 73 As in Chapter 3, in which I find that people are generally more likely to report having contacted the chief than the local councilor, I find that, in the majority of vignettes, people said the hypothetical person should go to the chief over the mayor or any other local official. As seen in Figures 4.1 and 4.2, for 71% of the vignettes overall, people recommended the chief, compared to 24% for the mayor and 4% for other local actors. The results are surprisingly consistent across three of the four attributes. For the vignettes in which the character was named Awa (a female name), people recommended going to the chief 71% of the time, compared to 24% for the mayor and 5% for another type of official. When people received Mamadou as the character, they recommended going to the chief 72% of the time, compared to 25% for seeing the mayor and 3% for seeing another type of local official. Again, these results are very similar to each other, suggesting that people don’t perceive gender to be leading to a large difference between who people will go to with their problems. Figure 4.1. Percentage of vignettes in which the respondent said the character would go to the chief, mayor, or other local official, by character’s gender and length of time living in the village 100% 80% 60% 40% 20% 0% 3.9 24.8 5.0 24.4 2.8 25.3 3.5 25.0 4.2 24.6 71.3 70.7 72.0 71.5 71.2 Overall Awa Mamadou Born Several Years Chief Mayor Other The results are even closer for vignettes in which the character was born in the village and those in which they are not. When the character was born in the village, 72% said they would go to the chief, compared to 25% for the mayor and 4% for other types of local officials. When the character had lived in the village for several years, 71% of the time, people said they would go to the chief, compared to 25% for mayor and 4% for other types of officials. Again, 74 whether someone is born in the village or not does not seem to lead people to perceive much of a difference in expected behavior for solving problems. Figure 4.2. Percentage of vignettes in which the respondent said the character would go to the chief, mayor, or other local official, by character’s issue type and timing of vignette 100% 80% 60% 40% 20% 0% 3.9 24.8 5.4 31.3 2.4 22.2 3.8 21.4 3.7 24.7 4.0 24.9 71.3 63.3 75.4 74.8 71.5 71.1 Overall Birth Certificate Buying Land Land Conflict Recently Ten Years Ago Chief Mayor Other There is a little bit more variation when looking at the type of issue the character in the vignette is having. For the vignettes with the issue of getting a birth certificate for a child, people said the person would go to the chief 63% of the time, compared to 31% for the mayor and 5% for other local actors. For vignettes where the character wanted to buy land, people said the character would go to the chief 75% of the time, compared to 22% for the mayor and 2% for other local actors. Similar to the buying land issue, for the vignettes in which the character had a land conflict with a neighbor, people also said they would go to the chief 75% of the time, compared to 21% for the mayor and 4% for other local actors. This suggests that the issue type might be a bigger driver than identity type in determining how people determine whom to approach for assistance with an issue. Lastly, there is very little variation when looking at the timing that the issue took place. This attribute is designed to pick up on change in behavior after the passage of Act III of Decentralization in 2014. For vignettes that took place recently, people said the person would go to the chief 72% of the time, compared to 25% for the mayor and 4% for other local actors. When people received a vignette that took place ten years ago, they gave nearly identical 75 answers. They said the person would go to the chief 71% of the time, the mayor 25% of the time, and other local actors 4% of the time. Thus, it does not seem that people are perceiving a change in behavior from before Act III to after. The story is similar when we look at the other two outcome questions: how likely is the person in the vignette to go to the chief and how likely is that person to go to the mayor? Figures 4.3 through 4.6 show frequencies for each of these two outcome questions for each attribute. Generally, there is quite a bit of consistency across attributes, with people overwhelmingly saying the person would be very likely to go to the chief and much fewer saying they would be very likely to go to the mayor. As seen in Figure 4.3, for 86% of the vignettes, people said the person was very likely to go to the chief. In another 7% of vignettes, people said the person was a little likely, compared to 5% of vignettes where people said the person was not very likely to go to the chief and in only 2% where they were not at all likely to see the chief for help. For 40% of the vignettes people received, they said the character would be very likely to go to the mayor and for another 21% of the vignettes, people said the character would be a little likely to go to the mayor for their problem. This compares to 27% of the vignettes in which people said the character would not be very likely to go to the mayor, plus not at all likely for 12% of the vignettes. This means that, for roughly 60% of the vignettes, the person would be a little likely or very likely to see the mayor for their issue. This compares to over 90% of the vignettes where the person would be a little likely or very likely to see the chief. Thus, the chief still seems to be the main point of contact for local issues, despite the transfer of power from the chief to the mayor in recent decades. Similarly, when people received Awa as the name of the character, they said she was very likely to go to the chief for 84% of the vignettes, compared to 89% when people received Mamadou. For the vignettes with Awa as the character, people said she would be a little likely to go to the chief 9% of the time, compared to 5% for those vignettes with Mamadou. Respondents said the character would not be very likely to go to the chief only 5% of the time and not at all likely to go to the chief only 2% of the time, regardless of whether the subject received Awa or Mamadou as the name of the character. There is some difference in the frequency that people said the person in the vignette would be very likely to go to the chief when they received a vignette with Awa compared to Mamadou. However, over 80% of people thought the character 76 would be very likely to go to the chief regardless of gender, meaning that when people have issues, they are probably very likely to start at the chief regardless of gender. When asked if the person in the vignette would go to the mayor, the results were quite different than when asked if the person would go to the chief. However, results were quite consistent for vignettes with Awa and those with Mamadou. For those vignettes with Awa, only 39% of the time people said she would be very likely to see the mayor, compared to 40% for those vignettes with Mamadou. For both those with Awa and those with Mamadou, people said they would be a little likely to go to the mayor 21% of the time and not very likely 27% of the time. For those vignettes with Awa, people said she would be not at all likely to go to the mayor 13% of the time, compared to those with Mamadou at 11% of the time. This means that the gender of the character in the vignette did not seem to change the likelihood that people said they would go to the mayor. Figure 4.3. Likelihood of going to the chief for help with an issue and likelihood of going to the mayor for help with an issue, by gender of the character in the vignette 2.1 4.9 6.7 2.3 5.0 8.9 1.9 4.8 4.6 86.3 83.9 88.7 12.0 27.1 21.3 13.2 26.9 21.2 10.9 27.4 21.3 39.6 38.7 40.4 100% 80% 60% 40% 20% 0% Overall Awa Chief Mamadou Overall Awa Mamadou Mayor Very likely A little likely Not very likely Not at all likely Table 4.4 shows the likelihood of going to the chief or to the councilor for vignettes in which the character was born in the village and those in which the character had lived there for several years. This is designed to get at whether the character was a local or non-local to the village. For those receiving vignettes with a character born in the village and those in which the character had lived there for several years, there is little difference in how likely people thought 77 they would be to go to the chief. For the vignettes in which the character was born in the village, people said they would be very likely to go the chief 87% of the time, compared to 86% for those vignettes in which they had lived in the village for several years. For both the vignettes with a character born in the village and ones with a character who had lived there for several years, people said they would be a little likely to go to the chief for help 7% of the time. Those with a character born in the village would be not very likely to go to the chief 4% or not at all likely only 2% of the time. For vignettes with a character who has lived in the village for several years, people said they would not be very likely to go to the chief 5% and not at all likely 2% of the time. Here, too, people consistently said the person would be very likely to go to the chief for help with their problem, regardless of whether they were born in the village or not. Figure 4.4. Likelihood of going to the chief for help with an issue and likelihood of going to the mayor for help with an issue, by length of time the character has lived in the village 100% 80% 60% 40% 20% 0% 2.1 4.9 6.7 2.0 4.2 6.7 2.2 5.5 6.8 86.3 87.1 85.6 12.0 27.1 12.2 26.9 11.9 27.4 21.3 21.0 21.5 39.6 39.9 39.3 Overall Born Several Years Overall Born Several Years Chief Mayor Very likely A little likely Not very likely Not at all likely When asked how likely the person in the vignette would be to go to the mayor for help with their problem, the answers were similarly consistent between those who were given a vignette with a character born in the village and those who were given a vignette with a character who had lived in the village for several years. However, there was less consensus overall as to how likely the character would be to go to the councilor, compared to the responses for going to the chief. For those vignettes with a character who was born in the village, people said they would be very likely to go to the mayor 40% of the time, compared to 39% of the time for those 78 vignettes with a character who had lived in the village for several years. As with characters named Awa or Mamadou, both vignettes with characters born in the village and those not born in the village, people said they were a little likely to go to the mayor 21% of the time, not very likely to go to the mayor 27% of the time, and not at all likely 12% of the time. In other words, being born in the village or not born in the village produces nearly identical responses and does not seem to affect the likelihood that people said they would go to the mayor. Figure 4.5 shows the likelihood of going to the chief or the councilor by the type of issue the character in the vignette is having. Again, the three issues are 1) obtaining a birth certificate for a child, 2) buying land, and 3) resolving a land conflict with a neighbor. Just as with the initial outcome question, there is more variation here across issue type than any other attribute. When people received a vignette in which the issue was getting a birth certificate, they said the person was very likely to go to the chief 80% of the time, compared to 90% for buying land and 89% for a land conflict issue. Thus, while people are very likely to go to the chief for a birth certificate, they are even more likely to go for a land issue. For 11% of the vignettes with birth certificates as the issue, people said the character would be a little likely to go to the chief, compared to only 6% for buying land and 4% for land conflict. For vignettes with birth certificate as the issue, 8% of the time, people said the character would not be very likely to go to the chief, compared to only 3% of the time for buying land and 4% for land conflict. Lastly, in only 2% of vignettes with birth certificate, people said the character would be not at all likely to go to the chief—the same as the 2% result for buying land and nearly the same as the 3% who said the same for land conflict. Thus, people are quite likely to go to the chief for any of these issues, but this seems to be especially true for land-related issues. When asked how likely the person would be to go to the mayor for these issues, there was less consensus. For nearly half (49%) of the vignettes that included a birth certificate issue, people said the person would be very likely to go to the mayor, compared to closer to a third for the land issues (34% for buying land and 37% for land conflict). People said that vignettes with a character with a birth certificate issue would be a little likely to go to the mayor 21% of the time, similar to those who wanted to buy land (23% of the time) and those with a land conflict (22% of the time). For 26% of the vignettes with a birth certificate issue, people said the character would not be very likely to go to the mayor, compared to 28% for buying land and 27% for those 79 vignettes with a land conflict. Lastly, people said the character would be not at all likely to go to the mayor for 6% of vignettes with a birth certificate issue, compared to 15% for either buying land or land conflict. Thus, there is a bit of a difference between the birth certificate issue and land issues with the likelihood of going to the mayor, though the difference is mainly at the degree to which the character is very likely to go to the mayor and less so with the lower levels of likelihood. Figure 4.5. Likelihood of going to the chief for help with an issue and likelihood of going to the mayor for help with an issue, by issue type 100% 2.1 4.9 6.7 1.9 7.9 10.5 1.5 3.2 5.9 2.9 3.7 4.0 86.3 79.7 89.5 89.4 80% 60% 40% 20% 0% 12.0 27.1 21.3 39.6 5.7 26.4 19.1 48.9 14.8 15.2 28.2 26.8 23.0 21.5 33.9 36.6 Overall Birth Certificate Buying Land Land Conflict Overall Birth Certificate Buying Land Land Conflict Chief Mayor Very likely A little likely Not very likely Not at all likely Lastly, the vignettes included a time dimension, gauging the difference between the vignette taking place recently or ten years ago. There was particularly little variation along this attribute. When the vignette took place recently, 86% of the time people said the person would be very likely to go to the chief, compared to 87% if the vignette took place ten years ago. Similarly, if the vignette took place recently, people said the character would be a little likely to go to the chief 7% of the time, compared to 6% if it took place ten years ago. People said the person would be not very likely to go to the chief only 5% of the time and not at all likely 2% of the time, regardless of whether the vignette took place recently or ten years ago. When asked how likely the character would be to go to the mayor, people said they would be very likely to do so 41% of the time if the vignette took place recently, compared to 39% if it occurred 10 years ago. People said the character would be a little likely to see the 80 mayor 20% of the time if the vignette took place recently, compared to 22% of the time if it took place ten years ago. For vignettes that were recent, people said the person would not be very likely to see the mayor 26% of the time, compared to 28% for vignettes that took place ten years ago. And finally, people said the person would be not at all likely to go to the mayor 13% of the time for vignettes that took place recently, compared to 11% of the time for vignettes that took place ten years ago. All in all, there is very little variation along this time dimension, suggesting that the changes put in place in Act III have not noticeably changed people’s perceptions of behavior. Figure 4.6. Likelihood of going to the chief for help with an issue and likelihood of going to the mayor for help with an issue, by timing of vignette 2.1 4.9 6.7 2.3 5.0 7.1 1.9 4.7 6.4 86.3 85.5 87.1 12.0 27.1 13.2 26.0 10.9 28.2 21.3 20.2 22.3 39.6 40.6 38.6 100% 80% 60% 40% 20% 0% Overall Recently Ten Years Ago Overall Recently Ten Years Ago Chief Mayor Very likely A little likely Not very likely Not at all likely Looking at these three outcome questions overall, it’s important to notice the differences in likelihood that someone would go to the chief compared to the mayor for help with an issue. When asked individually about going to the chief and the mayor for help, people overwhelming thought it was very likely that, regardless of attribute, people would see the chief for assistance. However, people were still reported that the character would be likely to see the mayor for help, as well. That said, when asked to pick who the character would most likely see, people overwhelmingly said the chief. This goes against the idea that chiefs are losing power or influence in people’s lives. It also suggests that even when people are given alternative avenues for local assistance, people still expect them to use the chief over the mayor. Indeed, even with 81 local governance and decentralization reform, chiefs continue to be important actors in the local context. 4.9.2 Regression Analysis: Model 1 How do these descriptive results hold up to statistical scrutiny? I next discuss the results of three OLS statistical models on the significance of the conjoint experiment. I first look at the results of Model 1, examining the outcome question, “Who is s/he most likely to go to for help solving this problem?” I then discuss Model 2, which shows the results from the outcome question, “How likely is it that s/he would ask for assistance from the chief for this problem?” followed by Model 3, which uses the last question, “How likely is it that s/he would ask for assistance from the mayor for this problem?” In Model 1, for three out of the four attributes, I find no statistical difference between the different attributes in the likelihood that they will visit the chief or the mayor. In this model, the dependent variable is coded 1 for chief and 0 for mayor, meaning that positive results indicate going to the chief, while negative results indicate going to the mayor. First, we see that the character’s gender has no impact on response. There is also no difference between those vignettes with a character born in the village and those where the character has lived there for several years. Lastly, there is no difference between whether the vignette occurred recently or ten years ago. As was seen in the frequencies presented earlier in this section, in all these cases, people believed the person in the vignette was much more likely to go to the chief than the mayor for assistance. The one exception is with the type of issue the character is having. For the vignettes with the land issues (buying land and land conflict), people said the character was more likely to go to the chief than those with a birth certificate issue. Even though people believe that those with a birth certificate issue are also likely to go to the chief, they think that someone with a land issue is even more likely to do so. I will discuss the implications of these results in turn. 82 Table 4.3. Effect of attributes on who the character would most likely go to for help solving their problem Awa Born in the Village Issue Buying Land Land Conflict Occurred Recently Female Respondent Age of Respondent Education Primary Secondary Lived Poverty Ethnicity Mande Other Constant Observations R-squared (1) 0.005 (0.018) -0.007 (0.016) 0.107*** (0.026) 0.111*** (0.025) 0.006 (0.017) 0.103*** (0.030) 0.000 (0.002) -0.008 (0.035) -0.121*** (0.040) -0.013 (0.008) -0.081* (0.045) -0.104* (0.054) 0.719*** (0.085) 2,634 0.053 Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Looking first at gender, regardless of whether the respondent received Awa or Mamadou as the name of the hypothetical character in the vignette, there is no significant difference in whether they thought the person would go to the chief or the mayor as their primary source for help. While this doesn’t align with my expectations, it does suggest that people do not see gender as a major distinction in how someone may decide to interact with local government officials, at least with regard to whether they will prefer the chief to the mayor. This result is consistent with my findings from Chapter 3, in which I found that while women were less likely than men to 83 contact chiefs or councilors overall, there was no difference for men and women in the gaps between going to the chief compared to the councilor. In other words, I found that the proportion of women seeking assistance from the chief compared to the councilor is similar to the proportion of men doing the same. Because the vignette here asks who Awa or Mamadou is more likely to see (and not whether they’re likely to see anyone), it makes sense that respondents wouldn’t perceive a difference between Awa and Mamadou in their likelihood of visiting either the chief or the mayor. Figure 4.7. Coefficient plot of attributes for who the character would most likely go to for help solving their problem Why might this be? The reasons for these findings mirror those discussed in Chapter 3. First, there may be less discrimination in the customary sector than expected. Perhaps the level of incorporation of the village chiefs into the deconcentrated system of governance provides enough motivation for village chiefs to provide a similar level of rational bureaucratic services to their people as the formal sector. Surprisingly, the majority of women in Senegal report never having experienced discrimination in the past year; however, this is more of an indicator of social desirability bias than an actual lack of discrimination. 84 Another possibility is that while discrimination exists in the customary sector, perhaps people don’t base their decisions of whom to go to on this. Perhaps instead, they choose to go to the chief or the mayor based on convenience or personal relationships. This could be especially true if there is also discrimination in the formal sector, as well. Again, as discussed in Chapter 3, women are less likely than men to report that the government is doing a good job promoting equal rights and opportunities for women. If there were discrimination in both sectors, then it would make even more sense that women could be prioritizing some other reasoning in selecting whom to go to—possibly custom or convenience—instead of trying to avoid discriminatory practices in the customary sector. Women often have fewer resources and less free time than men, so going to the chief may just be easier (Hyde et al. 2020). Lastly, while the conjoint did include gendered names in the vignettes to differentiate gendered behaviors, it’s possible that people’s responses did not take into account that women may act differently than men, even if in reality they do. This could mean that people were interpreting the question as whom should the person go to instead of whom would they go to. If the chief is the main person to see for these issues, then people might say either Awa or Mamadou would likely go to the chief, even in reality there is a difference in behavior. I don’t expect this to be the case, however, because the results here align so well with Chapter 3. If in reality, there were big differences in women’s and men’s behavior, we would have likely seen it in the Afrobaromter data, as well. Next, I look at the attributes for local or non-local status. Again, the attributes in the vignette to get at this factor were whether the character was born in the village or had lived there several years. I get similar results when looking at whether the character in the vignette was born in the village or moved there several years ago. I find that local/non-local status makes no difference in the perceived likelihood that the character in the vignette will seek help from the mayor versus the chief. This is unexpected, as I expected local and non-local status to affect how people chose to address their issues at the local level. This result also differs from my finding in Chapter 3 that minority ethnicities go to the councilor in a higher proportion than majority ethnicities. What explains this discrepancy? One possibility is that this attribute is not explicit enough to make a difference in the results, while minority and majority ethnicities, though blunt measures, are better at capturing the 85 idea of local/non-local status. Being born in a village or having lived there for several years is similar to—but not the same—as being a local or non-local to a village. For instance, if a woman marries into a “local” family in a nearby village from the one in which she is born, she would likely be considered local because she now belonged to a local family by marriage. For instance, in my experience in rural Mali (which is quite similar to rural Senegal) it was more common for women to marry men from nearby villages rather than to marry other families within the same village. However, these women would not be considered outsiders in the same way a migrant family who recently moved to a village would. Additionally, if a member of a local family moved to the city for work, but then moves home with their children who were born outside of the village, the children would still be local to the village because, again, they belong to a local family. In Senegal and across West Africa, it’s common for people to move to the city for work. For example, one of my research assistants had spent most of his career working for an American NGO in a city across the country from the village of his birth. When the covid pandemic hit, he lost his job and moved his family back to his home village to save money. His children were not considered outsiders when they returned, despite being born several hundred miles away. The kinship ties to the village are sufficient to determine local status. This differs from a family who does not have such kinship ties to the village moving in. In this instance, the family would be considered non-local or “outsiders,” even if they’ve lived there for many years. For instance, in Mali, in the village I lived in, there was a family that was still considered non-local, despite having lived there for decades, because they had moved in from another area of the country. The children of this family were not considered local, even though they were born in the village and the family had lived there for a long time. This dynamic likely applies more to more rural villages, as opposed to villages on the edge of expanding cities. The majority of the villages in this study are rural. While my theory looks specifically at this local/non-local dynamic, it’s possible that the attribute in the vignette isn’t clear enough to capture the difference and contributes to this finding. It’s possible that regardless of whether someone is native to the village or not, people don’t expect there to be a difference in how they behave when they have a local problem. As with gender, perhaps this is because there is no discrimination in the customary sector towards 86 non-locals. Senegal does have a cultural value of teraanga, or hospitality. This value is usually seen as an openness to welcoming guests. Perhaps chiefs feel compelled to treat non-locals well as a form of hospitality. This could discourage discrimination in the informal sector. As I discussed in Chapter 3, most minority ethnicities report never having experienced discrimination. However, this could be due to social desirability bias. Another potential explanation for the non-significance of this attribute is that there is discrimination against non-locals in the customary sector, but they face similar problems in the formal sector, as well. This seems unlikely, however. Discrimination by mayors against outsiders would seem relatively unlikely, because the mayor is much less likely to know who is an insider or an outsider in a given village compared to a village chief. The mayor of a rural commune is the mayor of several villages and is unlikely to know all the families in each, while the village chief is expected to know every family in his village and, to some extent, the histories of these families, as well. It’s also possible that the respondents don’t expect to see differences between locals and non-locals in the hypothetical vignette, even if such differences exist in reality. A local might presume that a non-local would act the way they would, even if that’s actually true. That is a drawback of this type of methodological approach; however, one would expect that if there were a big difference between how locals and non-locals act, that people would be familiar with it in general. While there are many possible reasons to explain the results for locals and non-locals in this chapter, much more research is needed to explore these possibilities. First, further research should work to determine a better measure of local/non-local status to ensure that future research is accurately capturing salient distinctions. Second, research should try to determine the extent of discrimination in both the formal and informal sector to understand how it may or may not affect people’s behavior. This is important because it’s long been thought that outsiders suffer from lower status at the local level (Lentz 2013) and thus, if that is the case, research to understand how it may affect their interactions with formal and informal institutions can provide insight into how to craft these institutions to better provide for such populations. Next, I look at the type of issue the person is having. Again, the respondent could get a character in need of a birth certificate, someone who would like to buy land, or someone with a 87 land conflict that they would like to solve. The results show a difference between both of the land-related issues and the birth certificate issue. If the character would like to buy land, they are seen as being much more likely to go to the chief than the mayor. This is particularly interesting because land issues are legally no longer officially the purview of the chief and haven’t been for some time.18 Power over land titling resides with the mayor (Interview with local councilor, Tambacounda Region, 7/29/2019). If someone wants to acquire land, the only legal path to ownership is through the mayor’s office. The mayor’s decision then gets approved by the sous- prefet (Interview with the secretary of the sous-prefecture, 7/29/2019). The mayor may involve a chief in the sale of a piece of land out of courtesy, but is not obligated to do so (Honig 2017). However, these results suggest that people are less accustomed to going to the mayor when they would like to buy land than they are for birth certificates. They still think the chief is the more likely place for someone to go for assistance on the matter, compared to getting a birth certificate. This suggests that even though the laws are changing and responsibilities have resided with the mayor for some time, people are not necessarily changing their habits of whom to see for help with such an issue. This is an important finding for understanding local governance in this type of hybrid context. When people have the choice between a formal and informal local actor, they may continue using the informal actor long after that official has lost the official power to deal with the issue. Decentralization policies should be taking this into account. The same holds for those who received a vignette with a land conflict issue. People are more likely to say the person would go to the chief than the mayor if they have a conflict over land than if they need a birth certificate. This makes sense and is in line with my expectations, as conflict resolution is a key component of the chiefs’ remaining responsibilities. Looking holistically at the issue results, however, it’s important to note that for all three issue types, people are more likely to say the character in the vignette would go to the chief over the mayor. As shown earlier in Figure 4.2, in 63% of vignettes with the birth certificate issue, people said the character would go to the chief over any other official, compared to 75% for those vignettes with either land issue. Thus, well over half the respondents are still choosing the 18 Since Act I of Decentralization of 1972, rural councils (now communes) have had power over rural lands (Faye 2008). 88 chief over the mayor for the birth certificate issue, even though they’re even more likely to suggest the chief for the land issues than the birth certificate. This suggests that regardless of the issue type, even with some differences between them, the chief is still the go-to person for these types of administrative tasks. Lastly, whether the vignette took place recently or ten years in the past does not affect who respondents think the person would see. There is no difference in the timing of the vignette on whether they would see the chief or the mayor. This is also a surprising result, as it signifies that Act III did not seem to significantly alter people’s behavior in terms of whom they go to locally for help with issues. One explanation is that people are not very good at remembering back ten years to determine how people’s behavior may have changed in that time. If that’s the case, then perhaps a conjoint experiment is not the best way to determine behavior change of this nature. However, in the absence of data to ascertain people’s perceptions both before and after the implementation of Act III of Decentralization, this approach is a justifiable way to examine change over time. It’s also possible that while there were changes in the division of power between the formal and informal sectors with decentralization reform, people are not well enough informed of these changes to alter behavior. This would make sense, as my survey found that only 40% of respondents had heard of Act III when asked. This suggests that much more can be done to inform local residents of changes in local governance structure. It also suggests that any reforms designed to improve the functioning of local government need to incorporate information programs for local populations to understand the changes in law. If this were done for Act III, there is little evidence for it in my survey. A final possibility is that Act III did not fundamentally change the structure of local governance, despite the reforms. This is quite likely the case, based on interviews I conducted in fall of 2021. I was told, “Act III was rushed” (Interview with local development agent, Kolda Region, 12/1/2021); “Act III hasn’t had a real impact” (Interview with secretary general of an opposition party, Kolda Region, 11/30/2021); “Act III is only a theory” (Interview with opposition candidate for mayor, Kolda Region, 11/30/2021); and, “Act III: the intention was good, but there are parts that aren’t respected” (Interview with the secretary to the mayor, Kolda Region, 11/29/2021). An adjoint to the mayor told me, “Act III is a great document. The problem 89 is the application” (Kolda Region, 11/29/2021). Multiple people told me one of the main issues with Act III is that while the law transferred competencies to the mayor, it did not transfer the means to implement projects in these areas (Interview with secretary to the mayor, Kolda Region, 11/29/2021; Interview with member of civil society organization, Kolda Region, 11/30/2021). This suggests that while Act III is the law in the books, it may not be making a discernable difference on the ground due to issues with implementation. Thus, it’s reasonable that without actual changes on the ground, people would not be changing their behavior in response. For the controls, female respondents are more likely than male respondents to say the person in the vignette would go to the chief compared to the mayor. This could suggest that women are more inclined to see the chief than the mayor themselves, though this is not what I find in Chapter 3. In fact, when self-reporting having contacted the chief or mayor, women are less likely than men to say they contacted either the chief or the mayor. However, women are more likely to trust the chief than men are. When comparing means for trust in chief, the rate for women (2.434) is significantly higher than the rate for men (2.349) (t = -2.722, p = 0.007). Women could be more familiar with the chief, since the chief resides in their village Women are also often less likely to have the means or the time to travel to a mayor who is not located in their village, as was discussed in Chapter 3. For these reasons, women may be more likely to recommend going to the chief in this hypothetical situation of the vignette than men are. Alternatively, women may lack of knowledge about who the proper authority is for a given problem, leading them to recommend the chief even when the mayor is the better option. Respondents with a secondary education are more likely to suggest the person go to the mayor than those with less than a primary education. There is no difference between those who obtained a primary education and those with less than a primary education. This suggests that, with more education, people are more likely to go to the mayor than the chief. This aligns with the results from Chapter 3, in which those with at least a secondary education were more likely to contact their councilor than those with less than a primary education, but there was no difference between those with a primary education and less than a primary education. Perhaps this means that those with a secondary education or higher have a better understanding of the various responsibilities of local governance actors. Indeed, in my survey, those with at least a 90 secondary education were more likely than those with less than a primary education to say they had heard of Act III of Decentralization19. The mean for those with at least a secondary education (0.74) is statistically higher than that for those with less than a primary education (0.259) (t = - 23.645, p = 0.000). Of those who have heard of Act III, those with at least a secondary education were more likely to say they had a better understanding of the law than those with less than a primary education20. The mean for those with at least a secondary education (1.768) is statistically higher than those with less than a primary education (1.011) (t = -11.11, p = 0.000). It’s also possible that those with more education are more confident to seek help from the mayor than those with less education. It could also mean that those with at least a secondary education have more means to seek help from the mayor over the chief, though by controlling for lived poverty, this is unlikely the reason for the difference. Regarding the ethnicity of the respondent, those who identify as Mande or other are less likely than those who identify as Pulaar to say the character should go to the chief; however, the significance of these variables is just outside the 95% threshold (p = 0.07 for Mande and p = 0.054 for Other). I find no difference in the age or the level of lived poverty of the respondent in whether they think the character should go to the chief or the mayor. 4.9.3 Regression Analysis: Models 2 and 3 Next, I look at the other two outcome questions: “How likely is it that s/he would ask for assistance from the chief for this problem?” and “How likely is it that s/he would ask for assistance from the mayor for this problem?” These two questions are a bit different from the previous one because they’re not asking whom the character in the vignette is most likely to see, but rather their perceived independent probabilities of seeing the chief and mayor. Again, the structure of these questions allows for the possibility that people see both sets of actors, or neither, which might reflect reality for many Senegalese. Table 4.4 shows the results for Model 2 (likelihood of seeing the chief) and Model 3 (likelihood of seeing the mayor). 19 Respondents were asked, “Have you heard of Act III?” Response options were “yes” and “no.” 20 Respondents were asked, “How much do you understand about Act III?” Response options included, “nothing,” “a little,” “some,” and “a lot.” 91 Table 4.4. Effects of attributes on the likelihood of the character seeing the chief or seeing the mayor for help solving their problem (2) Likelihood of seeing chief (3) Likelihood of seeing mayor Awa Born in the Village Occurred Recently Issue Buying Land Land Conflict Female Respondent Age of Respondent Education Primary Secondary Lived Poverty Ethnicity Mande Other Constant Observations R-squared -0.064** (0.026) 0.024 (0.024) -0.026 (0.024) 0.163*** (0.035) 0.123*** (0.037) 0.101** (0.039) -0.001 (0.002) 0.034 (0.046) -0.069 (0.052) -0.029** (0.012) 0.145*** (0.040) 0.032 (0.066) 3.793*** (0.118) 2,749 0.039 -0.072* (0.041) 0.004 (0.038) -0.011 (0.039) -0.361*** (0.057) -0.327*** (0.058) 0.016 (0.071) 0.006 (0.004) 0.036 (0.086) 0.443*** (0.088) 0.013 (0.019) 0.433*** (0.091) 0.421*** (0.103) 2.619*** (0.208) 2,743 0.084 Note: Robust standard errors in parentheses. For the issue variable, the reference category is “birth certificate;” for the education variable, the reference category is “no formal education;” for the ethnicity variable, the reference category is “Pulaar.” *** p<0.01, ** p<0.05, * p<0.1. In Model 2, when respondents are asked if the character in the vignette would go to the chief, there is a significant difference between whether respondents think Awa or Mamadou would go to the chief for their problem. Those whose vignette featured the name Awa said she was significantly less likely to see the chief for this problem than those who received Mamadou. 92 This suggests that subjects expect a woman to be less likely to see their chief for a problem than a man. In Model 3, when subjects were asked if the character would go to the mayor, the difference between the Awa or Mamadou characters in the vignette was significant, but only at the 0.08 level. Thus, when the gender of the character is female, people are less likely to say she should go to the chief or the mayor. This is in line with results from Chapter 3, in which women are found to be less likely to see either the chief or the councilor than men. In line with previous results, local or non-local status does not affect someone’s likelihood of going to the chief or the mayor. There is no difference in responses between those who received a character born in the village or one who has lived there several years in their likelihood of seeking help from either the chief or the mayor. When the character had a land issue in the vignette—either wanting to buy land or having a land conflict—subjects reported them to be significantly more likely to go to the chief and significantly less likely to go to the mayor for help with their issue, compared to those who wanted to obtain a birth certificate. This is in line with previous results, which showed those with land issues to be more likely to go to the chief than the mayor. Lastly, the element of time still has no effect on people’s decisions on whom to see for local issues. Those with a vignette that happened recently were no more likely to see the chief or the mayor than those with a vignette that happened ten years prior. Again, this suggests that Act III of Decentralization has not had a discernable impact on who people seek for help with their problems. When looking at whom the character is most likely to see, compared to whether they’re likely to see the chief/mayor, there are some changes in the control variables. Female respondents are more likely than male respondents to say there is a high likelihood the character would go to the chief, but there is no difference between female and male respondents in the likelihood the character would go to the mayor. This is in line with earlier results, in which female respondents are more likely to recommend the chief over the mayor. They are, however, slightly different from the results in Chapter 3, in which women are less likely than men to see either the chief or the mayor themselves. Those with at least a secondary education are more likely than those with less than a primary education to say that the character would go to the mayor, but there is no difference 93 between those two groups in saying that the character would go to the chief. Perhaps this suggests that regardless of education, people have a good understanding of when someone would go to the chief, but those with more education are more likely to also understand when someone would go to the mayor. This is in line with earlier results. Respondents with higher lived poverty are less likely than those with lower lived poverty to say the character would go to the chief. There is no relationship, however, between poverty levels and saying the character would see the mayor. This result goes against expectations that those with higher poverty would be more likely to recommend seeing the chief. This result requires further research for better understanding. Lastly, when looking at ethnicity, when asked about the character’s likelihood of going to the chief or the councilor, Mande respondents are more likely than Pulaar respondents to say the character would go to the chief. When asked if the character would go to the chief or the mayor, Mande respondents were also more likely to say the character would go to the chief, but at a much lower level of significance. It’s possible the lower significance for the initial question is due to the smaller sample size for Mande respondents. Because I did not expect ethnicity to play a big role in people’s decision making, I did not design my sample to have as much ethnic variation as would be helpful for such an analysis. That said, the Mande are a small minority group in Senegal. The Pulaar are a minority group overall in Senegal but are the dominant group in the regions in which this survey was conducted, particularly the Kolda region.21 Thus, it is surprising that Mande would be more likely to say the person would go to the chief than Pulaar. This could be a cultural difference between the two groups. It could also suggest that minority ethnicities are more likely to feel the need to go to the chief for assistance than majority ethnicities. Mande are also more likely than Pulaar to say that the character would go to the mayor for assistance. The “other” category for ethnic groups are also more likely to say the person should go to the mayor than Pulaar respondents. These results combined suggest that groups that are minorities in a given area are more likely to say the character should get help in the formal system than groups that are majorities. The result for the “other” groups could suggest that those 21 According to Afrobarometer, Mande make up roughly 7% of the population of Senegal, while Pulaar make up 29% of the population. 94 who are minority ethnicities in a given area are more likely to experience discrimination by the customary sector and instead are more inclined to recommend the formal sector. However, without a difference in their likelihood to recommend seeing the chief, we cannot be overly confident in this interpretation. 4.9.4 Exploratory Analysis Finally, in light of these findings, I also conducted some exploratory analyses to further understand the effects of subjects’ gender and ethnicity on their responses. I did not design this study to be adequately powered to discern heterogeneous treatment effects; thus, these results should be read with caution. First, I interact the gender of the respondent with the gendered name of the character they received in the vignette, Awa or Mamadou. Predictive margins are shown in Figure 4.8. While I saw no difference in how female and male respondents expected Awa to act, female respondents were significantly more likely than male respondents to say Mamadou would most likely to go to the chief for his problem compared to the mayor. This provides yet another example of how women are more likely to favor the chief than the mayor; however, in this instance, women favor the chief only in their expectations of how men will act. Again, women could be less educated about what type of problems the mayor could help with. Thus, maybe men and women have a good understanding of how women will act when they have a problem, but women are less sure about how men will act. This could be especially true if the responsibilities of getting a birth certificate, buying land, or solving a land conflict are more likely to be issues men solve. Regardless, there is a difference in how men and women expect others to use local government resources. This is important for any policy development addressing decentralized local power. Next, I look more closely at ethnicity. When looking at the regression results for the three outcome questions, there was some evidence of ethnicity playing a role in people’s choices. Again, the analysis was not designed to look into ethnicity; however, upon reviewing the results, it appears to have more of an impact that I would have expected. In most areas, there is little evidence that Pulaar are much different from Mande respondents. However, I do find some differences when looking specifically at the attribute about the length of time the character has lived in the village. This attribute is designed to get at whether the character has local or non- local status. 95 Figure 4.8. Predictive margins of character gender and respondent gender Figure 4.9. Predictive margins of the length of time the character has lived in the village and respondent ethnicity I ran an interaction between being Mande and the born-in-village attribute. As seen in Figure 4.9, the interaction shows that when the character in the vignette was not born in the 96 village but has lived there several years, Mande respondents are significantly more likely than Pulaar respondents to say that the character in the vignette will go to the chief for assistance. This did not hold when the person in the vignette was born in the village: there is no difference between Mande and Pulaar then. While much more research into this would be needed to fully understand the difference here, the fact that Mande are a minority ethnicity both in Senegal and in the Tambacoumba Region, where the majority of the Mande in my survey live could explain this difference, as was discussed earlier. To be clear, this does not make them non-local in the sense that I have described earlier. Mande have lived in Senegal since the pre-colonial era (Wilfahrt 2022). However, perhaps their minority status gives them the sense that those who are not local should be careful to check in with the chief when there are problems, at least more so than Pulaar, who make up a greater portion of the Senegalese population. If this is the case, then perhaps the opposite of my theory is true: that minority ethnicities think non-locals would be more likely to go to the chief with their problems than locals. This could be a way for minority ethnicities to maintain good relations with majority ethnicities in the area. It could also just be a cultural difference between the two groups. More research into this difference is necessary to gain a better understanding of these differences. 4.10 Conclusion This chapter extended the analysis from Chapter 3 to further understand how people choose who to seek for assistance at the local level. The results align well with previous results and expand our understanding around the importance of the type of issue and ethnicity of the subject. The most consistent finding in this chapter is that land issues are still heavily regarded as the purview of the chief, despite village chiefs not having legal authority over land and only an advisory role compared to the mayor and local council. This is important because while laws have changed who holds this power, people’s behavior does not seem to have changed in this area. This could leave many people vulnerable to losing their land if they are hesitant to use the legal channels to obtain land. It could also suggest that while the mayor holds this power, that formally titling land is out of reach for most rural Senegalese, particularly in the southern and western regions. 97 There are also differences along gender lines: when the character is named Awa, people are less likely to say she would go to the chief or the mayor, compared to when the character is named Mamadou. This aligns with findings in Chapter 3 that women are less likely than men to contact the chief or the local councilor. I did not, however, find any differences between locals and non-locals or in the timing that the vignette occurred. These attributes may be less important than I expected, but more research is needed to better understand how they affect which local actor people seek for assistance. All in all, Chapters 3 and 4 combine to tell a story of imagined continued power among village chiefs, despite their decrease in formal power at the state level. When people have a choice of who to go to for help, many groups still continue to choose the chief, though we find some variation across groups. Yet, is this a story about Senegal or does it extend to other contexts across Africa? Chiefs are common actors in local governance in many African countries, but the structures of chieftaincies and the institutional incorporation of chiefs into the state can vary quite widely. In Chapter 5, I will compare data from Senegal with that from Ghana and Malawi to see how groups may change their behavior in different institutional contexts. This will introduce institutional variation into my analysis and will provide some insight into how generalizable these results are. 98 CHAPTER 5: INSTITUTIONAL STRUCTURE AND USING FORMAL AND INFORMAL INSTITUTIONS 5.1 Introduction This chapter introduces country-level variance in institutional structure to the study of how people use formal and informal local government institutions. In previous chapters, I look specifically at Senegal, exploring how identity and issue type can affect how people decide which sector of local government they use when they have a problem. The analysis in Chapter 3 uses Afrobarometer data to explore how identity can affect whom people go to for help, focusing on gender and local/non-local status. I find that women are actually not different from men in their tendency to contact either the chief or the councilor: both women and men favored the chief over the councilor quite a bit. Men were, however, more likely to seek assistance overall. I also find that non-locals are less likely to contact the chief, but not different from locals in contacting the councilor. Additionally, the gap between contacting the chief and contacting the councilor is smaller for non-locals than for locals, suggesting that those with non-local status have a different pattern of using the formal and informal local institutions from locals. Chapter 4 expands on the analysis in Chapter 3 and uses novel data from a conjoint experiment to look at the effect of gender and local/non-local status, as well as issue type, on whom people contact in times of need. Chapter 4 finds little difference between the different identity types: men and women, locals and non-locals are all perceived to go to the chief more than the councilor. Issue area, however, is a significant factor in whom people seek for assistance. Those with land issues are perceived as going to the chief at much higher rates than those with more administrative issues. This suggests that historical responsibilities for chiefs are still a strong determinant in how people use the institution even today. Here in Chapter 5, I move away from my focus on only Senegal and add Ghana and Malawi to my analysis. I focus on how national-level institutional structure affects the relationship between formal and informal leaders. More specifically, I examine how the degree to which chiefs are integrated into local councils affects people’s choices on whom to visit for help with a problem. In a setting with high integration of chiefs into councils, chiefs sit as formal members of the council, as is the case in Malawi. At the other end of the spectrum, chiefs are not members of the council and are not involved in the decision-making of the formal local 99 institutions, as is the case in Ghana. Senegal sits somewhere in the middle, with chiefs not holding seats on the council, but still attending some meetings and working with the council when necessary. Exploring the variation across these cases is important, as it begins to show how much institutional structures can shape the dynamics on the ground in local government. This, in turn, can influence how people use the institutions available to them. The findings from the analysis show that institutional structure matters. In Malawi, a highly integrated system, people are more likely to contact the chief over the councilor for assistance. The gap between contacting the chief and the councilor is bigger than in Senegal and favors the chief. In Ghana, a less integrated system, people are more likely to contact the councilor instead of the chief. That said, the gap between seeing the chief and seeing the councilor is smaller than in Senegal, contrary to expectations. Thus, the lack of integration increases the rate at which people seek out the councilor but doesn’t necessarily decrease the rate at which people seek out the chief. This suggests that chiefs are still important local actors in Ghana, but they are less of a substitute for formal actors than in the more integrated systems. These findings show the implications of various types of decentralization policies and the ways they deal with the dual systems of authority at the local level. They also suggest that revised decentralization policies should take into account how the structure will affect the ways people use the systems. In this chapter, I first outline what we know about the relationship between chiefs and local governments, before discussing the variations in formal and informal local governing systems across Senegal, Ghana, and Malawi and outlining my theoretical expectations. I then discuss my methodological choices before presenting the results from my analysis and their implications. 5.2 Chieftaincy and Local Government in Senegal, Ghana, and Malawi In this chapter, I explore how variation in institutional choices affects the way people use local formal and informal institutions. In previous chapters, I looked more specifically at individual factors such as identity and issue area to understand how people use local institutions. Here, I look cross nationally at the degree of integration of chiefs into formal local governments. Thus, in this section, I discuss some of the literature on the relationship between formal and 100 informal local government actors. I then give some background on the local government and chieftaincy systems in Senegal, Ghana, and Malawi. Much of what we know about the relationship between formal local government and the chieftaincy across Africa focuses on whether chiefs cooperate or compete with local officials. Many studies frame the relationship between chiefs and local government actors as adversarial. For example, Ribot (2003) writes, “Because chiefly authority is undermined by the transfer of control over land allocation and other legitimating powers to democratic institutions, chiefs and their allies in central government—who may also lose their rural power base through local democratization—pose a serious threat to decentralization” (21). The argument is that chiefs, in other words, seek to maintain their power at the local level by undermining and minimizing the authority of local elected officials. Baldwin (2016) also notes that chiefs often oppose the formation of democratic formal institutions at the local level. Clayton, Noveck, and Levi (2016) even find that in Sierra Leone, close relationships between the chief and the local councilors resulted in decreased public goods provision, while higher rates of conflict between chiefs and local councils yielded higher public goods provision, suggesting that this adversarial relationship may even help improve local outcomes. In contrast, others find that cooperation between the chief and local actors is beneficial. Fearon, Humphreys, and Weinstein (2009), for instance, find that in Liberia, when chiefs cooperate with local development committees, social cohesion increases in post-conflict villages. Of course, it is unlikely to be so cut and dry that chiefs either always compete or always cooperate in the same way across various institutional settings. Kern, Holzinger, and Kromrey (2024) look at the extent that chiefs are legally integrated into the state. They expect that increases in the integration of chiefs into the state will lead to lower levels of political conflict between chiefs and the state. However, in their comparative case study of Kenya, Namibia, Tanzania, and Uganda, they only find support for this theory in Namibia. Henn (2023) also explores the issue of chiefs’ power across institutional settings, by looking at the effect of state capacity and level of institutionalization of the chieftaincy on the relationship between local officials and chiefs. He finds that in stronger states with higher levels of institutionalization, local officials can supplement chiefs, while in weaker states with less institutionalization, local officials and chiefs complement each other. Yet, Henn measures institutionalization by whether 101 the chieftaincy is included in the constitution or not. This misses the variation in levels of institutionalization across states and codes states such as Senegal, with legal integration outside the constitution, as the same as states lacking any integration at all. I argue that the level of institutionalization and integration of chiefs into the state is an important aspect of the chieftaincy. Chiefs can be integrated at the local level by serving on the local council, which gives them a seat at the table when decisions are made. On the other hand, the chieftaincy can exist in a separate parallel system, with limited interaction with local elected officials and minimal overlap of responsibilities. Much of the literature focuses on the relationship between chiefs and the state, yet we still know little about how these levels of integration affect the way people use these formal and informal institutions at the local level. This is an important factor in how local governance works and how people experience the state at the level closest to them. Thus, in this chapter, I look at variation in the levels of chiefs’ integration into the formal system at the local level and its effect on how people interact with formal and informal local institutions. In other words, does the extent to which chiefs have been incorporated into the local councils affect which institution people choose to contact when they have a problem? In this chapter, I compare Senegal with two additional cases: Ghana and Malawi. The relationship between local government and traditional chiefs varies quite a bit across all three countries. With Ghana at one end of the spectrum, chiefs and local councils are institutionally quite separate at the local level, as chiefs do not serve on the council and have little overlap in responsibilities. In Senegal, chiefs have an official role at the local level and a moderate amount of involvement with the local council, despite not holding seats on the council. Village chiefs in Senegal often have a working relationship with the council, wherein they attend council meetings when asked and work with the formal local government as necessary. They also have several duties that overlap with the council. Malawi sits at the other end of the spectrum from Ghana. There, chiefs are much more integrated into the formal system than in Senegal and especially more than in Ghana. Chiefs sit on the council and serve in an advisory role for local government. In this section, I’ll discuss the institutional structures of local government and the chieftaincy in all three countries and expand on my theoretical expectations for each. 102 5.2.1 Senegal I’ll start with Senegal, as its institutional structure has been discussed extensively in previous chapters. Senegal is a very good country for examining the relationship between formal and informal institutions. Local councils and village chiefs each have responsibilities over local governance in Senegal. An elected mayor is the head of the council, which is otherwise comprised of several elected local councilors. Councils have jurisdiction over several competencies, including primary education; local health centers; agriculture and livestock production; water resources; and land management, among others (Wilfahrt 2022). Councils also maintain public records such as registering marriages, births, and deaths within the jurisdiction and providing titles and certificates to citizens. The chieftaincy in Senegal, as in much of Africa, while rooted in precolonial governance structures, exists today in a much different form from the past. During colonization, the French took pieces of traditional authority structures and crafted them into a more uniform administrative structure much removed from what existed before French rule. This can be seen through the development of the canton chiefs, a position created by the French wherein indigenous Senegalese, often with some aristocratic lineage, were appointed to work under the French authorities in rural areas, mainly collecting taxes and enlisting labor (Wilfahrt 2022). As the canton chiefs worked to enforce French colonial policy, they lost legitimacy and were ultimately abolished upon independence (Keese 2011). Village chiefs, however, were much less threatening to authorities due to their limited power at the village level and their increased legitimacy on the ground since they were less influenced by French rule (Wilfahrt 2022). Today, village chiefs in Senegal are also considered part of the state, serving as the lowest level of the deconcentrated system of governance.22 Since independence, the chieftaincy in Senegal has not been structured hierarchically, which means village chiefs are particularly close to the population they represent. Chiefs act as a liaison between the village and other officials, such as the mayor and the local councilors. They resolve disputes in the village, maintain some 22 The deconcentrated system in Senegal exists parallel to the decentralized system of governance. The deconcentrated system is comprised of appointed officials at each level of government, while the decentralized system is comprised of either directly or indirectly elected officials (Vengroff and Johnston 1989). See Chapter 3 for more information. 103 degree of informal power over customary land, and can also provide some administrative services to villagers, such as providing birth certificates and collecting certain types of taxes. There is often some cooperation between chiefs and the council, but chiefs do not have a formal position on the council. It’s common for chiefs to go to the local council for meetings, but some chiefs admit that the councils tend to make decisions without considering the opinion of village chiefs. As one chief told me, “They call us for meetings, but they don’t want our advice” (Interview with village chief, Kaffrine Region, 08/09/2019). Other chiefs describe having close relationships with the mayor and the council. One chief told me, “Whenever there’s a problem or a conflict, I call [the councilors] and we deliberate. We’re very much in touch” (Interview with village chief, Tambacounda Region, 7/30/2019). Thus, chiefs are somewhat integrated into the formal system, but not fully. They have a formal role in the administration but are not fully incorporated into the decision-making of the formal government. Yet, in Senegal, chiefs’ responsibilities overlap with the responsibilities in the formal system, especially regarding administrative tasks such as obtaining birth and death certificates and issues around land tenure and conflict. This overlap in responsibility can give citizens a choice on who to go to with their problem. As I’ve shown in previous chapters, this overlap may lead to variation in how people interact with local government. 5.2.2 Ghana In Ghana, the institutional structure of formal and informal local governance differs from Senegal in some key ways. The center of local government in Ghana is the District Assembly (DA).23 Popularly elected Assemblymembers, who are directly elected from single-member districts, comprise about 70% of the DA, while roughly 30% of members are appointed by the president. The district’s member of Parliament also serves on the assembly as a non-voting member. The appointed members are often selected along party lines, even though, legally, none of the members are allowed to be explicitly partisan (Debra 2022). This contrasts with Senegal’s system, in which all councilmembers are popularly elected, and appointed officials remain in the parallel deconcentrated system without a role on the council. Senegal’s local councilors are also 23 District Assemblies come in three types: Metropolitan, Municipal, and District Assemblies. Metropolitan Assemblies are for larger urban areas, Municipal Assemblies are for smaller urban areas, and District Assemblies are for rural areas. All three types of assemblies have the same functions (Debra 2022). For clarity, I refer to all assemblies of this type as District Assemblies. 104 elected using a mixed proportionality-plurality electoral system, compared to the single-member system in Ghana. Thus, partisanship is much more of a feature in Senegal’s local councils than in Ghana. As in Senegal, Ghanaian DAs have responsibility over many sectors at the local level, including development and planning; social services such as education and health; natural resources, conservation, and forestry; public works and infrastructure; public safety; and industry and trade. They also provide some administrative services and collect some taxes (Friedrich Ebert Stiftung 2010). However, the DAs comprise larger geographic units than local councils in Senegal. In Ghana, there is a level of local government below the District Assembly called the Unit Committee; however, this lowest level has much less power than the District Assembly and serves primarily to enforce the DA’s decisions and mobilize citizens for the DA when necessary (Friedrich Ebert Stiftung 2010). In Senegal, the local council is the lowest form of decentralized governance. The chieftaincy varies between Senegal and Ghana, as well. Chieftaincy in Ghana under British colonial rule was somewhat similar to that under the French in Senegal. Many existing structures of the chieftaincy were empowered by the British to maintain order and collect taxes (Debra 2022). Britain’s approach to indirect rule gave chiefs more power than France’s approach in Senegal, as the chiefs replaced the need for local British administrators. However, after independence, the Nkrumah administration sought to limit the power of the chiefs by further institutionalizing them and putting boundaries around chiefs’ power. For instance, the state now had the power to remove chiefs from their positions (Knierzinger 2011). They also empowered local councils to take over some of the development responsibilities that chiefs had enjoyed (Bonoff 2016). Chiefs’ power ebbed and flowed during subsequent administrations but continued to be institutionalized and controlled by the state, which minimized their autonomy and their role over time. The structure of the chieftaincy in Ghana is hierarchical even today, compared to Senegal where once canton chiefs were abolished, all that was left were village chiefs with no customary power above them. The constitution today outlines five types of chieftaincy: Paramount chiefs sit at the top of the hierarchy, followed by Divisional Chiefs, then Sub-Divisional Chiefs, then Adikrofo, and finally a catch-all category of other chiefs (Knierzinger 2011). The lowest level of 105 the hierarchy is the village headman, which serves a similar role as the village chief in Senegal. Due to the hierarchy of chiefs, higher-level chiefs in Ghana enjoy more power than village chiefs in Senegal, however the chieftaincy overall is more sidelined than in Senegal in order to keep that power in check. Today, chiefs’ responsibilities include managing natural resources and stool land, resolving conflict, and implementing customary law (Sackey 2012). According to Wilfahrt and Letsa (2023), both chiefs and villagers in Ghana rated dispute resolution as the chief’s most important responsibility. Chiefs also play a role in development, though Wilfarht and Letsa (2023) find this to be lower of the list of responsibilities compared to dispute resolution, despite its prominence in the literature on chiefs. DAs have some responsibility over these types of issues, as well, including conflict resolution and natural resource management. However, where these responsibilities overlap, there is often little cooperation between chiefs and DAs. Instead of working together, chiefs are, for instance, not included in local security committees (Debra 2022). Instead, they operate parallel to each other on many of these issues. For instance, chiefs do not serve on DAs and are even forbidden from participating in political activities or advocating partisan viewpoints (Bonoff 2016). Thus, like in Senegal, there is a division between the formal and informal sectors of local governance; however, in Ghana, the roles of chiefs are kept even more separate from the roles of assemblymen and women at the local level. This is one of the main institutional differences between Ghana and Senegal at the local level. In Senegal, chiefs are not part of the local council, but there is a norm of including them in certain aspects of local governance, such as advising on land issues. Chiefs attend meetings at the mayor’s office on issues that pertain to them (Interview with village chief, Tambacounda Region, 7/27/2019). In Ghana, I have found no evidence of similar meetings at the DAs. Thus, while lower-tier chiefs and the DAs do maintain overlapping roles on certain issues, they tend to operate in more separate spheres of influence than in Senegal.24 24 While the chieftaincy is not incorporated into formal local government, Ghana does have a National House of Chiefs, as well as ten regional houses of chiefs for the higher levels of the chieftaincy. In these deliberative venues, the upper levels of chiefs are responsible for “maintaining and updating customary law” (Bonoff 2016). However, while the Houses of Chiefs do incorporate chiefs into governance in a way that does not exist in Senegal, these institutions are at such a higher level of government, that they have little effect on life at the local level in Ghana (Debra 2022). 106 Because of the variation in the level of integration of chiefs into local councils, people in Ghana and Senegal are likely to show different patterns of frequenting the chief and the councilor. In Ghana, chiefs are less integrated into the formal government at the local level compared to Senegal. Chiefs in Ghana do not sit on the council and are less incorporated into council business than chiefs in Senegal. While chiefs in Senegal are also not formal members of the council, they are better positioned to act as an intermediary between the population and the council and are included in some meetings and some administrative business. Therefore, I expect more people in Ghana to go directly to the councilor for assistance compared to people in Senegal, while fewer people will go to the chief. Ghana has both a more professionalized formal local government system, with more overall power given to the DAs than to the local councils in Senegal, while also having a more sidelined chieftaincy compared to village chiefs in Senegal. These two factors create an environment where people have less of a choice between the two systems in whom they must see to address a problem. As a result, not only will more people frequent the councilor than the chief in Ghana, but I expect the gap between those contacting the councilor and those contacting the chief in Ghana to be larger than the gap in Senegal. I also expect the gap in Ghana will favor the councilor, while the gap in Senegal will favor the chief. Again, due to the lower levels of integration in Ghana than in Senegal, I expect people to have more reason to go directly to the councilor for assistance, while in Senegal people will seek help from the chief over the councilor when possible. Therefore, my first hypothesis states: H1a: In Ghana, people will be less likely to contact the chief than in Senegal. H1b: In Ghana, people will be more likely to contact the councilor than in Senegal. H1c: In Ghana, the gap between visiting the chief and the local councilor will be larger than the gap for Senegal and will favor the councilor. 5.2.3 Malawi In contrast to Senegal and Ghana, Malawi’s system of local government incorporates chiefs to a much greater extent. Similar to Ghana, the local councils in Malawi are comprised of elected ward councilors, as well as other non-elected positions. Ward councilors are elected from single-member districts and may be affiliated with a political party but are not required to have partisan ties. Mayors are then indirectly elected by ward councilors, which is a system similar to 107 that of Senegal’s until 2022. The non-elected positions include MPs within the district, appointed representatives of special interest groups, and chiefs (Chiweza and Msiska 2020). Those representing special interest groups and chiefs are non-voting members of the council. Local councils in Malawi have purview over several different types of issues, including public works and infrastructure; agriculture and livestock; water resources; economic development; and peace and security. They also have administrative responsibilities, such as registering births, deaths, and marriages (Ministry of Local Government and Rural Development 2013). While chiefs’ power in Malawi has ebbed and flowed to some extent over time, chiefs have been integrated into the state since the colonial era. Similar to Ghana, chiefs were given legal recognition in Malawi through the British colonial government. As Chiweza (2007) writes, “The colonial regime wanted to incorporate indigenous structures into units of governance in order to fill an administrative gap” (55). Chiefs were put under the supervision of the District Commissioner, a type of British administrator (Kayira and Banda 2018). They were responsible for collecting taxes, keeping local records, maintaining law and order, and constructing roads. In this capacity, chiefs were the main structure of local governance in this era. Incorporating chiefs into the colonial system, however, also served to keep chiefs under British control (Chiweza 2007). After independence, in contrast to Ghana, chiefs maintained much of their power. As Kim (2022) writes, “While the ruling party operatives during the one-party regime usurped traditional leaders to some degree, local leaders remained influential and strengthened their influence in rural areas after the multiparty transition” (25). When local councils were created, chiefs were incorporated into the structure, and chiefs kept most of the same responsibilities they had under colonial rule. By using this structure, the central state was able to minimize the power of the local councils by continuing to work through the chiefs (Kayira and Banda 2018). This allowed the central state to retain their control over all areas of formal government. In other words, by working through the chiefs, the central government limited the ability of local elected officials to rise up the ranks and challenge the center. Chiefs’ roles have not officially changed since the 1967 Chiefs Act. This act leaves a fair amount of ambiguity around the role of the chief. Chiefs are charged with maintaining the peace, administering customary law, and collecting certain taxes (Kayuni et al. 2019). In practice, chiefs are heavily involved in local development, as well, including mobilizing local labor and serving 108 as a liaison between development agents and the community. As Power (2020) writes, “When villagers are born, when they die, when they need a passport, a school, a clinic, their first stop is the village headman. If they need development funds, they go to the chief or the senior chief. Some of the functions of the chieftaincy then are traditional, some colonial throwbacks, some distinctly modern (or even post-modern), but people still regard chiefs as central to their lives, as tribunes, guardians, and gatekeepers” (281). Similar to Ghana, chiefs in Malawi are hierarchical, with a paramount chief at the top of the hierarchy, followed by senior chiefs, sub-chiefs, group village headmen, and finally village headmen at the bottom (Kim 2022). However, chiefs at the lower end of the hierarchy are more likely to interact with villagers than those at the top. These are the leaders that people are likely to go to initially with an issue (Kim 2022). The Malawi case is one of much more integration of formal and informal systems than in Senegal and Ghana. The inclusion of local chiefs explicitly on the council integrates chiefs into the local council and gives chiefs a much larger role in local decision-making. Chiefs can also serve as the main point of contact for people who need help from the local council. This is different from either Senegal or Ghana, because while chiefs can operate as intermediaries in all three countries, they are insiders on the council in Malawi and not in the other two contexts. Local councils in Malawi have also been sidelined by the central state more than in Senegal or Ghana. From 2005 to 2014, local council elections were postponed, and only administrative staff served at the local level at that time (Chiweza and Msiska 2020). This left a vacuum at the local level, which chiefs happily filled (Hussein and Sambo 2022). As a result, local councils have had less time in Malawi to gain popular legitimacy, while chiefs’ role has not substantially changed since independence. This makes chiefs much more central to local governance and much more powerful on local councils in Malawi than in Senegal or Ghana today. Because chiefs are highly integrated on the councils, people are more likely to frequent the chief in Malawi than in Senegal. While the chief can assist in some aspects of local government in Senegal, chiefs in Malawi have a broader mandate, and their role on the council gives them access that chiefs in Senegal do not have. I expect people to contact their councilor less in Malawi than in Senegal and to contact their chief more because chiefs are a more consistent, direct line to the council, whereas in Senegal, the councilor has more independent 109 power to solve people’s problems. I also expect the gap between contacting the chief and contacting the councilor to be larger in Malawi than in Senegal. This is because in Malawi, the chief can serve as the main point of contact on the council, while in Senegal, the chief can serve as an intermediary, but isn’t included in all council business—in other words, the chief is more integrated onto the council in Malawi than in Senegal. Thus, people in Senegal will be more likely to go to the chief for some things and the councilor other things, while in Malawi they can go to the chief for a broader array of problems. I expect the gap to favor the chief in Malawi, as well, the same as in Senegal. This is because both countries’ integration of the formal and informal allows people to go to the chief for assistance if they have an issue under the purview of the chieftaincy or an issue the council could assist with. Therefore, my second hypothesis states: H2a: In Malawi, people will be more likely to contact the chief than in Senegal. H2b: In Malawi, people will be less likely to contact the councilor than in Senegal. H2c: In Malawi, the gap between visiting the chief and the local councilor will be larger than the gap in Senegal and will favor the chief. I discuss the methodological approach and each of the variables used in my analysis in the next section. 5.3 Methods In this analysis, I use Afrobarometer data from Senegal and Ghana to explore variation in use of formal or informal local institutions across the three countries. I use Senegal and Ghana rounds 4, 6, 7, and 8, and Malawi rounds 3, 4, 7, and 8. Rounds 1, 2, and 5 did not contain the contact variable for either traditional leaders, local councilors, or both, so those rounds were excluded. Round 6 in Malawi also did not include the contact variable for local councilors. For the dependent variables, I use the same two variables as were used in Chapter 3: two questions about people’s frequency of contacting various officials. One asks about contacting traditional leaders and the other about contacting local councilors.25 Because I’m exploring 25 The question asks, “Over the past year, have you contacted any of the following persons about some important problem or to give them your views?” The question is asked for several political actors, including Members of Parliament, political party officials, local councilors, and traditional and religious leaders. The response options for these questions include, “never,” “only once,” “a few times,” and “often.” 110 country-level variation in this chapter, the independent variable is country, denoting Senegal, Ghana, or Malawi. In addition to these two independent variables, I use a number of control variables in the analysis, as was done in Chapter 3. First, I control for gender, which is a binary variable for female and male. I expect women to be less likely to contact either chiefs or councilors than men, as cultural norms across the three countries likely encourage men to tackle these types of tasks compared to women. Next, I control for whether the person is local or non-local to the community. I expect locals, or those with stronger ties to the village, to be less likely to contact the chief than those who are non-local, or those who are seen as outsiders. Locals will likely feel more comfortable using the customary institutions because they are accustomed to them and more familiar with the norms around them. Non-locals may be more likely to use formal institutions, though depending on the composition of the formal institutions, they may feel less comfortable there, as well. To capture local/non-local status, I create a variable for minority ethnicity. This variable is constructed in the same way as in Chapter 3, using the Ethnic Power Relations (EPR) dataset (Wucherpfenning et al. 2011) to code dominant ethnicities in each geographic area of the country. Then, I used the geocoded Afrobarometer data on ethnicity to code each observation as either minority or majority within the corresponding EPR geographic area.26 As discussed in Chapter 3, there are some downsides to this approach. The main issue is that the geographic areas delimited in the EPR data can be quite large and, thus, may not capture the nuance of local/non-local status on the ground. Someone may be a local in their village or area, but still be coded as a minority ethnicity because they are not part of the dominant ethnicity in the broader region. In other words, while an area delimited in the EPR may have a dominant ethnicity overall, it may code other groups as minorities despite localized power dynamics within the larger region. Similarly, if an ethnic minority lives in a village that primarily consists of their same ethnicity, they may be considered locals despite their minority status in the overall region. 26 For Senegal, the majority ethnicities in the EPR data include Wolof, Pulaar, Serer, Mande, and Diola. For Ghana, the majority ethnicities include Akan, Ewe, Ga-Adangbe, and Northern Groups.26 For Malawi, the majority ethnicities are divided into three regions: Northerners, Central, and Southerners. Northern majority ethnicities include Tumbuka, Tonga, and Ngonde; Chewa is the only majority ethnicity for the Central region; and Lomwe, Mang’anja, Nyanja26, and Yao are all coded as majority ethnicities in the Southern region in the EPR data. Any ethnicities aside from these found in each geographic region from the EPR were coded as minority ethnicities. 111 In contrast, someone that is a minority ethnicity in a village composed primarily of a majority ethnicity would more likely be considered a non-local. However, in this coding scheme, both would be coded as a minority ethnicity, despite their differing lived experiences. Despite this lack of nuance, this measure of minority ethnicity status is the best possible way to measure local/non-local status given the data available. It allows us to see overall patterns of minority/majority ethnic status and gain insight into how people choose between formal and informal institutions for solving their problems. I also control for trust in political parties. Trust may influence how someone interacts with formal and informal local government, as those who don’t trust or support the party of an elected official may be less inclined to seek help from them. Thus, trust in one party over another may affect how someone interacts with a local council. In these instances, chiefs may be a preferred option because they may be seen as less overtly partisan. To control for the partisanship of elected officials, I create the trust in political parties variable by combining two Afrobarometer questions about trust in political leaders: trust in ruling party and trust in opposition. The questions in Afrobarometer ask for both the ruling party and the opposition political parties, “How much do you trust each of the following, or haven’t you heard enough to say?” The response options include “not at all,” “just a little,” “somewhat,” and “a lot.” I create the new trust variable by coding those who reported trusting both the ruling party and the opposition “not at all” or “just a little” as trusting neither. Those who reported trusting the ruling party “somewhat” or “a lot,” but trusted opposition parties “not at all” or “just a little” were coded as trusting the ruling party. Those who reported trusting the opposition “somewhat” or “a lot,” but trusted the ruling party “not at all” or “just a little” were coded as trusting opposition parties. Finally, those who reported trusting both the ruling party and the opposition parties “somewhat” or “a lot” were coded as trusting both the ruling party and the opposition. Unfortunately, I do not have data on the political leaning of governments at the local level in Senegal, Ghana, or Malawi. In Senegal, the overwhelming majority of local governments are run by the ruling party. This is not the case in Ghana because local councils are non-partisan, making the political leanings of local councilors much harder to discern. This means that the trust variable is much harder to interpret in this chapter. Knowing that someone supports the ruling party at the national level doesn’t necessarily predict how they will interact with the local 112 council. However, I include this variable in this analysis for two reasons. First, including it keeps consistency with the analysis in Chapter 3. Second, I do expect that those with higher trust in institutions generally may be more likely to contact both the chief and the councilor. Next, I control for whether the subject lives in an urban or rural area, as rural areas are expected to have more influence from chiefs than urban areas (Baldwin 2016). Lastly, I control for education, lived poverty27, and age, as well. My analysis here mirrors that of Chapter 3. I again use a difference-in-differences approach to explore how institutional structure affects the way people use local institutions. I compare the gap between the frequency of contacting the chief and the frequency of contacting the councilor in Ghana to that of Senegal and then do the same for Malawi. I include all the control variables above and use round fixed effects in my analysis. This type of analysis allows me to focus first on which type of institution is generally favored in each country, and second, by how much the institution is favored. By examining the gap between frequency of contacting the chief and the frequency of contacting the councilor in each country, we can understand both how people interact with these institutions and also the distribution of power in each system. This approach gives us a useful comparison into how people use these institutions across the three cases. 5.4 Results In this section, I will go through my results. I first present the frequencies of contacting leaders across countries. I then use regression analysis to present these differences across countries. Lastly, I present the difference-in-difference analysis to compare the gap in contacting the chief or the councilor in Ghana and Malawi to Senegal. In Chapter 3, I showed that most people in Senegal contact neither the chief nor the councilor. I find a similar situation in Ghana and, to a lesser extent, in Malawi. In Senegal, only 27 The lived poverty variable is an index of five Afrobarometer questions, which ask, “Over the past year, how often, if ever, have you or anyone in your family: Gone without [enough food to eat/enough clean water for home use/medicines or medical treatment/enough fuel to cook your food/a cash income]? The responses include, “Never,” “Just once or twice,” “Several times,” “Many times,” and “Always.” The responses are coded from 0 to 4, and then the response from each of the five variables are added together for each observation. The index variable has possible values ranging from 0 (responses of “Never” on all five questions) to 20 (responses of “Always,” on all five questions). 113 30% have contacted the chief and only 22% have contacted the councilor in the past year. In Ghana, more people report having contacted the councilor than the chief, with only 26% having contacted the chief, but 30% having contacted the councilor. Thus, a similar percentage of people in Ghana contacted the councilor as contacted the chief in Senegal. In Malawi, people favor the chief much more than the councilor, with 43% contacting the chief and only 17% contacting the councilor. It’s likely that, across all three countries, most people in a given year don’t have any reason to contact either type of official. Figure 5.1. Frequency of contacting the chief and the councilor, by country 100% 80% 60% 40% 20% 0% 13.1 11.7 5.5 69.8 6.8 8.8 6.3 10.1 10.3 5.1 78.1 74.5 9.4 13.0 7.4 70.2 4.0 5.7 7.0 83.3 21.9 13.4 7.5 57.3 Chief Councilor Chief Councilor Chief Councilor Senegal Ghana Malawi Never Only once A few times Often Figure 5.1 shows the frequency with which people report contacting the chief and the councilor in Senegal, Ghana, and Malawi. The variables indicate how often the respondents have visited each type of local official in the past year. The response options are never, only once, a few times, and often. In Ghana, three quarters of respondents report not having contacted the chief in the past year, compared to 70% in Senegal and only 57% in Malawi. Five percent of people in Ghana and Senegal contacted the chief only once, which is similar to the 7% who contacted the chief in Malawi. Ten percent of respondents in Ghana reported contacting the chief a few times, compared to 12% in Senegal and 13% in Malawi. Similarly, 10% reported contacting the chief often in Ghana, compared to 13% in Senegal and 22% in Malawi. Thus, of the three countries, Malawians appear more likely to contact the chief than Senegalese, while Ghanaians appear less likely to do so. 114 While Malawians are more likely to contact the chief than Ghanaians or Senegalese, Ghanaians are more likely to contact the councilor. In Ghana, 70% of respondents said they never contacted the councilor, compared to 78% in Senegal and 83% in Malawi. There were similar reports of contacting the councilor only once across all three countries, with 7% saying so in Ghana, compared to 6% in Senegal and 7% in Malawi. Thirteen percent of respondents in Ghana reported contacting the councilor a few times, compared to 12% in Senegal and only 6% in Malawi. Lastly, 9% reported contacting the councilor often in Ghana, compared to 7% in Senegal and only 4% in Malawi. Ghanaians appear more likely to contact the councilor than Senegalese, while Malawians appear much less likely to do so. Figure 5.2. Mean values for contacting the chief (0-4 scale), by country (with 95% confidence intervals) 1.2 1 0.8 0.6 0.4 Senegal Ghana Malawi These frequencies, however, do not show us if these differences are significant. Figures 5.2 and 5.3 show the means of the two contact variables for each country with 95% confidence intervals around the means. Figure 5.2 shows the means of contacting the chief for Senegal, Ghana, and Malawi. As you can see, people are indeed significantly more likely to contact the chief in Malawi than in Senegal, while people are significantly less likely to contact the chief in Ghana than in Senegal. This is in line with my expectations that the more integrated the chief is with the local council, the more people will contact the chief. 115 Figure 5.3 shows the means of contacting the councilor for Senegal, Ghana, and Malawi. This figure shows the opposite pattern as Figure 5.2: people are significantly more likely to contact the councilor in Ghana than in Senegal and significantly less likely to contact the councilor in Malawi than in Senegal. This, too, follows my expectations that the more integrated chiefs are into the formal system, the fewer people will contact the councilor. Figure 5.3. Mean values for contacting the councilor (0-4 scale), by country (with 95% confidence intervals) 0.7 0.6 0.5 0.4 0.3 0.2 Senegal Ghana Malawi These results provide some evidence to support my theory, however, they are still limited. To examine this relationship further, it is important to include possible confounders in the analysis. Next, I run an ordinal logit for each of my dependent variables: contacting the chief and contacting the councilor. My independent variable is country, using Senegal as a baseline. I also include control variables for gender, minority ethnicity, urban or rural location, trust in the ruling party and opposition, level of education, lived poverty, and age. I include round fixed effects, as well. 116 Table 5.1. Effect of country on contacting chiefs and councilors (1) Contacting chief (2) Contacting councilor Country Malawi Ghana Female Minority ethnicity Rural Trust Trust opposition Trust ruling party Trust both Education Primary Secondary Post-secondary Lived poverty Age 0.300*** (0.049) -0.203*** (0.047) -0.684*** (0.034) -0.077** (0.036) 0.914*** (0.040) 0.344*** (0.050) 0.257*** (0.046) 0.210*** (0.045) -0.060 (0.047) 0.001 (0.049) 0.122* (0.071) 0.034*** (0.004) -0.000 (0.000) -0.652*** (0.059) 0.516*** (0.049) -0.731*** (0.037) -0.148*** (0.038) 0.479*** (0.041) 0.330*** (0.055) 0.321*** (0.050) 0.277*** (0.048) 0.101* (0.053) 0.270*** (0.053) 0.578*** (0.071) 0.031*** (0.004) 0.000 (0.000) 17,220 Observations Note: Standard errors in parentheses. Includes round fixed effects. For the country variable, the reference category is “Senegal;” for the trust variable, the reference category is “trust neither;” for the education variable, the reference category is “no formal education.” *** p<0.01, ** p<0.05, * p<0.1. 17,333 These models, seen in Table 5.1, show whether the relationship shown in Figures 5.1, 5.2, and 5.3 still holds once controlling for confounding factors. Model 1 looks at the frequency of contacting the chief. I find that in Malawi, people are significantly more likely to contact the chief compared to people in Senegal, while those in Ghana are significantly less likely to contact the chief than those in Senegal. This is in line with my expectations in H1a that integration of chiefs into the formal system will result in increased contact with chiefs, as well as my 117 expectations in H2a that a lack of integration of chiefs will result in increased contact with councilors. Model 2 looks at the frequency of contacting the councilor. Here, too, the results are in line with expectations for H1b and H2b. In Malawi, people are significantly less likely to contact the councilor than people in Senegal, while in Ghana, they are significantly more likely to contact the councilor than those in Senegal. This shows that the general pattern of contact across these three countries is in line with my theory. To explore the variation in the patterns of use of these formal and informal local institutions, it’s important to understand not just whether people frequent these institutions at different rates in different countries, but also how the pattern of frequenting them varies, as well. To see this, I use a difference-in-differences test using OLS to see whether the gap between seeing the chief and seeing the councilor differs across the three countries. The results are shown below in Table 5.2. The difference-in-differences estimates are shown by interacting the frequency of contacting either the chief or the councilor with a variable that denotes either Ghana or Senegal, as seen in column 1. I then use the same format shown in column 2 with a variable that denotes either Malawi or Senegal. This approach allows us to see both which institution is favored in each country and by how much. For H1c, I find that people do favor the councilor in Ghana, while they favor the chief in Senegal, as expected. Again, though, I also expected that the gap between contacting the councilor and chief—the strength of people’s preferences for the councilor over the chief— would be larger than that same gap in Senegal, because the lack of integration in Ghana and the corresponding weakening of chiefs would cause more people to seek assistance from the council and fewer to seek assistance from the chief. In Senegal, the distinction between these two groups in terms of power is less stark, which I expected would lead to comparatively smaller gaps (i.e. people’s favoritism towards one group over another would not be as stark). However, I found the opposite: the gap between contacting the chief and contacting the councilor in Ghana is actually smaller than the gap in Senegal. In other words, while people in Ghana do favor the councilor, the extent to which they favor the councilor is actually smaller than the extent to which Senegalese favor the chief over the councilor. 118 Table 5.2. Difference-in-differences estimates, by country Chief or Councilor Ghana Chief or Councilor x Ghana Malawi Chief or Councilor x Malawi Female Minority ethnicity Rural Trust Trust opposition Trust ruling party Trust both Education Primary Secondary Post-secondary Lived poverty Age Constant (1) Ghana 0.235*** (0.019) 0.212*** (0.020) -0.294*** (0.025) -0.342*** (0.013) -0.050*** (0.013) 0.321*** (0.014) 0.139*** (0.019) 0.154*** (0.017) 0.116*** (0.016) -0.012 (0.018) 0.035** (0.018) 0.146*** (0.024) 0.014*** (0.002) 0.001*** (0.000) (0.000) 0.213*** (0.034) (2) Malawi 0.235*** (0.020) -0.290*** (0.023) 0.467*** (0.029) -0.229*** (0.015) -0.046*** (0.016) 0.282*** (0.018) 0.133*** (0.021) 0.127*** (0.020) 0.085*** (0.020) 0.048** (0.020) 0.092*** (0.022) 0.141*** (0.033) 0.014*** (0.002) -0.000 (0.000) 0.172*** (0.033) Observations R-squared Note: Standard errors in parentheses. Includes round fixed effects. For the trust variable, the reference category is “trust neither;” for the education variable, the reference category is “no formal education.” *** p<0.01, ** p<0.05, * p<0.1. 19,371 0.103 25,765 0.074 119 This suggests that although chiefs have been sidelined in Ghana and are not formally part of the council, they still appear to play a big enough role in local governance that people seek them out for assistance at similar rates as the formal officials. In Senegal, on the other hand, councilors are not nearly as popular as chiefs, resulting in a larger gap. The direction of the relationship, however, is in line with my expectations, with more people contacting the councilor in Ghana, while more people contact the chief in Senegal. This suggests that the professionalized local council does attract more interaction from local populations in Ghana than the comparable local councils in Senegal. For H2c, I find that the gap between contacting the chief and contacting the councilor in Malawi is larger than the gap in Senegal and that people favor the chief over the councilor. This means that people in Malawi are much more likely to contact the chief than the councilor compared to people in Senegal. This makes sense, as chiefs are much more integrated into the council in Malawi and can therefore more easily act as an intermediary for local population in Malawi than in Senegal. This finding supports my second hypothesis and suggests that the institutional structure may affect how people interact with formal and informal institutions. 5.5 Discussion In this chapter, I explore how variation in institutional structure can affect the way people use formal and informal local institutions. More specifically, I look at the degree to which chiefs are integrated into formal local institutions in Senegal, Ghana, and Malawi and how that affects the distribution of power in these systems. In all three countries, colonial and post-colonial central governments made choices that have had lasting impacts on the way formal and informal local institutions are structured to serve people. Colonial governments worked through chiefs in all three countries, cementing the informal institutional structures of the chieftaincy into useful administrators for the colonial regime. In Ghana and Malawi, this institutionalized the hierarchical structure to the chieftaincy that remains more or less intact today (Kim 2022, Knierzinger 2011). In Senegal, this meant that the upper level of chieftaincy—the canton chief— was left weak and easy to dissolve upon independence (Wilfahrt 2022). In the post-independence era, Ghana sought to placate the upper level of chiefs, while sidelining the remaining chiefs and keeping them out of political positions in the local government (Debra 2022). In Malawi, in 120 contrast, post-independence governments kept chiefs in power, favoring them over other local officials (Kayira and Banda 2018). This is a structure that remains today. These legacies have shaped the way local governments are structured in each country, causing variation in the degree to which chiefs are integrated into formal local government institutions. I theorized that this variation in the level of integration affects the way people use the local institutions available to them. In Malawi, the chiefs are more integrated into the local council than in Senegal and as a result, I expected to find a larger gap between people contacting the chief and contacting the councilor compared to the gap in Senegal. Indeed, Malawians overwhelmingly prefer to contact the chief when they need assistance, compared to the councilor. The gap in contacting the chief and contacting the councilor in Malawi is much larger than the gap in Senegal. This all suggests that national institutional choices have a meaningful impact on the ways in which people use local institutions for solving everyday problems. In Ghana, chiefs are less integrated into the local government compared to Senegal. I, therefore, expected to find a larger gap between visiting the chief and the local councilor in Ghana than in Senegal, as more people would seek the councilor than the chief. Instead, I found a smaller gap between the chief and the councilor, meaning people frequent these two institutions at more similar rates in Ghana than in Senegal. What could explain this finding? One explanation could be that while Ghanaian chiefs are not formally integrated onto the council, as they are in Malawi, chiefs may still maintain close informal ties to local councilors, meaning they can serve a similar role as chiefs in Senegal, even with less overt overlap in responsibilities. If this is the case, chiefs may still have a voice on the council, even if they are not formally provided one. This would give people a reason to continue seeking assistance from the chief for their problems, particularly when chiefs can act as an intermediary between the local population and the council. This explanation would suggest that formal institutional rules have less impact on the relationship between chiefs and councilors than I expected. We could better understand the nuance of the relationship between chiefs and councilors by using interviews instead of survey data to gain insight into the informal relationships between chiefs and councilors. Another explanation could be that the roles of chiefs and councilors are more distinct in Ghana than in Senegal, meaning the responsibilities of the two sectors in Ghana have much less 121 overlap than in Senegal. If this is the case, and the responsibilities are somewhat evenly divided, it could mean that people simply go to the chief frequently for the issues under the chief’s purview and go to the council for those issues under the council’s purview. Therefore, the smaller gap in Ghana is showing the chiefs are still important and useful to local populations, just for issues that are separate from those issues the council can help with. Further research could focus much more on the effect of overlap of responsibilities in how people use local institutions instead of focusing on the level of integration of chiefs into formal institutions. This could be achieved by conducting interviews on not only whether people contacted the various local officials, but also what types of issues they contacted the officials for and how that may vary across institutional settings. For both of these possibilities, more research is needed to better understand the relationship between chiefs, local government officials, and local populations. One area where we need a much better understanding is into how the structure of the chieftaincy affects the way people use the institution. Research on the role of chiefs in Ghana specifically is often more ambiguous than that of Senegal because the chieftaincy is more complex in Ghana than in Senegal. The influence of a paramount chief is likely to be much stronger than that of a village headman, but people’s access to paramount chiefs is likely to be much more limited than their access to a local headman. Research on chiefs rarely specifies which level of chieftaincy it is discussing, making it difficult to understand the various points in the system to which local people have access. This level of nuance in Malawi and Ghana is beyond the scope of this project, but further research should examine these nuances to better understand which parts of the chieftaincy people tend to contact and what relationship those parts of the chieftaincy have with the local council. Regardless of the reason, the fact that people still contact chiefs at high rates in Ghana, even when they are less integrated into formal local institutions suggests that chiefs remain highly important actors in local government, even when central states have tried to reign them in. This finding deserves further study into the nuances of the dynamic between local officials and chiefs in Ghana. 122 5.6 Conclusion In this chapter, I used Afrobarometer data to explore variation in formal and informal local institutional structure and its effect on how people use local institutions. I looked at three cases, comparing structures in Ghana and Malawi to that of Senegal. Across the three cases, the level of integration of chiefs into local councils varied, which I expected would affect the way people used the local institutions available to them. I found support for my theory when comparing Malawi to Senegal: in Malawi, where chiefs are highly integrated, people are much more likely to contact the chief compared to the councilor, and the extent to which they favor chiefs over councilors is larger than Senegalese citizens’ favoritism toward chiefs. This suggests that the more of a role chiefs play in local government, the more people will contact them with their problems. In Ghana, however, where chiefs are less integrated into the council and where councils are more professionalized, I found that people actually favor councilors more than chiefs, a pattern that is different from what we observed in Malawi and Senegal. However, chiefs are still quite important at the local level. People continue to contact chiefs for assistance, despite efforts to sideline their power over the years. In Chapters 3 and 4, I looked specifically at the relationship between chiefs and formal local officials in Senegal. By expanding my analysis to two other cases, I am able to show that these relationships between local actors are quite influenced by local institutional rules in addition to the individual-level variables I explored in previous chapters, such as gender and local/non-local status. All in all, in this chapter, I find evidence for variation in how people use formal and informal local institutions across these three countries. This suggests that the ways that countries structure their local institutions will have an impact on how people use them. It also suggests that understanding formal local governments is not complete without also understanding informal local governments—the ways people seek assistance from local councils is influenced by the availability of other avenues to solve problems, especially the chieftaincy. 123 CHAPTER 6: CONCLUSION Across Africa, local government systems often include aspects of formal and informal government, with varying degrees of overlap. Often, the formal government consists of a local council, while the informal system is made up of chiefs. When these two systems overlap, people have choices as to whom they go to for assistance when they have a problem. However, we know little about how people make these choices. Do some people favor one type of local institution over the other? Do some people prefer formal leaders, while others prefer the customary? Are there situations where one type of institution is more appropriate than the other? On the one hand, we could imagine that as the formal system is developed, people will increasingly prefer that system over the informal one, as it has the backing of the rational, bureaucratic state to provide fair and equitable services compared to the informal sector (Weber 1978). In other words, when people are given a choice, some predict that people will move away from the chieftaincy to the more democratic and accountable local elected officials (Ribot 2003). On the other hand, many have found chiefs to be incredibly resilient. Chiefs are well trusted and maintain high levels of legitimacy across many contexts in Africa (Logan 2013). Thus, we can also imagine that people will continue to prefer the customary sector, even when other choices are available. In this dissertation, I have argued that more nuanced theories are needed in understanding how people use the formal and informal institutions at the local level. Different types of people may have different reasons to go to the chief or the local councilor, while the same person may use one institution in some situations, while using the other in another situation. More specifically, I have looked at nuances along three dimensions: identity, issue area, and institutional structure. Regarding identity, I examined the effect of gender and local/non-local status on who people contact in times of need. I expected these identities to affect such choices because customary institutions are more likely to operate under systems of bias and discrimination compared to the formal sector, which could encourage women and non-locals to favor the formal sector more than men and locals when they need assistance. Issue area is another area that may affect how people use formal and informal local institutions. Some issues are more the purview of the formal sector, while others are in the wheelhouse of the informal sector. Therefore, people will decide who to go to based on how 124 helpful that official is likely to be for the specific issue they need assistance with. For example, I expected administrative issues such as acquiring a birth certificate to be the purview of the formal sector, while people would use the chief more for issues around conflict resolution. Lastly, I expected institutional structure to affect people’s choices of who to see for assistance. In other words, I expected to find variation across institutional contexts in how people made these choices. The more integrated the chief was into the local council, the more expected people to choose the chief over the councilor. In contrast, when chiefs were not very integrated into the local council, the more I expected people to choose the councilor over the chief. These three theories around how people choose which institution to use at the local level all get at different dimensions of people’s choice calculus. Again, this shows that these decisions are complex and that variation in how these choices will be made is expected across and within groups, depending on circumstances. To test my theory in Chapter 3, I used Afrobarometer data to explore how identity affected the way people choose which institution to use when they have a problem. I looked specifically at the effect of gender and local/non-local status on which institution people contact. I expected women and non-locals to be more inclined to contact the formal sector compared to men and locals because they are more likely to face bias and discrimination in the customary sector over the formal sector. I used difference-in-differences analysis to explore the gaps across different groups in contacting the chief or contacting the councilor. I found that women actually have the same size tendency as men to go to the chief over the councilor, though men are more likely to see assistance overall. Non-locals, however, are less likely to contact the chief than locals, but show no difference in contacting the councilor. Also, the gap in contacting the chief and contacting the councilor for non-locals is smaller than the gap for locals. In Chapter 4, I continued my analysis into how identity affects the way people use local institutions, while also incorporating issue area into the analysis, as well. This chapter used novel data from a conjoint experiment to examine these issues. The use of novel survey data was useful in that I could better match terminology and questions to the local context in rural Senegal. The conjoint experiment used a vignette to ask respondents about a hypothetical situation and whether the character in the vignette would contact the village chief or the local mayor. Regarding identity type, I found little difference between the men and women or local/non-local 125 status and people’s proclivities to go to the chief or the mayor. In fact, most people expected the character in the vignette to choose the chief over the mayor, regardless of gender or local/non- local status. However, for issue area, I did find a difference between needing a birth certificate and both land issues (buying land or having a land conflict). People were much more likely to expect the character in the vignette to see the chief for land issues than for administrative ones. Chapter 5 broadened the scope of the dissertation to compare the local government systems across three African countries: Senegal, Ghana, and Malawi. I used Afrobarometer data again to explore this question, with much of the analysis mirroring what was done in Chapter 3. The main focus of this chapter was on the effect of chiefs’ integration into the formal system on how people use the formal and informal sectors. Malawi is a case where chiefs are highly integrated into the formal system, and I expected people to be more likely to contact the chief in Malawi compared to Senegal. I also expected the gap between visiting the chief and the councilor to be larger in Malawi than in Senegal and to favor the chief. Ghana, on the other hand, is a case where chiefs are not very integrated into the formal system, so I expected people to be more likely to contact the councilor in Ghana compared to Senegal. I also expected the gap between contacting the chief and the councilor to be larger in Ghana than in Senegal, but to favor the councilor. I found support for my theory in Malawi, where the gap was indeed larger and favoring the chief. However, in Ghana, I found that while people were more likely to contact the councilor than the chief in Ghana, the gap was actually smaller, suggesting that the chiefs are still widely used despite the increase of use in the formal sector. These findings overall show that chiefs are still quite popular across multiple contexts. In Senegal, despite the likelihood of discrimination in the customary sector, women do not have a greater tendency to use the formal system than men. In fact, men and women are both more likely to see the chief than the councilor. Non-locals are also more likely to see the chief than the councilor, though are less likely to contact the chief than locals. This suggests that chiefs are still incredibly important to people’s relationship with local government in Senegal. That all said, there are variations in how people use the formal and informal local government systems, and these nuances are important in developing our understanding of how these institutions operate at the local level. The chieftaincy is a diverse institution across contexts, both within countries and across countries. This work seeks to begin outlining some of 126 the variation in how people use this institution, especially when they have other options available to them. As with any project, there are several limitations to this dissertation. First, because this project primarily focuses on one case, the findings from Chapters 3 and 4 may be somewhat Senegal-specific. This is an issue with much of the research that looks at chiefs, as much of it examines a single case. However, the nuance of the chieftaincy across contexts makes cross- country comparisons difficult. The comparison of Senegal, Ghana, and Malawi in Chapter 5 attempts to address this, but still suffers from the limitations of such a small number of cases. Future research should continue to find ways to conduct large-N analyses to better understand the broad patterns of chieftaincy across context. That said, the Senegal case does share similarities with other francophone countries across West Africa. I would expect the results to travel reasonably well to Mali, Burkina Faso, and Cote d’Ivoire. Again, more research can help us better understand the degree to which people’s choice calculus changes across institutional contexts. Another limitation is geographic variation within Senegal. While Chapters 3 and 5 use Afrobarometer data, which is nationally representative, Chapter 4 focuses on three departments in southeast Senegal. These areas are good for understanding the relationship between the formal and informal local context because they are fairly rural and not overly dominated by religious leaders such as marabouts. In other areas of Senegal, the village chief is likely less important, while marabouts hold more political power and may compete or usurp the role of the chief in helping people solve their problems at the local level. By avoiding the areas dominated by marabouts, I can draw a simpler comparison between chiefs and local councils; however, the results of this chapter may be less applicable to areas around Touba where the Mouride brotherhood are dominant. Another limitation of this study is the focus on quantitative methods. While I conducted several elite interviews to better understand the relationship between the formal and informal sectors around Senegal, more qualitative work can be done to understand how people make choices in whom to contact when they have a problem. Indeed, interviews and focus groups with different groups of citizens would expand our understanding of how people make choices on whom to contact when they have a problem. 127 Considering these limitations, I recommend some additional avenues of future research. First, one thing that was made clear throughout this dissertation was that people still overwhelmingly use chiefs when they need help with a problem. This makes sense in some respects, particularly in Senegal, where village chiefs may be more convenient or accessible than local councilors or mayors since village chiefs operate at the village level, but not every village has its own representative on the local council. However, little has been written about the barriers that people face in using the formal system. Thus, it’s possible that people would prefer to use the formal system more than they do, but face challenges in accessing it. Because mayors are in charge of a whole commune in Senegal, they serve a much higher number of constituents than village chiefs do. That means that wait times to see the mayor can be high. Mayors are also often absent, traveling to the capital frequently or even living far from their constituents. These obstacles might make the village chief a more reliable actor in times of need, even if chiefs have less power than other local officials overall. Regardless, future research could examine these barriers more closely to understand better what factors people are taking into account when they seek assistance at the local level. Another avenue of future research could focus more closely on the ways that power is divided between chiefs and local councils. The nature of the overlap in powers may affect how people choose to use the formal and informal local institutions, so a cross-national analysis of the various ways that responsibilities are divided at the local level would go a long way in furthering our understanding in this area. In Senegal, for instance, many of the powers that overlap between chiefs and local councils are not formal divisions, but rather informal ones. This means that any study of this topic would need to incorporate more qualitative research into outlining where the responsibilities overlap in reality, compared to what is outlined in formal decentralization policy. Chapter 5 shows that institutional structure matters for how people interact with local institutions. However, it is limited to only looking at three case countries. There is a lot of variation in how local government is structured across Africa, but we would benefit from future research that looks more broadly across a number of countries to better understand the patterns of use across contexts. Finally, the systems of chieftaincy vary quite a bit across contexts. For example, Ghana and Malawi have hierarchical systems of chieftaincy, while Senegal, along with some other West 128 African countries, have only a single-tier system of village chiefs. The differences in these structures likely affect how people use the customary system. More could be understood about how people use the hierarchical chieftaincies—how do the different levels of chiefs interact with each other; what types of services do the different levels of chiefs provide; when discussing the power of the chieftaincy, how much power is held at each rung? Answers to these questions would contribute quite a bit to understanding the nuance in how people use the customary system when the chieftaincy is hierarchical. There is also a key policy implication of this work. Formal and informal local institutions are the government actors that people interact with the most. When people have everyday problems, they don’t turn to the president or their member of parliament. Instead, their first step is likely to be to go to the chief or to their local councilor for help. As states continue to refine their decentralization policies, understanding the way different types of people interact with these systems can allow policymakers to create systems that better serve all groups. Ultimately, the way power is distributed between the formal and informal sectors at the local level is up to the central government. Decentralization policy can be refined to ensure that local systems serve all groups. Making the formal system more accessible in rural areas may help certain groups access the services that the formal system provides. Likewise, further formalizing chiefs’ roles may help ensure that customary biases do not limit some groups’ access to the services provided by the informal sector. In sum, this dissertation shows the continued resiliency of the chieftaincy. People continue to contact the chief with their issues, despite parallel institutions often serving similar functions at the local level. This suggests that thirty years after democratization across Africa, chiefs continue to provide valuable services to their communities. While more can be learned in this area, especially around the scope of the services that chiefs provide and the extent to which powers are divided between the formal and informal sectors, this project shows that while there is some variation across groups, people still continue to choose the chief over other local actors when they have a problem. 129 BIBLIOGRAPHY Adida, Claire, Ferree Karen, Dan Posner, and Amanda Robinson. 2016. “Who’s Asking? Interviewer Co-Ethnicity Effect in African Survey Data.” Comparative Political Studies 49(12): 1630-1660. Alexandre, Pierre. 1970. West African Chiefs: Their Changing Status under Colonial Rule and Independence. New York: Africana Publishing. Amoateng, Acheampong Yaw, Tim B. Heaton, and Ishmael Kalule-Sabiti. 2014. “Gender and Changing Patterns of Political Participation in Sub-Saharan Africa: Evidence from Five Waves of the Afrobarometer Surveys.” Gender & Behaviour 12(3): 5897-5910. Baldwin, Kate. 2016. The Paradox of Traditional Chiefs in Africa. Cambridge: Cambridge University Press. Baldwin, Kate and Katharina Holzinger. 2019. “Traditional Political Institutions and Democracy: Reassessing Their Compatibility and Accountability.” Comparative Political Studies 52(12): 1747-1774. Baldwin, Kate and Pia Raffler. 2019. “Traditional Leaders, Service Delivery, and Electoral Accountability.” In Decentralized Governance and Accountability, Edited by Jonathan A. Rodden and Erik Wibbels, 61-90. Cambridge: Cambridge University Press. Bansak, Kirk, Jens Hainmueller, Daniel J. Hopkins, and Teppei Yamamoto. “The Number of Choice Tasks and Survey Satisficing in Conjoint Experiments.” Political Analysis 26(1): 112-119. Bansak, Kirk, Jens Hainmueller, Daniel J. Hopkins, and Teppei Yamamoto. 2021. “Conjoint Survey Experiments.” In Advances in Experimental Political Science, Edited by James Druckman and Charles Crabtree, 42-55. Cambridge: Cambridge University Press. Bardasi, Elena and Quentin Wodon. 2010. “Working Long Hours and Having No Choice: Time Poverty in Guinea.” Feminist Economics 16(3): 45-78. Barnes, Tiffany D. and Stephanie M. Burchard. 2013. “‘Engendering’ Politics: The Impact of Descriptive Representation on Women’s Political Engagement in Sub-Saharan Africa.” Comparative Political Studies 46(7): 767-790. Beall, Jo, Sibongiseni Mkhize, and Shahid Vawda. 2005. “Emergent Democracy and ‘Resurgent’ Tradition: Institutions, Chieftaincy, and Transition in KwaZulu-Natal.” Journal of Southern African Studies 31(4): 755-771. Beck, Linda. 2008. Brokering Democracy in Africa: The Rise of Clientelist Democracy in Senegal. New York: Palgrave MacMillan. Bonoff, Nicole. 2016. Traditional Authority in the State: Chiefs, Elections and Taxation in Ghana. PhD Diss. University of California, San Diego. 130 Boone, Catherine. 2003. Political Topographies of the African State: Territorial Authority and Institutional Choice. Cambridge: Cambridge University Press. Boone, Catherine. 2014. Property and Political Order in Africa. New York: Cambridge University Press. Bratton, Michael. 2012. “Citizen Perceptions of Local Government Responsiveness in Sub- Saharan Africa.” World Development 40(3): 516-527. Buur, Lars and Helene Maria Kyed. 2007. State Recognition and Democratization in Sub- Saharan Africa. New York: Palgrave MacMillan. Camara, Bakary. 2013. “The Dynamics of Land Tenure Systems in the Niger Basin, Mali.” Africa 83(1): 78-99. Chiweza, Asiyati L. 2007. “The Ambivalent Role of Chiefs: Rural Decentralization Initiatives in Malawi.” In State Recognition and Democratization in Sub-Saharan Africa, Edited by Lars Buur and Helene Maria Kyed, 53-78. New York: Palgrave MacMillan. Chiweza, Asiyati L. and Yapikachi Msiska. 2020. “Practices and Perceptions of the Politics- Administration Dichotomy in Malawian Local Government.” Journal of Public Administration and Development Alternatives 5(3): 29-48. Clayton, Amanda. 2015. “Women’s Political Engagement Under Quota-Mandated Female Representation: Evidence from a Rnadomized Policy Experiment.” Comparative Political Studies 48(3): 333-369. Clayton, Amanda, Jennifer Noveck, and Margaret Levi. 2015. “When Elites Meet: Decentralization, Power-Sharing, and Public Goods Provision in Post-Conflict Sierra Leone.” Policy Research Working Paper; no. WPS 7335. Washington, DC: World Bank Group. Coffe, Hilde and Catherine Bolzendahl. 2011. “Gender Gaps in Political Participation Across Sub-Saharan African Nations.” Social Indicators Research 102: 245-264. De Kadt, Daniel, and Heratio A. Larreguy. 2018. “Agents of the Regime? Traditional Leaders and Electoral Politics in South Africa.” Journal of Politics 80(2): 382-399. Debra, Emmanuel. 2022. “Participation of Chiefs in Decentralised Local Governance in Ghana.” Commonwealth Journal of Local Governance 26: 53-73. Dickovick, J. Tyler, and James S. Wunsch. 2013. Decentralization in Africa: The Paradox of State Strength. Boulder, CO: Lynne Rienner Publishers. Dzivenu, Setri. 2008. “The Politics of Inclusion and Exclusion of Traditional Authorities in Africa: Chiefs and Justice Administration in Botswana and Ghana.” Political Perspectives 2(1): 1-30. 131 Erk, Jan. 2018. Federalism and Decentralization in Sub-Saharan Africa. London: Routledge. Fearon, James, Macartan Humphreys and Jeremy Weinstein. 2009. “Can Development Aid Contribute to Social Cohesion after Civil War? Evidence from a Field Experiment in Post-Conflict Liberia.” American Economic Review 99(2): 287-291. Fonjong, L., Lawrence Fombe and Irene Sama-Lang. 2013. “The Paradox of Gender Discrimination in Land Ownership and Women’s Contribution to Poverty Reduction in Anglophone Cameroon.” GeoJournal 78: 575-589. Friedrich Ebert Stiftung. 2010. A Guide to District Assemblies in Ghana. Geschiere, Peter. 1993. “Chiefs and Colonial Rule in Cameroon: Inventing Chieftaincy, French and British Style.” Africa 63(2): 151-75. Geschiere, Peter. 2009. The Perils of Belonging: Autochthony, Citizenship, and Exclusion in Africa and Europe. Chicago: University of Chicago Press. Gottlieb, Jessica. 2016. “Why Might Information Exacerbate the Gender Gap in Civic Participation? Evidence from Mali.” World Development 86: 95-110. Henn, Soeren J. 2023. “Complements or Substitutes? How Institutional Arrangements Bind Traditional Authorities and the State in Africa.” American Political Science Review 117(3): 871-890. Hern, Erin Accompo. 2022. “Why Do Women Win in Legally Plural Systems? Evidence from Ghana and Senegal.” Journal of Modern African Studies 60(4): 527-546. Holzinger, Katharina, Florian G. Kern, and Daniela Kromrey. 2016. “The Dualism of Contemporary Traditional Governance and the State: Institutional Setups and Political Consequences.” Political Research Quarterly 69(3): 469-481. Honig, Lauren. 2017. “Selecting the State or Choosing the Chief? The Political Determinants of Smallholder Land Titling.” World Development 100: 94-107. Hussein, Mustafa K. and Gift Sambo. 2022. “Democracy and Electoral Politics Amidst Traditional Leadership Resurgency in Africa: Village-Level Paradoxes and Contradictions in the Context of Malawi.” African Journal of Democracy and Election Research 2(1): 21-41. Hyde, Elizabeth, Margaret E. Greene, and Gary L. Darmstadt. 2020. “Time Poverty: Obstacle to Women’s Human Rights, Health, and Sustainable Development.” Journal of Global Health 10(2): 1-5. Isaksson, Ann-Sofie. 2014. “Political Participation in Africa: The Role of Individual Resources.” Electoral Studies 34: 244-260. 132 Isaksson, Ann-Sofie, Andreas Kotsadam, and Mans Nerman. 2014. “The Gender Gap in African Political Participation: Testing Theories of Individual and Contextual Determinants.” The Journal of Development Studies 50(2): 302-318. Joireman, S. F. 2008. “The Mystery of Capital Formation in Sub-Saharan Africa: Women, Property Rights and Customary Law.” World Development 36(7): 1233-1246. Kaag, Mayke, Yaram Gaye, and Marieke Kruis. 2011. “Land Conflicts in Senegal Revisited: Continuities and Emerging Dynamics.” Land, Law and Politics in Africa: 141-161. Kayira, Gift Wasambo and Paul Chiudza Banda. 2018. “Chiefs and the Politics of Local Government in Malawi: A Historical Perspective.” Journal of Public Administration and Development Alternatives 3(1.1): 84-97. Kayuni, Happy, Michael Chasukwa, Boniface Dulani, and Gift Sambo. 2019. “Perceptions on the Legitimacy of Traditional Leaders in Democratic Malawi.” Journal of Public Administration and Development Alternatives 4(1): 42-54. Keese, Alexander. 2011. “Understanding Colonial Chieftaincy from its Final Phase.” Africana Studia 15: 11-28. Kern, Florian G., Katharina Holzinger, and Daniela Kromrey. 2024. “Between Cooperation and Conflict: Tracing the Variance in Relations of Traditional Governance Institutions and the State in Sub-Saharan Africa.” Third World Quarterly 45(1): 113-132. Kim, SangEun. 2022. “Competition over Traditional Leadership and Public Welfare: Evidence from Central Malawi.” PhD diss. Vanderbilt University. Knierzinger, Johannes. 2011. “Chieftaincy and Development in Ghana: From Political Intermediaries to Neotraditional Development Brokers.” Institut fur Ethnologie und Afrikastudien Working Paper No. 124. Mainz: Johannes Gutenberg Univeritat. Koter, Dominika. 2016. Beyond Ethnic Politics in Africa. Cambridge: Cambridge University Press. Lentz, Carola. 2013. Land, Mobility, and Belonging in West Africa. Bloomington: Indiana University Press. Leon, Gianmarco and Leonard Wantchekon. 2019. “Clientelism in Decentralized States.” In Decentralized Governance and Accountability, Edited by Jonathan A. Rodden and Erik Wibbels, 229-247. Cambridge: Cambridge University Press. Liu, Mingnan. 2017. “Data Collection Mode Differences between National Face-to-Face and Web Surveys on Gender Inequality and Discrimination Questions.” Women’s Studies International Forum 60: 11-16. Logan, Carolyn and Michael Bratton. 2006. “The Political Gender Gap in Africa: Similar Attitudes, Different Behaviors.” Afrobarometer Working Paper No. 58: 1-29. 133 Logan, Carolyn. 2009. “Selected Chiefs, Elected Councillors and Hybrid Democrats: Popular Perspectives on the Co-Existence of Democracy and Traditional Authority.” Journal of Modern African Studies 47(1): 101-128. Logan, Carolyn. 2013. “The Roots of Resilience: Exploring Popular Support for African Traditional Authorities.” African Affairs 112(448): 353-376. Lukac, Martin and Alberto Stefanelli. 2020. Conjoint Experiments: Power Analysis Tool. Retrieved from https://mblukac.shinyapps.io/conjoints-power-shiny/ Mamdani, Mahmood. 1996. Citizen and Subject: Contemporary Africa and the Legacy of Late Colonialism. Princeton: Princeton University Press. Ministry of Local Government and Rural Development. Guidebook on the Local Government System in Malawi. October 2013. Myers, William M. and Kevin S. Fridy. 2017. “Formal versus Traditional Institutions: Evidence from Ghana.” Democratization 24(2): 367-382. Ntsebeza, Lungisile. 2005. Democracy Compromised: Chiefs and the Politics of Land in South Africa. Leiden: Brill. Poteete, Amy R. and Jesse Ribot. 2011. “Repertoires of Domination: Decentralization as Process in Botswana and Senegal.” World Development 39(3): 439-449. Power, Joey. 2020. “Chieftaincy in Malawi: Reinvention, Re-emergence or Resilience? A Kasungu Case Study.” Journal of Southern African Studies 46(2): 263-281. Rangan, Haripriya and Mary Gilmartin. 2002. “Gender, Traditional Authority, and the Politics of Rural Reform in South Africa.” Development and Change 33(4): 633-658. Ribot, Jesse. 2003. “Democratic Decentralization of Natural Resources: Institutional Choice and Discretionary Transfers in Sub-Saharan Africa.” Public Administration and Development 23(1): 53-65. Robinson, Amanda Lea and Jessica Gottlieb. 2019. “How to Close the Gender Gap in Political Participation: Lesson from Matrilineal Societies in Africa.” British Journal of Political Science 51(1): 68-92. Rodden, Jonathan and Erik Wibbels. 2019. Decentralized Governance and Accountability. Cambridge: Cambridge University Press. Scott, James C. 1999. Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed. New Haven: Yale University Press. Teele, Dawn Langan, Joshua Kalla and Frances Rosenbluth. 2018. “The Ties that Double Bind: Social Roles and Women’s Underrepresentation in Politics.” American Political Science Review 112(3): 525-541. 134 Tripp, Aili Mari. 2004. “Women’s Movements, Customary Law, and Land Rights in Africa: The Case of Uganda.” African Studies Quarterly 7(4): 1-19. Ubink, Janine. 2008. Traditional Authorities in Africa: Resurgence in an Era of Democratization. Leiden: Leiden University Press. van Rouveroy van Nieuwaal, E. Adriaan B. 1987. “Chiefs and African States.” Journal of Legal Pluralism 37-38: 39-78. Vengroff, Richard and Alan Johnston. 1989. Decentralization and the Implementation of Rural Development in Senegal. Lewiston, New York: Edwin Mellen Press. Villalon, Leonardo. 1995. Islamic Society and State Power in Senegal: Disciples and Citizens in Fatik. Cambridge: Cambridge University Press. Weber, Max. 1978. Economy and Society: An Outline of Interpretive Sociology. Translated by Ephraim Fischoff. Berkeley: University of California Press. Wibbels, Erik. 2019. “The Social Underpinnings of Decentralized Governance.” In Decentralized Governance and Accountability, Edited by Jonathan A. Rodden and Erik Wibbels, 14-39. Cambridge: Cambridge University Press. Wilfahrt, Martha. 2018a. “Precolonial Legacies and Institutional Congruence in Public Goods Delivery: Evidence from Decentralized West Africa.” World Politics 70(2): 239-274. Wilfahrt, Martha. 2018b. “The Politics of Local Government Performance: Elite Cohesion and Cross-Village Constraints in Decentralized Senegal.” World Development 103: 149-161. Wilfahrt, Martha. 2022. Precolonial Legacies in Postcolonial Politics. Cambridge: Cambridge University Press. Wilfahrt, Martha and Natalie Wenzell Letsa. 2023. “Judge, Landlord, Broker, Watchman: Assessing Variation in Chiefly Duties and Authority in the Ghana-Togo Borderlands.” Modern African Studies 61(3): 439-462. Winters, Matthew S. and Jeffrey Conroy-Krutz. 2021. “Preferences for Traditional and Formal Sector Justice Institutions to Address Land Disputes in Rural Mali.” World Development 142: 1-11. Wucherpfennig, Julian, Nils B. Weidmann, Luc Girardin, Lars-Erik Cederman, and Andreas Wimmer. 2011. “Politically Relevant Ethnic Groups across Space and Time: Introducing the GeoEPR Dataset.” Conflict Management and Peace Science 28(5): 423-437. 135