Abstract
In the developing world, politicians often use public office to redistribute resources to their core constituencies. This form of clientelistic exchange motivates ethnic voting in Africa and may also shape broader attitudes toward the state. But does clientelism retain its power as cross-ethnic contact increases, or might new forms of political linkage emerge? This article uses public opinion data from urban Nigeria to investigate how social position affects trust in elected local officials. The article finds that local ethnic minorities are less trusting of local officials, but this trust deficit does not diminish as cross-ethnic contact rises. For members of locally dominant ethnic groups, however, greater cross-ethnic contact and lessened ethnic attachment dampen expressed trust in local elected officials. The article argues that ethnic clientelism is resilient in urban contexts but that scholarship must take a more nuanced approach to assessing membership in clientelistic coalitions.
Keywords
Introduction
In the developing world, group-based inequalities are often created and institutionalized through political strategies that give some groups greater access to public goods and to elected office. To understand citizens’ political attitudes, political scientists have tended to prioritize ethnicity or other ascriptive differences in explaining access to benefits and to power. There is good reason for this focus. Across the developing world, public goods distribution often benefits groups that hold the presidency or are part of national ruling coalitions (Burgess et al., 2015; Franck & Rainer, 2012). Categorical distinctions—such as ethnicity—often provide the basis for political mobilization and voting (Chandra, 2007; Ferree, 2011; Horowitz, 1985). Elections reinforce relationships between citizens and government officials that are based on clientelistic exchange, especially where voters are poor (Stokes et al., 2013; Van de Walle, 2003; Wantchekon, 2003; Weitz-Shapiro, 2012). In Africa, election campaigns may even heighten the salience of ethnic identities, as suggested by higher rates of ethnic identification during election periods across a number of countries (Eifert et al., 2010).
When it comes to explaining how individual citizens navigate ethnic clientelism, the political science literature has largely focused on ethnic voting. The mechanism underlying voter support for coethnics is one of expected benefit and, to a lesser extent, limited information. Thus, voters support coethnic candidates because they expect these individuals are more likely to favor their own communities in the distribution of benefits and access to government (Bates, 1983; Chandra, 2007; Posner, 2005). Other information, notably cues about positive or negative performance when in office, may mediate—but rarely removes—this effect (Carlson, 2015; Conroy-Krutz, 2013). Despite a large amount of scholarly attention to these dynamics, existing literature has not adequately examined two important questions. First, to what extent do state-citizen linkages built on clientelistic exchange affect individuals’ broader attitudes toward the state and toward representatives? Is clientelism trust-boosting, and, if so, for whom? Second, under what conditions do ethnicity and clientelism begin to weaken their hold over individuals’ actions and beliefs? Does increasing urbanization result in more “cosmopolitan” orientations toward community membership, thereby undermining the power of ethnic clientelism?
These questions relate to a larger debate about how identities and attitudes toward the state shift over time. Some have argued that economic development and state-making lead to the loosening of “primordial” ties and the reorientation of citizens toward more “civic” forms of engagement (Almond & Verba, 1963; Robinson, 2014). Putnam suggests that hierarchical clientelism can crowd out trust and impede engagement with the state, harming institutional performance in the process (Putnam, 1994). But the pathway by which a society moves from one kind of political culture to another remains unclear. Urban settings in the developing world provide an opportunity to look at the effect of individual changes in identity and integration on attitudes toward political representatives. Some argue that greater contact between diverse groups, as often occurs in urban settings, can build bridging social capital and more encompassing civic trust (Kasara, 2013; Paluck et al., 2019; Varshney, 2003). A contrasting school of thought argues that informal institutions like ethnicity and clientelism can adapt and prove durable despite significant changes in social context (Bates, 1983; Cohen, 1969 [2013]; Habyarimana et al., 2009; Mitchell, 1956; Nathan, 2016).
This article examines how ethnic identity and social position affect trust in elected officials in urban Nigeria. I argue that individual access to clientelistic networks shapes trust in local elected officials, but access and exclusion are not determined by group membership alone. Local minority status, level of cross-ethnic contact, and reliance or attachment to an ethnic community can all be seen as collectively composing an individual’s “social position” in her place of residence. To an overwhelming extent, Africanist political science has concentrated on ethnicity as the main determinant of social position. This is despite findings that clientelistic benefits are unevenly distributed within ethnic groups (Keefer, 2007; Wantchekon, 2003) and that the salience of ethnicity varies markedly across contexts (Eifert et al., 2010; Miguel, 2004; Posner, 2004). The challenge posed by urbanization has not been neglected. Newer research has found that urban residence and residence in more diverse communities affects ethnic voting, the importance that individuals attach to ethnic identity, and individuals’ trust in members of other ethnic groups (Grossman & Honig, 2017; Ichino & Nathan, 2013; Kasara, 2013; Nathan, 2016; Robinson, 2020).
This article builds upon this emerging literature but introduces new markers of social position, which take into account whether individuals have cross-ethnic relationships and whether they rely on ethnicity when making everyday decisions. Individuals who report higher degrees of integration or less attachment to their own ethnicity could potentially be seen as more “cosmopolitan” or as less attached to the trust communities that underlie clientelism. These changes may be adaptive, as cross-ethnic cooperation can give individuals access to new networks, resources, and economic opportunities. But lessening reliance on ethnicity can also expose individuals to costs, as they may be excluded from group-provided club goods or other community resources.
How does social position affect trust in local officials in a context of ethnic clientelism? First, I expect, uncontroversially, that local ethnic minorities will be less trusting of their local representatives. Second, the effect of integration and lessened attachment to ethnic trust communities will vary based on the community from which the individual defects. For ethnic minorities, we might expect cosmopolitanism and cross-ethnic friendship to reduce exclusion and thereby boost trust in local officials. Integration presumably creates links to the power-holding group and may yield more access to the clientelistic benefits afforded to members of the dominant group. For members of the dominant group, however, greater cross-ethnic contact and less ethnic attachment may mark them as marginal members of the group. These individuals may be disadvantaged during clientelistic redistribution and will express, therefore, less trust in their elected representatives.
This article uses public opinion data from Nigeria to investigate how social position affects urban residents’ attitudes toward elected officials, especially at the local level. The article presents two main findings. First, I find that local ethnic minorities express much lower levels of trust in local officials than members of dominant groups, and this trust deficit does not diminish for individuals who invest in more cross-ethnic contact or otherwise distance themselves from ethnic minority networks. This finding runs contrary to expectations. But it suggests that political clientelism in urban areas is ethnically based and fairly resistant to inclusion, even of those minority group members who signal a willingness to affiliate with the dominant group. It is possible that local ethnic minorities act strategically by “crossing the line” to vote for parties associated with the dominant group (Ichino & Nathan, 2013), but the findings here suggest that these actions do not result in greater trust in the representatives they elect.
Second, the article provides further support that ethnic coalitions are not homogeneous. For members of dominant groups, higher levels of cross-ethnic contact and lessened ethnic attachment both dampen expressed trust in local officials. This finding suggests that these more cosmopolitan or less “ethnic” members of dominant groups are less likely to benefit from their position in winning coalitions. In terms of how to think about this, party machines, ethnic brokers, and representatives are likely to funnel benefits to their core constituents, since these constituents’ votes are more “reliable” (Nichter, 2008). By signaling independence from ethnic trust communities, cosmopolitans create uncertainty about their political loyalties and their willingness to engage in clientelistic exchange, therefore making brokers less likely to provide them with private or club goods.
The article makes three significant contributions to the growing literature on how clientelism and ethnicity operate, especially in urban settings. First, I show that the existing literature on clientelism might be improved with a more nuanced approach to individual membership in clientelistic coalitions, one that encompasses categorical membership, cross-boundary interaction, and an individual’s attachment to or investment in ethnic identity. This approach might yield more accurate predictions about whom benefits from ethnic clientelism and also the conditions under which clientelism becomes weaker or more inclusive as a form of political linkage. Second, the article’s findings run contrary to the optimistic narrative of Varshney, Putnam, and others, who suggest that greater cross-ethnic contact and other processes inherent to urbanization will result in less group-based discrimination or conflict, more “civic” forms of politics, and better governance (Putnam, 1994; Varshney, 2003). Instead, even in cities, even when it comes to local representatives with whom residents presumably have more contact, trust in government remains ethnically marked. The article’s findings are consistent with a view of clientelism as ethnically exclusionary and as fairly resilient in urban settings. Finally, the article provides a set of measures that can be used to construct a more fine-grained operationalization of an individual’s social position, possibly lessening the discipline’s reliance on ethnicity as the primary driver of political behavior and attitudes in sub-Saharan Africa. 1 Hopefully, it also provides an example of how research conducted in urban settings can contribute to refinement of existing theories and development of new ones (Auerbach et al., 2018).
The article proceeds as follows. The first section situates the article in a larger discussion of ethnicity, clientelism, and urban politics, and it introduces three hypotheses that will guide empirical analysis. This section also situates the article within the broader literature on how clientelism and ethnicity adapt to settings of greater social diversity, including cities. The subsequent section discusses patterns of ethnic inequality in Nigeria, introduces the data, and discusses key differences across the cities and between this article’s urban sample and the Afrobarometer’s nationally representative survey. The next sections of the article discuss measures and results, and the conclusion points out directions for future research and refinement of the analysis presented here.
Trust Networks and Orientations to the State
The study of African politics stresses the tension between fairly weak states and much stronger social institutions, usually built around ethnic identities. Some suggest that these social institutions possess a claim to moral authority that “amoral” state institutions lack, thereby justifying or possibly even mandating theft from the civic realm and redistribution to more legitimate ethnic or “primordial” publics (Bayart, 1993; Ekeh, 1975). These institutions often play a pivotal role as political intermediaries: politicians may rely on local leaders and ethnic brokers to “deliver” blocks of votes (Arriola, 2012; Koter, 2013), and traditional rulers—in rural and urban Africa alike—are often responsible for local governance, the implementation of state policy, or the distribution of public goods (Baldwin, 2016; Boone, 2003; Paller, 2019). Informal institutions also play social and economic roles that have little to do with politics. Ethnic networks are often underlain by norms that support cooperation and reciprocity within the group or by “technologies” that allow members to detect and punish those who violate commitments to other members of the group (Greif, 1989; Habyarimana et al., 2009). These norms and technologies make cooperation within ethnic groups easier than across ethnic lines. This means that ethnic groups often play a direct role in providing club goods, such as security and contract enforcement, and thereby shape societal orientations toward the state (Bodea & LeBas, 2016; Hydén, 1980; MacLean, 2010).
This article suggests that ethnic groups might be profitably viewed as identity-based “trust communities” that arise in the absence of a state to provide collective goods to their members, to facilitate interactions between members, and to allow for opportunity-hoarding behavior by the group (Greif, 1994; Tilly, 2005). For local ethnic minorities and majorities alike, there are clear advantages to investing in trust communities rather than throwing in one’s lot with weak and unreliable state institutions. These communities provide informational resources and can serve as focal points for collective action and “self-help” public goods provision (Habyarimana et al., 2009; Miguel & Gugerty, 2005). Group membership may also mediate individuals’ access to power and patronage. As noted above, the literature on clientelistic voting relies on this assumption: voters support coethnics because they expect these politicians, if elected, will favor their own ethnic communities in the distribution of benefits and access to government (Bates, 1983; Bratton & Kimenyi, 2008; Carlson, 2015). Individuals’ attachments to their trust communities may also be driven by an affective rather than an instrumental logic, and this could serve as a rival mechanism explaining higher levels of trust in elected officials who are coethnic. Individuals may simply prefer their coethnics as elected representatives, either because they view coethnics as more competent and trustworthy or because they are swayed by negative associations with members of other ethnic groups (Ferree, 2011; Horowitz, 1985).
This discussion provides the foundation for the expectation that local ethnic minorities will express lower levels of trust in local elected officials. But does this logic hold in urban contexts? In many parts of Africa, it can be hard to exclude local ethnic minorities from many of the goods that politicians provide, such as schools or roads (Harding, 2015; Ichino & Nathan, 2013). These kinds of public goods are even more likely to be nonexcludable in cities, where higher levels of ethnic diversity and anonymity make policing of access and exclusion of ethnic minorities from state-provided services more difficult. If we expect an instrumental logic to drive individuals’ assessments of politicians, then might the effect of ethnic minority status be attenuated in urban settings? Recent work by Auerbach and Thachil find that slum dwellers in India do not prioritize coethnicity when evaluating potential slum leaders (Auerbach & Thachil, 2018). In some multiethnic slums, associations and politicians mobilize cross-ethnically, improving governance in the process (Auerbach, 2016; Paller, 2019). Urban ethnic minorities may also cope with their demographic disadvantage by voting for noncoethnics in hopes of improving their public goods access (Ichino & Nathan, 2013). Finally, common economic interests and shared discontent about services provision can forge new cross-ethnic coalitions in urban spaces (Resnick, 2013).
There does also exist, however, powerful evidence that clientelism adapts. In two linked articles, Nathan lays out compelling evidence that clientelistic mobilization continues to dominate in urban Ghana. First, he suggests that politicians cannot credibly commit to promises of public goods delivery, leading to declining turnout among the middle class and wealthy voters who demand public goods and are more resistant to ethnic clientelism (Nathan, 2019). In poor neighborhoods, on the contrary, politicians can rely on vote-buying, a form of private transfer that allows them to effectively target their coethnics. This reinforces ethnic voting among the poor voters who compose the bulk of the electorate in African cities and in much of the urban developing world (Nathan, 2016). Elected officials may also employ more sophistical strategies to maintain discrimination in resource distribution on group lines. For instance, Ejdemeyer, Kramon, and Robinson find that Malawian members of parliament from ethnically segregated constituencies use their constituency development funds to provide nonexcludable public goods to areas inhabited primarily by their coethnics (Ejdemyr et al., 2017). In constituencies where high levels of residential integration render this strategy difficult, MPs instead provide private transfers, such as subsidized fertilizers, rather than public goods from which local minorities cannot be excluded. Thus, local ethnic minority status is associated with less access to resources than would be the case for members of locally dominant groups. Local ethnic minorities may therefore be unlikely to materially benefit from political patronage, even if they live in diverse neighborhoods.
In addition to the material costs of exclusion, local ethnic minorities may be politically excluded in more direct and coercive ways. In urban Ghana, some parties attempt inclusive mobilization and build cross-ethnic constituencies, yet party campaigns in other areas are characterized by exclusionary and even violent political campaigns (Klaus & Paller, 2017). The strongest predictors of this divergence are group identity and electoral dominance. In constituencies where one group claimed that they were indigenous and where that group had a stronger incumbency advantage, party campaigns used fear and intimidation to target members of other groups, Klaus and Paller argue. Other work also suggests that ethnically diverse cities can be vulnerable to ethnic mobilization, the spread of distrust, and even group-based violence (Brass, 1997; LeBas, 2013; Osborn, 2008).
Thus, there are two reasons that local ethnic minorities might express lower levels of trust in elected officials, even in cities. The first stems from expectations about the purposes of voting and political support in settings of ethnic clientelism. Although ethnic diversity does make it more difficult for politicians to target their coethnics with public goods, clientelistic redistribution can occur via private transfers or other strategies that allow preferential targeting of core constituencies or “trusty” coethnics. Even putting aside the material disadvantage of out-group membership, the salience of ethnicity is often reinforced in urban political campaigns. Ethnic minority status is therefore expected to dampen expressed trust in local officials, either because it disadvantages individuals in terms of patronage or because it limits the development of affective ties between these citizens and their elected representatives.
Local ethnic minorities are therefore likely to express low levels of trust in local officials and institutions, though their assessments of politicians and institutions at the national level or even possibly the state level can remain unaffected.
Individuals vary, however, in their orientation to the trust communities afforded by their ethnic group membership. Some individuals have more diverse social networks, or they may choose to invest less in the “trust communities” afforded by ethnic, religious, or other identities. For minorities, we could imagine that this kind of integration-oriented behavior would give them greater access to clientelistic rewards and other benefits that typically accrue to members of politically dominant groups. If these integration-seeking individuals are penalized by coethnics for their cosmopolitan ways, either in terms of lost business opportunities or lost access to group-provided club goods, these penalties are unlikely to impact their orientation toward elected officials or toward the state. Local ethnic minorities rarely wield substantial electoral power at the local level, and a more marginal position within that group would not necessarily entail diminished access to politician-provided goods or services. So, for members of local minority groups, integration has potential rewards but few apparent costs when it comes to politician-provided resources. We therefore expect integration to boost expressed trust in local officials for members of ethnic minorities.
The situation is, however, different for members of locally dominant groups. Patronage is rarely distributed equally within clientelistic networks. Although age and sex are likely to be determinants of one’s status within a network (e.g., Wantchekon, 2003), politicians use other cues to target private transfers and other clientelistic benefits. For instance, where party machines are well-developed, local brokers will assess whether individuals constitute core versus marginal supporters and will steer resources toward core supporters (Nichter, 2008; Stokes, 2005). In settings of ethnic clientelism, we would expect this evaluation of core versus marginal support to be based on individual’s orientation toward the ethnic trust community. As ethnic clientelism operates as a form of opportunity-hoarding, we would expect individuals who are seen as less committed to prioritizing group identity or privileging coethnics in economic transactions to be viewed as possible defectors in terms of vote choice. More ethnically diverse social networks and a lack of preferential treatment for coethnics in economic transactions would be visible to one’s local ethnic community. We would therefore expect these traits to be associated with less access to clientelistic benefits for members of dominant groups.
The different expected effects of cross-ethnic contact and proethnic attitudes for members of local minority and locally dominant ethnic groups suggest the following two hypotheses:
We examine these hypotheses below, focusing especially on attitudes toward local officials.
I should make one final note before proceeding to discussion of the case and the presentation of results. The focus of this article is trust in local officials, but being an ethnic minority in one’s place of residence need not shape trust in all public officials. Individuals who are “losers” in local electoral contests may very well be “winners” at the national level. African parties often rely on multiethnic alliances, as they do in Nigeria. This gives a larger number of groups a share of the spoils, but the greatest beneficiaries of the system tend to be individuals who live in more ethnically homogeneous areas. Individuals who live outside these zones may not experience the direct benefits of membership in the national winning coalition, but their attitudes toward national institutions and elected officials may still be shaped by whether their ethnic group is in power at the national level. I briefly discuss this below, but the gaps between trust in local state institutions and representatives and less proximate ones would be a profitable direction for future research. Many countries in the developing world now match Nigeria’s high levels of group-based inequality and intracountry migration, making it likely that citizens discriminate between levels of government or between government officials they can trust and those they cannot.
Inequality and Political Attitudes in Nigeria
Urban Nigeria is an especially good context in which to examine questions of identity, marginalization, and political trust. Ethnic politics is central to political competition at the local level, and ethnicity is a primary determinant of how Nigerians interact with each other and relate to politicians (Osaghae, 2003; Ukiwo, 2003). When asked to choose between national and ethnic identity, only 17% of Nigerians chose to identify with the nation, versus an average of 41% across the set of 16 African countries surveyed in that round of the Afrobarometer (Robinson, 2014). The consequences of strong ethnic attachments and ethnically polarized national politics have been significant. Antisystem movements in Nigeria, including full-scale insurgencies, have often capitalized on group grievances and feelings of ethnic and ethno-regional exclusion (Adesoji, 2011; Adunbi, 2015). Ethnic inequality and differential access to political power have created a fertile terrain for communal violence (Bratton, 2008; Collier & Vicente, 2014; Smith, 2006). Although Nigerian parties attract cross-regional and cross-ethnic support in national elections, campaigning at the local and state levels relies on ethnic mobilization and, in some places, the encouragement of violence against ethnic minorities and migrants (Kendhammer, 2010).
All of these dynamics are more pronounced in urban spaces in Nigeria. Cities vary in their levels of diversity, but competition for resources, public sector jobs, and public office is shaped by ethnicity. Cities attract migrants from far away, meaning that some ethnic groups and individuals are viewed as nonindigenous or as “strangers.” 2 The designation of “nonindigene,” which may overlap but is not coterminous with ethnic group membership, has different implications than being a member of a local ethnic minority. In some states, nonindigeneity can be the basis for formal discrimination in employment, educational access, and access to land; nonindigenes also face other forms of discrimination in their places of residence, such as informal bars to political participation and social services (Human Rights Watch, 2006). In the urban survey data used in this article, more than 60% of respondents agree or strongly agree with the statement that “only indigenes of this state should be eligible for election into public office in this state.” Indigeneity is also linked to violent group conflict over land, especially in the more diverse cities of Nigeria’s “Middle Belt” (Higazi, 2008).
Nigeria’s institutional structure makes these local patterns of belonging and exclusion even more important in shaping citizens’ experiences with the state. As a federal system, Nigeria has three tiers of government: the national, the state level, and the local. For ordinary citizens, state and local governments are the most visible actors, especially in terms of services delivery. This is especially true in the country’s fast-growing cities. Few urban residents have meaningful interaction with federal authorities; instead, they are most likely to pay taxes and rates to their local councils and, in a small number of reformist states, to state governments. 3
Data used in this article are taken from a survey of 2,750 individuals conducted in 11 Nigerian cities in December 2010. These cities differ in terms of ethnic composition, riot propensity, and level of government services provision. The selected cities range across Nigeria’s major geographic zones: there are three northern cities (Bauchi, Kano, Sokoto), three middle belt cities (Jos, Kaduna, Lafia), three cities in southeast Nigeria (Aba, Enugu, Onitsha), and two in southwest Nigeria (Ibadan, Lagos). In each city, a city-representative sample was selected using a stratified, clustered sampling procedure, and respondent households were selected using a random-walk protocol from GIS-selected starting points. To prevent undersampling of women in parts of Muslim northern Nigeria, gender parity was imposed.
Figure 1 shows the substantial differences across cities in terms of expressed trust in public officials across the three levels of government. For each city, stacked bars represent the percent of respondents who reported that they trusted the specified official “somewhat” or “a lot.” Readers should keep in mind that the survey was conducted in December 2010, roughly 11 months after Goodluck Jonathan had taken over presidential duties due to the medical incapacity of elected President Umaru Yar’Adua, a Muslim from northern Katsina State. Jonathan contested the presidency in 2015, but his party lost the presidency and many of the state governments it had previously controlled. Voting patterns were starkly regional: Jonathan and his party received less than 25% of the vote across the northern states, though they were significantly more competitive in the Middle Belt. The same regional patterns of support for Jonathan are visible in the data. The reported trust levels for the president are associated with ethnicity, as levels of presidential trust are highest in southeastern cities of Aba, Enugu, and Onitsha where Igbos—one of Jonathan’s core ethnic constituencies—are dominant. Reported trust in governors and local councilors tends to be lower, but there is a great deal of cross-city variation.
Percentage of city residents who trust official “a lot” or “somewhat.”
How do urban respondents in this article’s urban survey compare with other Nigerians? Trust in government institutions is low across Nigeria, which likely reflects both the general weakness of state penetration and the corruption and ineffectiveness of state institutions. Table 1 shows the differences on key attitudinal measures between the 2010 urban survey used in this article and a nationally representative survey conducted by the Afrobarometer in May 2008. Both sets of respondents report fairly low levels of institutional trust, but urban residents are not appreciably different from other Nigerians on these and other questions. There are also few significant differences between respondents to the urban survey and those surveyed by the 2008 Afrobarometer when it comes to various measures of political participation and engagement (e.g., voting, attendance at community meetings, partisanship, meeting with a government official, and so on).
Common Indicators of Institutional Support.
On other dimensions, however, urban residents are distinct. Respondents from this article’s urban sample express significantly weaker abstract support for democracy and are less willing to reject the use of violence in politics than nationally representative samples collected by the Afrobarometer. 4 Although there is some evidence that support for democracy and democratic norms have eroded in Nigeria since founding elections in 1999, it seems unlikely that the 2 to 3 years between surveys explains these differences given the stability of other attitudinal measures. Rather than reflecting a trend over time, the observed attitudinal differences seem to reflect the greater political alienation of urban residents. When asked if democracy was preferable, a full 18.6% of the sample agreed with the statement “for someone like me, it doesn’t matter what kind of government we have.” This is roughly double the proportion of the Afrobarometer national sample who agreed with the statement.
When asked about how democracy worked in Nigeria, 18.6% of the urban sample volunteered the response “Nigeria is not a democracy,” versus only 3% of the Afrobarometer national sample. Attitudes toward the legitimacy of the use of violence also differ markedly: roughly 14% of the urban sample agreed very strongly that violence is sometimes necessary in support of a just cause, while only 5% of those surveyed by the Afrobarometer responded similarly. As noted in Table 1, the level of agreement that “violence is never justified” is 20% lower in the urban sample. This greater degree of political alienation is consistent with surveys completed in slums in Nairobi and Lagos, which found that slum dwellers expressed much higher levels of alienation and weaker commitment to democratic norms than their nonurban peers (LeBas, 2013).
Another important factor that differentiates urban respondents is the higher level of ethnic diversity and intergroup mixing in cities. The Afrobarometer does not include measures of ethnic intermarriage or other types of integration behavior, such as the ethnic diversity of individuals’ discussion networks. These kind of data are rarely gathered by survey researchers working in the developing world. But cities in Nigeria, like other fast-growing urban centers in the developing world, are more ethnically diverse and have more various forms of intergroup interaction than rural areas. We would expect that urban residents are more likely to have greater contact with members of other ethnic groups than residents in rural areas. Even so, there is significant variation in cross-ethnic friendship and intermarriage across the selected cities. In ethnically diverse “Middle Belt” towns like Lafia and Jos, more than 70% of respondents report having a member of another ethnic group as one of the people with whom they speak most; in contrast, in more ethnically homogeneous Enugu and Onitsha, fewer than 15% of respondents report similar cross-ethnic interaction.
Measures
The main dependent variable used in the article is an individual’s expressed trust in her elected local councilor, which is measured as their response to “how much would you say you trust [your elected local councilor] or have you heard enough about them to say?” 5 The identical question was asked about four other representatives of the state: the president, the individual’s state governor, the police, and the Nigerian electoral commission. Responses range on a 4-point scale from “don’t trust at all” to “trust a lot” for each of the three questions. Looking at expressed trust in the lowest level of government, the local councilor, is the right choice given the article’s interest in the complexity of local social position. Ethnic and religious identity shapes trust in the president in fairly straightforward fashion, since presidential contests divide the electorate on ethnic, religious, and regional lines. We would not expect other markers of local social position, such as intermarriage or other integrative behavior, to have systematic effects on trust in the president, police, or other national state institutions. Trust in state and local officials, on the contrary, is likely to be more directly impacted by individuals’ social positions within their cities, notably whether they are members of local ethnic minorities and the strength of their association with their own ethnic groups.
Measures of Social Position
In terms of major explanatory variables, the article examines local minority status, degree of integration across ethnic lines, and individual ethnic attachment. I first establish whether an individual is a member of an ethnic minority in her city of residence. Minority status depresses institutional trust as minorities are presumably more likely to be socially excluded, to lack political influence, and to have less access to patronage. Where the sizes of groups are roughly similar or where the population is diverse enough that there is no clear dominant group, this alienation is less likely to occur. None of the cities are characterized by ethnic balance between two groups, so there exist clear ethnic minorities in all sampled cities. 6 In most cities, levels of ethnic homogeneity are high, especially in the Southeast. In this area of the country, Igbos constitutes more than 90% of the city samples. In Lagos and Ibadan, Yoruba are more than 70%; in the northern cities of Sokoto, Kano, and Bauchi, Hausas are similarly dominant. The Middle Belt cities of Jos, Kaduna, and Lafia are significantly more ethnically diverse, though Hausa still constitute the largest group in each city (between 25% and 45% of the population). Ethnic minority status is operationalized as minority ethnic status or, in the case of the Middle Belt towns, as non-Hausa ethnicity. It is reasonable to expect that the majority of local councilors are coethnic with the majority group, and the findings are consistent with this assumption. 7 Even if some local councilors were elected from local minority groups, it is not clear that minority councilors would necessarily receive higher evaluations from their coethnics. Due to the patrimonial character of Nigerian electoral politics and the economic dependence of local councils on state government transfers, minority councilors would run the risk of being seen as part of an overall political system that benefits the locally dominant ethnic group.
Because indigeneity in an important concept in the Nigerian context, I also included it as an alternative measure of group-based exclusion in the initial analysis of trust in officials across levels of government. The survey asked respondents if they are indigenous or nonindigenous to their state of residence “by virtue of place of origin of your family.” The distinction between place of birth and place of family origin would be readily apparent to Nigerians. Both ethnic minorities and members of locally dominant ethnic groups may be nonindigenous to their place of residence. For instance, in Lagos, 76% of the dominant Yoruba group in the city identify as nonindigenous; in Kaduna, the proportion of nonindigenous Hausa is smaller, only about 27%, but is still significant. Overall, 40% of the sample identifies as nonindigenous, versus 29% with local ethnic minority status. 8
The article uses three separate measures of integration: ethnic diversity of an individual’s close discussion networks; marriage to a member of a different ethnic group; and reported use of a nonnative language as the language of primary daily use. In terms of ethnic diversity of discussion networks, the survey asks respondents to provide information about “the people [to whom] you talk the most,” and we then asked individuals to tell us the ethnic identity of each of these individuals. This measure is associated with a high rate of refusals, about 20% of the sample. Of those who did answer, the majority (about 59%) reported no cross-ethnic friendships, and about 18% reported networks that were more than 50% noncoethnic. To operationalize this as network diversity, the number of reported cross-ethnic friendships is divided by the respondent’s reported network size, yielding a measure of network diversity that ranges between 0 (purely co-ethnic) and 1 (entirely cross-ethnic).
Of those respondents who are married, 11% reported that they were married to at least one spouse of a different ethnic group. This translates into 4% of the sample that is married across ethnic lines. Finally, respondents were asked which language “on a normal day” they spend the most time speaking. If they reported mostly speaking a language other than their mother tongue, then the language-swapping variable used here takes on a value of 1. About 32% of the sample reports speaking mostly a nonnative language. Of these language-swapping respondents, about 30% report speaking Hausa, about 47% report speaking mostly English, and another 8% report speaking Pidgin.
To examine an individual’s attachment to their own group—effectively the opposite of integration—my preferred measure captures costly ethnic preference in investment. This question asks individuals to make a choice between two different investment opportunities: they could invest with a coethnic and earn 10% after 6 months, or they could invest with a member of a different ethnic group and earn 15% after the same amount of time. Roughly 49% of the sample preferred or strongly preferred the idea of investing with a member of their own ethnic group at the lower rate of return. I have retained the 4-point scale associated with this question, with 4 indicating the strongest preference for investing with a coethnic and 1 indicating strong preference for investing with a noncoethnic at the higher rate of return. In the analysis below, I include a second measure of investment in ethnicity. The survey asks individuals which qualities they would value or think about when choosing to invest money with another person. Individuals were read a list of eight qualities that are generally thought to shape business skill, such as track record with investments, age, education, or standing in the community, and respondents could choose up to three qualities. I use this to construct a binary variable that takes on the value of 1 for those who said they would value ethnic group and 0 for those who specified only other qualities. Roughly 24% of the sample said that they would value ethnic group in making this decision. I do not believe either of these measures necessarily reflect actual investment decisions, but they serve as solid indicators of individuals’ proethnic bias or attachments to ethnicity, especially when making low-information choices.
Controls
Standard demographic controls are used: the respondent’s age, education, religion, religiosity, socioeconomic status, and gender. To capture an individual’s socioeconomic status, the analysis includes as controls a measure of food deprivation (“in the past year, how often did your household go without enough food to eat?”) and occupational controls (e.g., employment in the informal sector; white-collar employment; and so on). Household income is included in the survey, but it is characterized by a large amount of missing data and is somewhat unreliable, as both poor respondents and those who work in the informal sector have greater difficulty accurately reporting their income. The findings reported here are similar when alternative measures of socioeconomic status, including household income, perceived access to safe water, and asset ownership, are used.
The analysis also include three measures of state performance as controls, since individuals’ past contact with the state will shape their evaluations of local officials. These measures are intended to capture both positive and negative experience with government. The first is a measure of concrete public goods receipt. The survey asks if the current administration had built any of the following in the respondent’s community: a hospital or clinic, a primary or secondary school, pipes for water, boreholes, roads, a police post, or new market stalls. The index used in analysis can take on values ranging from 0, in which none of these goods were provided under the current administration, to 7. Roughly 73% of respondents reported the construction of at least one public good, and 40% reported more than two public goods constructed in their local community. The second measure establishes whether individuals were affected by state-sponsored evictions or slum clearances, which are common in Nigerian cities and in cities throughout the developing world. Respondents were asked if they were affected by state clearances of illegal structures, and they could report either economic losses, such as seizure or destruction of property, or loss of residence. Just over 8% of respondents reported losing their place of residence due to state clearances, and an additional 21% of respondents reported economic losses. I use these responses to construct a binary variable for which any experience of victimization associated with state eviction takes the value of 1 and no experience of victimization has the value of 0. The second measure used for negative state contact is bribe victimization. Roughly 29% of respondents reported having been solicited for a bribe by a state official in the past 12 months. Notably, for both of these measures of victimization or negative contact with the state, there were only very slight or nonappreciable differences in rates of victimization between local ethnic minorities and members of locally dominant groups.
Finally, since both level of interest in public affairs and voting behavior are common political engagement variables that are usually correlated with trust in elected officials and government institutions, they are included as controls here. The political interest question simply asks respondents how interested they are in “public affairs,” and responses range on a 4-point scale from “not at all interested” to “very interested.” If respondents said that they voted in the 2007 elections in Nigeria, the voting behavior variable takes on a value of 1. Finally, I include recent urban migrant status as a control, as these individuals are likely less knowledgeable about and less engaged in local politics. Of the 25% of the sample who reported moving within the past 5 years, only 16% reported moving from outside their city of residence. Overall, these new migrants represent only 4% of the sample.
Effects of Minority Status and Integration on Trust in Local Officials
Due to the ordinal nature of the dependent variables and substantial unobservable heterogeneity across cities, all regressions use ordered logit with city-fixed effects. Before proceeding to the main analysis of the effects of minority status and integration on trust in local officials, it makes sense to situate the article’s findings about local officials with respect to trust in other government officials. Table A2 in the online Appendix presents analysis of how well local ethnic minority status predicts trust in state governors, the president, and local councilors. Overall, the findings are consistent with my intuition: membership in national-level identity groups—that is, overall ethnic and religious group membership—shape trust in the president, but these factors have no significant effect on individuals’ trust in their state governors or their local councilors. Second, local minority status in the sampled cities is not significantly associated with trust in governors. Like presidents, governors must build encompassing coalitions to win their elections. City residents who are local ethnic minorities may belong to ethnic groups that are dominant in other areas of the state, and their ethnic group may be included in the state-level ruling coalition, even if they are excluded at the local level. Consequently, local ethnic minorities may sometimes be hesitant to sanction their governors for poor governance in their places of residence.
These findings suggest two things, which are important for establishing the validity of the measurement of trust in elected officials and our understanding of the significance of the article’s overall findings. First, individuals do differentiate between levels of government, and individuals may be winners at one level of politics but still view themselves as losers at other levels. This should increase our confidence that the article’s measure of local level trust captures real experience of local politics and an evaluation of more proximate government officials that is independent of individuals’ feelings about other power holders. Second, these findings suggest that diversity may not be as dangerous for clientelistic democracy as some might presume (e.g., Horowitz, 1985). Groups that are excluded from national power may be able to gain power and access at lower levels, while individuals who are ethnically excluded at the local level still express trust and support for elected leaders at other levels.
Turning now to the analysis of local officials and the effects of cross-ethnic contact, Table 2 presents the article’s key findings. Across all models, local ethnic minority status is significantly and negatively associated with expressed trust in local councilors at the p < .001 level. Despite its importance in Nigerian politics, indigeneity has no significant effect on trust in local officials across multiple specifications. Table 2 presents results for two sets of variables that I hypothesize shape trust in local officials: cross-ethnic contact or integration and proethnic attitudes. Model 1 shows little support for the idea that cross-ethnic contact or integration uniformly boosts trust in local councilors, at least for the full sample. Ethnic minority status remains significant and negative in its effect on expressed trust. Cross-ethnic marriage has no significant effect on trust, but linguistic assimilation—or primary use of a nonnative language—does have a significant and positive effect on trust. On the contrary, those who have diverse social networks are significantly less likely to express trust in local councilors. 9 It was expected that integration yielded heterogeneous effects based on ethnic minority status, and these effects will be examined more fully in a moment. Model 2 presents results for the two measures of reliance on trust communities, which could alternatively be seen as ethnic preference. The results show that both measures of preference for investment with coethnics, which seem the best proxies for proethnic orientation, are strongly and positively associated with expressed trust in local councilors.
Effects of Social Position on Trust in Elected Local Councilors.
Robust standard errors in parentheses. Models include city fixed effects and all control variables. The dependent variable takes higher values for higher trust in elected officials (maximum 4). Variables that are consistently insignificant are not reported. Models with network diversity and cross-ethnic marriage also include controls for refused network responses and unmarried status.
p < .10. **p < .05. ***p < .01.
The final combined model in Table 2 shows strong support for several of the article’s hypotheses, though the table does not present results regarding possible heterogeneous effects of cross-ethnic contact and proethnic orientation. All the variables that reach significance in the other models retain their significance in the combined model. As in Model 1, greater levels of network diversity dampen trust, while linguistic assimilation seems to boost it. These seemingly contradictory findings will be discussed in the next section.
Many of the control variables also have effects in the expected directions, and the significance and size of effects is remarkably consistent across model specifications. Both positive and negative contact with the state are consistently and strongly associated with trust in local officials, suggesting that trust is partly a function of individuals’ assessments of government performance. Female respondents are more likely to express trust in local councilors, while the poor and those working in the informal sector are less likely to express trust. Ethnic and religious affiliation, education, and level of interest in public affairs have no statistically significant effects on trust. Interest in public affairs and past voting are not significantly associated with greater trust in elected officials.
The size of the effects is substantial. Ordered logit does not allow this to be directly inferred from coefficient values in the combined model, so predicted probabilities are calculated to assess the marginal effects. Being a member of a local ethnic minority dampens the likelihood of reporting the highest level of councilor trust by about 31%, while movement from a homogenously coethnic social network to one with half noncoethnics decreases the likelihood of the highest level of trust by roughly 18.5%. Moving from the lowest to the highest levels of expressed proethnic investment preference increases the likelihood of high trust in one’s councilor by 40%, while even moving between intermediate levels (from “agree with B” to “agree with A”) on this variable is associated with a nearly 12% increase in trust in the local official. There are similarly large marginal effects associated with individuals’ experiences of state contact. Experience of eviction lessens the likelihood of expressing “a lot” of trust in one’s local councilor by 23%, while a roughly modest increase from 0 to 2 on the index of public goods receipt increases the same by over 16%. These are substantial marginal effects, suggesting that future research should pay greater attention to local social position, integration or proethnic orientation, and negative experiences with state contact.
Differential Effects on Minority and Majority Groups
Readers will remember, however, that the effects of integration were not expected to be consistent across the sample. It was instead hypothesized that greater cross-ethnic contact or lessened attachment to ethnicity would mediate the effect of group membership on expressed trust in local officials. The initial intuition was that ethnic minority status’s trust-dampening effects operated mainly through exclusion from patronage networks. Thus, one might expect that minorities who have greater contact with the dominant group would experience less exclusion and would be more likely to trust local officials than their less integrated peers. Members of dominant groups who integrate, however, would be expected to be more marginal in patronage networks than their less integrated peers and therefore express lower levels of trust. The same lower levels of trust would be expected for members of dominant groups who express less coethnic preference in investment decisions for the same reasons.
I conducted a series of ordered logit regressions, some of which are reported in Table A3 in the online appendix, to explore the possibility of interactions between local ethnic minority status and measures of integration and proethnic sentiment. Interaction terms are consistently nonsignificant across a large number of model specifications. This is true both for the measures of integration and proethnic preference that reached significance in the models above, as well as those that did not have significant effects on overall trust. In terms of our core hypotheses, we find no evidence that the effect of integration on trust is substantially different for minorities than it is for nonminorities, and the most integrated ethnic minority members do not seem to be appreciably more trusting than their less integrated minority peers. Interestingly, several interaction terms are positive even though they fail to reach statistical significance. This may suggest that the negative effects of network diversity on trust, for instance, are smaller for members of local ethnic minorities than for members of dominant groups. We are not powered to observe this, however.
Despite the insignificance of interaction effects, there remain interesting differences across the two groups. Running ordered logit models for separate samples of local ethnic minorities and members of locally dominant groups allows us to better understand which groups are driving the results reported in Table 2. To graphically depict these differences, Figure 2 shows the predicted probabilities of the two groups for the highest response category of “trusts a lot” (4) on the dependent variable. Both ethnic minorities and members of dominant groups are appreciably less likely to express the highest level of trust in local representatives when they have cross-ethnic friendships. Receipt of public goods increases the probability of trusting local representatives “a lot” for both goods, while employment in the informal sector (not depicted) similarly dampens trust in local officials for members of both groups. The effects of other variables on predicted probabilities are less consistent across the two groups. Notably, the positive effect of nonnative language use applies only to minorities, while the negative effects of eviction and bribe solicitation work through members of dominant ethnic groups.
Differential effects of social position for minorities and nonminorities.
Overall, further investigation of how trust operates within the two groups suggests that the bulk of the trust-dampening effects reported in Table 2 operate via the effects on members of locally dominant groups. One can compare the size of these effects across the full sample and a smaller sample composed only of those 1,917 respondents who belong to members of locally dominant ethnic groups. The size of the effects of integration and ethnic preference are roughly the same in these two sets of regressions. I should also note that our two best measures of integration and cross-ethnic contact are contradictory when it comes to minorities. Local ethnic minorities that have cross-ethnic friendships are less likely to express high levels of trust in local representatives, but minorities who spend much of their time speaking a language other than their mother tongue express higher levels of trust in local representatives. As cross-ethnic friendships have a uniformly significant and negative effect on trust for both groups, it may be that these relationships serve as vehicles for the interpersonal transmission of low institutional trust in a way that more casual cross-ethnic contact does not.
Conclusion
To a large extent, the political science literature has focused on what ethnic clientelism means for voting and for the distribution of public goods. But state-citizen linkages built on clientelistic exchange have implications for citizens’ orientations toward the state, including expressed trust in institutions and in elected officials. This article has established that local ethnic minority status and other aspects of individuals’ local social position affect how individuals view the lowest level of government. In Nigeria and in other countries, this is the level of government with which individuals are most likely to have contact and through which most services are delivered. Local ethnic minorities may engage in cross-ethnic cooperation and may lessen their reliance on ethnic trust communities when living in urban environments. The article suggests that these accommodating behaviors will not make minorities more trusting of the politicians who represent them. I interpret this as evidence that clientelism is not ethnically inclusive in urban settings, at least not in urban Nigeria.
The findings on cross-ethnic integration are even more interesting for members of dominant groups: individuals who integrate and who are more open to cross-ethnic contact express lower levels of trust in local elected officials. It may be that these individuals are, like Nathan’s middle-income voters, impatient with ethnic clientelism and the local politicians who practice it. But the findings are also consistent with a narrative in which integration makes these individuals marginal in clientelistic networks and therefore less likely to receive the material benefits that may shape their nonintegrated peers’ higher levels of trust in local officials. These findings challenge the trust literature that relies on macro-comparisons across countries, which argue that individual trust in government institutions is driven by better-performing or less corrupt state institutions (e.g., Hutchison & Johnson, 2011).
Overall, this article suggests that local dynamics are important drivers of citizens’ orientation to the state. Existing literature has shown that identity shapes individual trust in government institutions and individuals’ expectations regarding government delivery. The bulk of this literature, however, has either fixed its attention on aggregate measures, such as the level of ethnic diversity, or on an individual’s position vis-à-vis national ruling coalitions. This is indicative of a general bias in the study of distributional politics in favor of positioning citizens in larger clientelistic networks, that is, those organized on the national level. This article calls attention to the importance of local winning coalitions and local patronage networks, and it argues for a more nuanced way of understanding who belongs and who is excluded from these networks. Future research will hopefully pursue some of the questions raised by the findings in this article, as well as explore the complicated patterns of identity change and cross-ethnic contact in urban areas in the developing world. Although integration and the emergence of new identities may eventually erode the sway of ethnic clientelism in the developing world, this article suggests that cross-ethnic contact and diminished reliance on ethnic trust communities is unlikely to boost trust in state institutions and representatives in the short run.
Supplemental Material
CPS_2020_Appendix – Supplemental material for Who Trusts? Ethnicity, Integration, and Attitudes Toward Elected Officials in Urban Nigeria
Supplemental material, CPS_2020_Appendix for Who Trusts? Ethnicity, Integration, and Attitudes Toward Elected Officials in Urban Nigeria by Adrienne LeBas in Comparative Political Studies
Research Data
CPS_replication_data for Who Trusts? Ethnicity, Integration, and Attitudes Toward Elected Officials in Urban Nigeria
CPS_replication_data for Who Trusts? Ethnicity, Integration, and Attitudes Toward Elected Officials in Urban Nigeria by Adrienne LeBas in Comparative Political Studies
Supplemental Material
LeBas_2020_Replication_code – Supplemental material for Who Trusts? Ethnicity, Integration, and Attitudes Toward Elected Officials in Urban Nigeria
Supplemental material, LeBas_2020_Replication_code for Who Trusts? Ethnicity, Integration, and Attitudes Toward Elected Officials in Urban Nigeria by Adrienne LeBas in Comparative Political Studies
Footnotes
Acknowledgements
The author thanks Cristina Bodea, Jeffrey Conroy-Krutz, Rachel Gisselquist, Alisha Holland, Janica Magat, Matthew Wright, and participants at a UNU/WIDER workshop for comments on earlier drafts.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was funded by the Department for International Development (DFID) of the United Kingdom.
Notes
Author Biography
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
