Strangers and Foreigners: Trust and Attitudes toward Citizenship in Sub-Saharan Africa

Abstract

This study explores the factors that shape natives’ attitudes toward citizenship acquisition for foreigners. The hypothesis is that, in Sub-Saharan Africa, the slave trade represents a deep determinant of contemporary attitudes toward citizenship, through a proximate determinant which is the level of trust. Accordingly, individuals belonging to ethnic groups with higher exposure to historical slave exports are more likely to exhibit a sense of distrust toward strangers, and are consequently more likely to oppose citizenship laws that favor the inclusion of foreigners. The findings indicate that individuals with higher levels of trust toward other people do exhibit more favorable attitudes regarding the acquisition of citizenship at birth for children of foreigners, that these attitudes are also negatively related to the intensity of the slave trade, and that the underlying inverse relationship between trust and the slave trade is confirmed. Other factors such as conflict, kinship tightness, and witchcraft beliefs, which could also influence attitudes toward citizenship through the channel of trust, do not yield the same distinct pattern of associations as observed with the slave trade.

1. Introduction

You are no longer foreigners and strangers, but fellow citizens.

Paul, Ephesians 2:19, The Bible (ESV)

The perception of foreigners as strangers, or unfamiliar people deemed unworthy of trust, has endured at least since biblical times. Paul the Apostle, in his letter to the Ephesians, highlights the distinction between the condition of foreigners and strangers, on the one hand, and that of “fellow citizens” on the other. This paper explores the relationship between natives’ attitudes toward citizenship acquisition for foreigners and the level of trust in strangers. Recognizing trust as a proximate determinant of these attitudes calls for an investigation of their underlying fundamental determinants.

The recent surge in migration and refugee flows has brought global attention to the forefront of citizenship policy discussions. However, this issue has consistently been particularly charged in Sub-Saharan Africa, where artificial post-colonial state borders, coupled with the historical impact of war and famine, have been associated with huge population movements. Internal migration across African countries, encompassing both forced migrants and refugees, continues to account for a substantial portion of contemporary flows (Herbst 1997, 2000).

The hypothesis driving the present study posits that, in Sub-Saharan Africa, the legacy of the slave trade—one of the most significant forced displacement experiences in history—may constitute a fundamental factor influencing contemporary sentiments toward foreigners and their prospects for acquiring citizenship. Accordingly, the channel linking the slave trade to attitudes toward citizenship is the influence of the slave trade on trust in other people. More intense exposure to slave exports may indeed have triggered a distrust of strangers, manifesting in current opposition to citizenship laws that favor the inclusion of foreigners. Thus, the hypothesis implies that, while trust is a proximate determinant of attitudes toward citizenship acquisition, in Sub-Saharan Africa the slave trade represents an underlying deep determinant.

While Nunn and Wantchekon (2011) have already established the detrimental impact of the historical slave trade on trust in Sub-Saharan Africa, this paper makes a dual contribution. First, it provides evidence regarding the relationship between contemporary measures of trust and citizenship attitudes and, second, it also uncovers an effect of the slave trade on citizenship attitudes, which is channeled through trust. The main measure of sentiments toward foreigners is represented by attitudes toward inclusive citizenship laws exemplified by a jus soli regime—where citizenship is automatically granted to the offspring of foreigners—in contrast to a jus sanguinis regime, which attributes the citizenship of the parents to offspring.

The empirical investigation employs individual-level survey data sourced from Afrobarometer, a research network dedicated to gauging public attitudes on economic, political, and social issues in African countries. The surveys also address respondents’ attitudes regarding citizenship acquisition for foreigners. A specific question probes the respondent’s opinion on the right to be citizens for individuals born in a country from non-citizen parents. The responses to this question can serve as a proxy for attitudes toward birthright citizenship, i.e., citizenship acquisition at birth according to the jus soli legal tradition. While the ability to acquire citizenship at birth is by far the most salient issue as far as citizenship policy is concerned (Manby 2018, 2020), the study also explores additional related survey questions. These include inquiries about the right to be a citizen for a non-citizen who has lived and worked in the country, which proxies for attitudes toward naturalization, and the right of an individual to hold dual citizenship.

Afrobarometer is also the source for various measures of trust. Although attitudes toward granting citizenship to foreigners are likely to align more naturally with trust in other people, this study takes into account all available measures of trust, by increasing closedness of the relationship between respondents and a reference group, namely, trust in other people, neighbors, and relatives. Data on the slave trade at the ethnicity level are from Nunn and Wantchekon (2011), who are also the source for a number of geographic and historical potential confounders. Ethnographic variables are drawn from the Ethnographic Atlas by Murdock (1967).

The first result of the analysis is that individuals with higher levels of trust in other people do exhibit more positive attitudes toward the acquisition of citizenship at birth through jus soli. This confirms the role of trust as a proximate determinant of citizenship attitudes. The novel association between trust and citizenship attitudes that this study uncovers remains robust across specifications including country fixed effects and an increasing number of controls (initially geo-historical, and then also ethnographic ones). This key association—which motivates the subsequent analysis—is also robust to the inclusion of ethnicity fixed effects that should filter out any other ethnicity-specific factors.

Second, attitudes toward citizenship are found to be negatively related to the intensity of slave exports for the ethnic group to which individuals belong. This implies that the slave trade does represent one of their underlying deep determinants.

Third, the negative impact of the slave trade on trust in other people, as established by Nunn and Wantchekon (2011), is validated within the present sample. This confirmation strengthens the hypothesis that the impact of the slave trade on citizenship attitudes is indeed channeled through trust.

The above results are corroborated by a battery of extensions and robustness checks. First, the uncovered channels are also activated for other citizenship policy provisions, including naturalization and dual citizenship, as well as an index capturing all three dimensions of citizenship policy, albeit in an attenuated fashion. This affirms that citizenship acquisition at birth is the most salient dimension of citizenship policy. Furthermore, in line with the tested hypothesis, the impact of alternative measures of trust—in neighbors and relatives—on jus soli attitudes is weaker, particularly for the latter. This aligns with the expectation that trust in closer groups should have less relevance to feelings toward foreigners. Conversely, the results hold when replacing trust in others with the trust gradient, defined as the difference between trust in other people and trust in relatives.

Next, to assess whether the evidence so far produced can be generalized, thus enhancing its external validity, the focus shifts to a sample of individuals residing outside their ethnic homeland, defined as movers. As expected, movers constitute a selected sample of younger, more educated, and more urbanized individuals. However, restricting the analysis to this sample allows us to concentrate on the portable component of ancestral influence on current outcomes, filtering out the influence of local factors. Specifically, introducing fixed effects for the ethnicity historically inhabiting the area of current residence, rather than for the ancestral homeland, should eliminate all potential local confounders, i.e., time-invariant characteristics of the geographic, economic, institutional, and cultural environment where movers currently reside. The analysis over the movers sample confirms the influence of the slave trade on citizenship attitudes, channeled through trust.

The final part of the paper is devoted to testing alternative potential factors—beyond the slave trade—that, through trust, may also influence attitudes toward citizenship. The literature has proposed other determinants of trust that could potentially impact such attitudes. This study tests three alternative hypotheses, selected from those deemed especially relevant within the history of African development, and centered respectively on conflict, kin ties, and witchcraft beliefs. However, none of these alternatives demonstrates the ability to generate the distinctive pattern of associations observed for the slave trade.

The rest of the paper is organized as follows. After a section that provides a summary of the related literature, another section presents background information on citizenship policy and its evolution in Sub-Saharan Africa. Next, one section is devoted to the description of the data and the empirical strategy, followed by a section with estimation results. A specific section focuses on the movers sample, and another on discusssing alternative potential determinants of citizenship attitudes. A final section draws conclusions from the analysis. Supplementary online appendices provide background information on the African slave trades as well as additional tables and a figure.

2. Literature

This paper intersects with various strands of literature. In the first, relatively smaller stream looking at citizenship policy and attitudes, Bertocchi and Strozzi (2010) compile a data set on citizenship laws across countries of the world and investigate their evolution in the post-war period, emphasizing the potential role of the perceived threat posed by immigrants. Bertocchi and Strozzi (2008) assess the impact of inclusive citizenship policies based on the jus soli principle on migration decisions during the nineteenth-century mass migration from Europe to America. Imam and Kpodar (2020) demonstrate that more inclusive jus soli laws tend to promote contemporary economic development. Herbst (1997, 2000) and Manby (2018, 2020) discuss the formation of citizenship laws in Africa. A closely related literature investigates broader attitudes toward immigration. Herreros and Criado (2009) find a positive impact of social trust on attitudes toward immigration in the European context. Using the same Afrobarometer data used in this study, Zhou (2018) shows how attitudes toward citizenship acquisition and trust are jointly influenced by the presence of refugees. Others explore the relationship between immigration and natives’ voting behavior (e.g., Voigtländer and Voth 2012; Barone et al. 2016; Halla, Wagner, and Zweimüller 2017; Dustmann, Vasiljeva, and Piil Damm 2019; Tabellini 2020) and the political economy of enfranchisement of ethnic and racial minorities (e.g., Alesina, Glaeser, and Sacerdote 2001; Cascio and Washington 2014; Bertocchi and Dimico 2017; Koukal, Schafer, and Eichenberger 2021). By extending its focus to the citizenship issue, the present investigation also contributes to the interdisciplinary literature on the impact of intergroup contact on attitudes, building on Allport, Clark, and Pettigrew (1954) and including recent contributions such as Bursztyn et al. (2021).

The second interconnected stream of work is the extensive literature on trust, with classic contributions from Fukuyama (1995), Knack and Keefer (1997), Putnam (1993), and Guiso, Sapienza, and Zingales (2008). This paper is particularly close to Nunn and Wantchekon (2011), who establish how trust in Sub-Saharan Africa is influenced by the slave trade. Other studies that have linked trust to conflict (Rohner, Thoenig, and Zilibotti 2013; Besley and Reynal-Querol 2014), kinship (Enke 2019; Moscona, Nunn, and Robinson 2017), and witchcraft beliefs (Gershman 2016) are also highly relevant, since they suggest potential alternative explanations, beyond the slave trade, for the deep determinants of citizenship attitudes as channeled through trust.

This paper also adds to the broader understanding of the long-run consequences of the slave trade on African societies, following among others Nunn (2008), Nunn and Wantchekon (2011), Whatley (2014), Bertocchi and Dimico (2019), and Teso (2019) (comprehensive surveys are available in Bertocchi (2016) and Nunn (2017)). Within the expanding research on the impact of historical legacies across various dimensions of African development, as surveyed by Michalopoulos and Papaioannou (2020), many contributions intersect with the issues addressed here. This is the case, for instance, for the connection between the slave trade and conflict (Fenske and Kala 2015, 2017; Boxell 2019; Boxell, Dalton, and Leung 2019; Cherniwchan and Moreno-Cruz 2019), the slave trade and witchcraft beliefs (Gershman 2020), kin ties and conflict (Moscona, Nunn, and Robinson 2020), and kin ties and institutions (Tedeschi 2021).

Since the present focus is on the slave trade as a fundamental determinant of trust and attitudes toward foreigners, as shaped by the process of historical economic development, this work is also linked to research on persistence of culture and social preferences, including among others Bisin and Verdier (2001), Alesina and Giuliano (2015), and Giavazzi, Petkov, and Schiantarelli (2019). Finally, this paper complements the literature on long-term development which distinguishes between proximate and deep (or fundamental) factors, following Hall and Jones (1999), Acemoglu, Johnson, and Robinson (2002), and Ashraf and Galor (2013).

3. Historical and Institutional Background

Each country in the world has established laws that govern the attribution of citizenship. Citizenship (a term often used interchangeably with nationality) is the legal institution that designates full membership in a nation, entailing a set of associated rights and duties contingent upon on a country’s legislation. Rights commonly include the voting franchise, permission to reside and work in the country, unrestricted travel in and out of the country, legal protection in the face of criminal charges, and the ability to secure a visa for a relative. Duties may encompass compulsory voting, participation in the military draft, and the renunciation of original citizenship in the event of naturalization.

Citizenship can be acquired at birth, by naturalization, or through marriage.¹ The vast majority of individuals obtain citizenship of a country at birth.² The regulation of citizenship at birth, which crucially determines citizenship acquisition by second-generation immigrants, is enshrined in the legal frameworks of common and civil law. Common law traditionally employs the jus soli principle, where citizenship is granted based on birthplace. Under this principle, a child born in the country of immigration to immigrant parents is automatically considered a citizen. Civil law, on the other hand, follows the jus sanguinis principle, attributing citizenship by descent, so that a child inherits citizenship from their parents, independently of where they are born. Despite these foundational principles, many countries underwent a process of adaptation during the second half of the twentieth century. This adaptation was influenced by global events such as decolonization, the collapse of the socialist system, and the increasing pressure of international migration. However, while citizenship policy is a component of broader migration policy, it differs from other migration policy measures such as quotas and visa requirements. The latter measures are typically adjusted in response to the business cycle and the ideology of the current government in power. In contrast, reforms to citizenship laws tend to be the result of long-term processes of adaptation, often involving constitutional amendments.

In Sub-Saharan Africa, the process of post-war decolonization had a profound impact on citizenship policy.³ The vast majority of African colonies that adhered to the jus sanguinis principle of their civil law metropolitan countries continued with this approach after gaining independence. Conversely, many former British and Portuguese colonies opted to reject the jus soli tradition, adopting a frequently ethnically tinged version of jus sanguinis. This shift was often seen as a means to more easily control the formation of national entities. Not only were rules devised to exclude the descendants of Europeans, but some countries sought to exclude from citizenship even those lacking an ancestral link to the land. Similar distrust was extended to potential dual citizens, as dual citizenship was long prohibited in most countries. Certain countries enacted legislation aimed at reinforcing racial or ethnic elements within their citizenship policies. For instance, Sierra Leone’s 1961 Constitution established that citizenship is transmitted solely by descent, specifically to children whose father and a grandfather were Sierra Leoneans of African descent. Similarly, the 1964 Congolese Constitution responded to massive immigration from Rwanda by conferring citizenship only to individuals whose parents belonged to tribes established within the territory by 1908, the initial year of Belgian colonization. In 1981, President Mobutu signed a new nationality law mandating an ancestral connection to the population residing in the territory dating as far back as 1885.

To this day, in several Sub-Saharan African countries, ethnic conflict lies at the root of adaptation of citizenship legislation in favor of one ethnic group over others. The emphasis on ethnic identity becomes even more challenging in the face of the artificial state borders established by European colonial powers during the 1884–5 Berlin conference that marked the end of the Scramble for Africa, which were not renegotiated after independence. A consequence of these citizenship policies is the marginalization and de facto statelessness of significant strata of the population.

A distinctive aspect of Africa that heightens the significance of citizenship policy is its history of intense migration, beginning with substantial flows of immigrants from Europe during the era of empires. Large-scale population movements also occurred as a result of the political and economic transformations brought about by colonization. Today, internal migration across African countries continues to account for a significant portion of overall migration, encompassing forced migrants and refugees. Migration can pose challenges to a country’s stability by questioning its territorial sovereignty, threatening cultural identity, and imposing economic burdens. For all these reasons, it heavily influences citizenship policy and shapes the attitudes of citizens toward it. As a result, in recent decades, the prevailing trend in the African continent has been to curtail automatic birth rights to citizenship and to make naturalization and dual citizenship more stringent. This shift has resulted in citizenship laws becoming less inclusive for foreigners.

Within the present sample of 27 Sub-Saharan African countries, based on the classification by Bertocchi and Strozzi (2010), 18 countries apply citizenship laws at birth based on the jus sanguinis principle, 4 apply jus soli, and 5 have a mixed regime.⁴ In the following empirical investigation, cross-country differences in citizenship laws are absorbed by country fixed effects, together with a plethora of other country-specific characteristics. On the other hand, the actual legal implementation of citizenship policy is complemented and reinforced by natives’ sentiments toward foreigners, which can be even more deeply rooted in culture and history.⁵ Thus, the round 5 Afrobarometer data on individual attitudes toward citizenship acquisition offer a unique opportunity to explore their correlates and underlying determinants.

4. Data and Empirical Strategy

4.1. Afrobarometer Data

Data on attitudes toward citizenship acquisition and the level of trust are available at the individual level from the Afrobarometer surveys.⁶ Implemented by national partners, Afrobarometer gauges the social, political, and economic atmosphere in African countries through national sample surveys, capturing citizens’ attitudes toward democracy, markets, civil society, and other developmental aspects. The questionnaire is standardized to facilitate comparison between the covered countries. These nationally representative surveys are based on interviews conducted in the local languages on a random sample of 1,200–2,400 adult citizens in each country. The sampling universe normally includes all citizens aged 18 and older. The data are geocoded at the village and town levels.⁷

The fifth round surveys, completed in 2013, cover the following 29 Sub-Saharan African countries: Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cape Verde, Ghana, Guinea, Ivory Coast, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritius, Mozambique, Namibia, Niger, Nigeria, Senegal, Sierra Leone, South Africa, Swaziland, Togo, Tanzania, Uganda, Zambia, and Zimbabwe. Cape Verde and Swaziland, lacking ethnicity-level information to be matched with Afrobarometer, are excluded from the sample, resulting in 27 countries.

In round 5 only, Afrobarometer specifically surveys respondents’ attitudes toward citizenship granting by asking them whether people in any specific situation, in their opinion, have the right to be citizens. We focus on the question regarding the right to citizenship for a person born in a country with two non-citizen parents, which can be likened to the acceptance of birthright citizenship under a jus soli regime. Respondents can answer yes, no, or don’t know. We code a binary variable that takes value 1 when the answer is yes, and 0 when the answer is no.⁸  Figure S2.1 displays the geographical distribution of jus soli attitudes for the countries in the sample (panel A). As explained by the legend, the shades reflect five intervals for the country-average values of the variable, with darker shades indicating a more favorable attitude. Attitudes toward related citizenship policy provisions can be proxied by respondents’ answers to questions regarding the right of a non-citizen person who has lived and worked in the country to become a citizen—corresponding to the right to naturalization—and the right of a person to hold dual citizenship.

The Afrobarometer surveys also include inquiries about measures of interpersonal trust, captured in the fifth round by three variables: trust in other people you know, trust in neighbors, and trust in relatives.⁹ A categorical variable is coded for each trust measure, with values ranging from 0 to 3, corresponding to the responses “not at all,” “just a little,” “somewhat,” or “a lot.”¹⁰  Figure S2.1 (panels B–D) illustrates the geographical distribution of trust in others, neighbors, and relatives for the countries in the sample. The country-average values of each variable reveal that, as expected, trust in others tends to be lower than trust in neighbors, which is in turn lower than trust in relatives.

Afrobarometer also collects information on other individual characteristics, including age, gender, religion, education, urban location, living conditions, and employment status. Some of these variables serve as proxies for income, which is not directly measured. Lastly, using the Afrobarometer district level of aggregation,¹¹ we compute the share of the population belonging to the same ethnicity as the respondent.

4.2. Slave Exports Data

Data on historical slave exports are taken from Nunn and Wantchekon (2011), who generate ethnicity-level estimates based on country-level slave export figures from Nunn (2008). Among the four slave trades, only the transatlantic and Indian Ocean trades have ethnicity data detailed enough to construct reliable estimates of the number of slaves taken from each ethnicity. In estimating the number of slaves taken from each ethnic group, Nunn and Wantchekon (2011) match the ethnic identities in historical records to the ethnic classification in the Afrobarometer surveys, linking the original ethnic groups to the classification constructed and mapped by Murdock (1959).¹² Their baseline measure of the slave trade, which is adopted here, is the natural log of 1 plus slave exports normalized by land area. This measure is normalized by the size of the ethnic groups and employs a precisely measured denominator which is available for all ethnic groups in the sample.

4.3. Additional Geo-historical Data

From Nunn and Wantchekon (2011) we also take a number of geo-historical controls that they collect from a variety of sources. These controls include various ethnicity-level measures of the influence of colonial rule, such as the number of Christian missions established during the colonial period and two binary variables respectively reflecting contact with the routes of European explorers and the colonial railway network. They also include a measure of the malaria ecology as a proxy for the initial disease environment, which according to Acemoglu, Johnson, and Robinson (2001) contributed to determining settlement patterns. Finally, to capture the initial level of prosperity, they code a binary variable capturing the presence of a city in 1400 on the land inhabited by each ethnic group.

4.4. Ethnographic Data

We rely on data from Murdock (1967) for a variety of ethnographic controls aimed at capturing ancestral characteristics of ethnicities that may influence contemporary attitudes toward migrants. Again closely following Nunn and Wantchekon (2011), the main analysis includes pre-colonial settlement patterns and the number of jurisdictional hierarchies beyond the local community. For extensions of the main analysis, we add four variables describing family structure and descent systems which, according to Enke (2019), measure kinship tightness. Accordingly, from the corresponding categorical variables in Murdock (1967)—domestic organization, transfer of residence at marriage, descent type, and community marriage organization—we define four binary variables which respectively capture the presence of extended (rather than nuclear) families, post-marital coresidence, unilinear (rather than bilateral) descent, and localized clans.

Additional variables, which will be employed in the section on alternative explanations of jus soli attitudes, will be discussed as they are introduced. Variable description and sources are detailed in table S2.1, while table S2.2 reports summary statistics, separately for individual-level and ethnicity-level data.

4.5. Empirical Strategy

To investigate the long-term impact of the slave trade, through trust, on jus soli attitudes, we estimate variants of the following empirical models:

where equation (1) captures the proximate association between trust and jus soli attitudes, equation (2) the reduced-form role of the slave trade as an underlying deep determinant of jus soli attitudes, and equation (3) the channel of transmission represented by the effect of the slave trade on trust. The variables JusSoli_i,e,d,c and Trust_i,e,d,c are indexed by individual i, ethnic group e, Afrobarometer district d, and country c, while SlaveTrade_e is measured at the ethnicity level. Country fixed effects α_c are included to control for country-specific institutional and policy factors (e.g., actual citizenship laws and migration policies).¹³

The models also control for a large set of potential confounders at various levels of disaggregation, including, at the ethnicity level, geo-historical (Christian missions, colonial routes, colonial railway, malaria ecology, and cities in 1400) and ethnographic (settlement patterns and jurisdictional hierarchies beyond the local community) variables. Variants of the models further control for individual-level characteristics (age, age squared, gender, religion, education, urban location, living conditions, and employment status) and the district-level share of the population belonging to the same ethnicity as the respondent. To account both for the fact that ethnicity-level controls are constant at such a level, causing serial correlation within ethnic groups, and for the Afrobarometer sampling design, robust standard errors are computed by two-way clustering at the ethnicity and Afrobarometer district levels.

5. Estimation Results

The empirical analysis begins by estimating the relationship between the level of trust and citizenship attitudes toward granting birthright citizenship through the application of the jus soli principle to second-generation immigrants, as expressed by equation (1). We start with the measure of trust in others, which should more closely capture attitudes toward foreigners. In table 1, model 1 presents a parsimonious specification that only adds country fixed effects to the focal regressor. In models 2 and 3, ethnicity-level geo-historical and ethnographic controls are added sequentially. Despite the associated loss of observations, a positive correlation (after partialling out the other controls) with stable size and high statistical significance for the coefficients is confirmed in all specifications, offering support for the hypothesis that trust is a reliable proximate determinant of jus soli attitudes, with more trusting individuals showing more open attitudes toward the integration of foreigners. With reference to the full specification in model 3, a one-standard-deviation increase in trust produces an average increase of 2.6 percentage points, or of 4.6 percent, in the dependent variable, relative to its sample mean (0.563 in the estimated sample, as reported in the table).

One concern with the regression results in table 1 is that other ethnicity-specific factors may be correlated with jus soli attitudes and trust as well. In order to filter them out, table S2.3 runs variants with ethnicity fixed effects. Since ethnicity-level variables are absorbed by the fixed effects, in model 1 we only control for trust and ethnicity fixed effects, while in model 2 we also reinsert country fixed effects. The smaller coefficient size suggests that the effect of trust on jus soli attitudes is somewhat attenuated as the bias is reduced, but previous results are confirmed, pointing to a robust association between attitudes toward citizenship for foreigners and trust in others.

A further concern is that the estimates may be biased because of unobserved heterogeneities across groups. Despite the large number of covariates employed to control for observable factors, there may still be unobservable ones that may be correlated with the focal variables. In order to gauge this possibility, we rely on the method provided by Oster (2019) which, building on Altonji, Elder, and Taber (2005), assesses how large the bias due to unobservables should be, in comparison to that due to observables, in order to explain away the estimated effect. The ratio between the two components of the bias is denoted by δ. Table 1 reports the values of δ that ensure a zero value for the β coefficient of each model.¹⁴ For all models, the value of δ is reassuringly greater than 1, which implies that selection on unobservables would have to be quite large to explain away the results.

Next, we turn to equation (2) and test the hypothesis that the slave trade can be identified as a fundamental determinant of jus soli attitudes. Indeed, the intensity of slave exports exerts a negative effect on jus soli attitudes (table 2), suggesting that individuals belonging to ethnic groups that were more exposed to the slave trade are more opposed to jus soli. This is true across all specifications. In the fully controlled model 3, the effect of a one-standard-deviation increase in the measure of the slave trade produces an average decrease in trust of 4.1 percent, relative to its sample mean.

As a third and last step, to investigate the channel running from the slave trade to contemporary jus soli attitudes, we estimate equation (3) and confirm that trust is negatively related to the slave trade (table 3), as already established by Nunn and Wantchekon (2011).¹⁵ Again, this result holds across specifications. In model 3, the effect of a one-standard-deviation increase in the slave trade produces an average decrease of almost 8 percent in trust, relative to its sample mean. This finding corroborates the hypothesis that the effect of the slave trade on citizenship attitudes is indeed channeled through trust.¹⁶

In the following, a number of robustness checks and extensions are performed, using as a benchmark the fully specified models (model 3 in tables 1–  3).¹⁷

Accounting for “Don’t Know” Responses So far, the analysis has relied only on categorizations of responses excluding “don’t know” responses. This choice could be problematic, since being informed about an issue is itself a potentially endogenous decision. To address this issue, we rerun the regressions presented in tables 1–  3 using alternative codifications. The first consists in assigning the middle value (respectively, 0.5 and 1.5) to “don’t know” responses about citizenship and trust, while the second consists in assigning the minimum value (0). As a result of this adjustment, we gain approximately 1,000 additional observations for citizenship and 200 for trust. Reassuringly, under both alternatives, the results remained nearly identical (though not reported for brevity).

Alternative Citizenship Policy Provisions The analysis is extended to dependent variables capturing attitudes toward naturalization and dual citizenship, which are considered less salient issues relative to jus soli (table S2.4). Predictably, a positive albeit attenuated influence of trust in others is confirmed for both. Likewise, a still negative but no longer significant impact of the slave trade is confirmed. This conclusion holds when the dependent variable is an index capturing the three dimensions of citizenship acquisition together.¹⁸

Alternative Measures of Trust A further robustness check consists in probing whether previous results concerning trust in other people extend to trust in neighbors and relatives. While attitudes toward foreigners are expected to align more naturally with trust in other people, it is well known that alternative measures of trust tend to be correlated between themselves. As expected, the influence of trust in neighbors and relatives on jus soli attitudes is still positive, but the size of the coefficients and, for the latter, its significance, are reduced (table S2.5, models 1 and 2). The impact of the slave trade on both measures of trust is confirmed (models 4 and 5).¹⁹ Rather than the absolute level of trust in a specific group, Moscona, Nunn, and Robinson (2017) and Enke (2019) look at the difference between the in-group and out-group levels of trust. The main results hold when trust in others is replaced with the trust gradient, defined as the difference between the level of trust in others and in relatives (table S2.5, models 3 and 6).

Individual-Level Controls Variants of the main models additionally control for individual-level covariates and, in a further specification, for the contemporary district-level share of the population belonging to the same ethnicity of the respondent. Previous results are largely confirmed, despite a loss of significance for the coefficient on the slave trade in the regressions where jus soli is the dependent variable (table S2.6).²⁰ However, we prefer to continue with the previous benchmarks that exclude these additional variables since they may represent “bad controls,” even more so in our long-term perspective.²¹

The Spatial Distribution of Slave Exports One potential concern with the results is that, even though the slave trade affected a large portion of the African continent, they may be driven by its spatial distribution. In order to ensure that the findings do not depend on a broad comparison between individuals belonging to ethnicities that were affected by the slave trade and those that were not, we perform a robustness check by dividing individuals in two sub-samples. The estimates show an even stronger effect of trust on jus soli attitudes over the slave trade sample (table S2.7, model 1), while the effect is considerably reduced over the sample that was not affected by the slave trade (model 4), suggesting that the underlying variation in the intensity of slave exports is an important driver of the association between jus soli attitudes and trust. In models 2 and 3, which are again restricted to the slave trade sample, the coefficients on the slave trade remain similar to those reported in tables 2 and 3.

6. A Sample of Movers

This section investigates whether the results we have obtained so far extend to an alternative sample, as an external validity argument. In order to do so, it zooms in on the behavior of movers. The epidemiological approach based on the behavior of migrants has been applied to outcomes related to the labor market (Ichino and Maggi 2000), financial decisions (Guiso, Sapienza, and Zingales 2004), the position of women in society (Fernández 2007; Fernández and Fogli 2009), living arrangements (Giuliano 2007), and family formation (Bertocchi and Dimico 2020). Following Nunn and Wantchekon (2011), Michalopoulos, Putterman, and Weil (2019), and Tedeschi (2018), who refer to a context similar to the present one, we focus on the movers in the sample to investigate the portable component of ancestral influence on current outcomes.

By exploiting the map of ethnic borders provided by Murdock (1959), movers are defined as individuals who are no longer living in the ancestral homeland of the ethnic group they belong to, even though they may have not moved during their lifetime. Movers represent 56 percent of the full sample (table S2.2).²² To ensure that the distribution of movers by ethnicity is not skewed, in such a way that a few ethnicities associated with a large number of movers may be driving the results, we also check the average share of movers across ethnicities, which is even higher at 65 percent.²³  Table S2.8 compares the characteristics of movers and non-movers using a t-test for differences in means. While attitudes toward jus soli and trust in others are similar across the two samples, and the same applies to historical and ethnographic characteristics, movers exhibit rather predictable differences at the individual level, such as lower age, higher education, better living conditions, and higher chances to live in an urban area and with a smaller share of people of the same ethnicity. Hence, the movers sample is likely affected by selection bias. Nevertheless, restricting the analysis to it allows us to filter away the influence of local factors that may bias our results within the full sample.

Table 4 replicates tables 1–  3, separately for the movers and non-movers in the original sample.²⁴ For movers, in models 1, 4, and 7, we enter previous controls including country fixed effects. In models 2, 5, and 8, we replace the latter with fixed effects for the ethnicity of destination. That is, rather than the movers’ ancestral homeland, these fixed effects capture the ethnicity whose historical homeland coincides with the area where they currently live. Adding these fixed effects should eliminate all potential local confounders, including time-invariant characteristics of the geographic, economic, institutional, and cultural environment where they currently reside. Comparing each pair of regressions for movers (i.e., the regression with country fixed effects and the regression with ethnicity of destination fixed effects) shows that the results hold after applying the second, more stringent strategy, despite a decrease in the size and significance of the coefficient on the slave trade in model 8. In models 3, 6, and 9, we apply an even more demanding strategy by including both country and ethnicity of destination fixed effects, and again our results substantially hold.

The main takeaway from table 4 is that both the proximate relationship between jus soli and trust and the underlying ones involving the slave trade are confirmed over the movers sample, which provides validation of the hypothesis that the deep influence of the slave trade on sentiments toward foreigners is channeled through trust.

7. Alternative Explanations

The last part of the paper is devoted to investigating whether other factors—other than the slave trade—can represent alternative deep determinants, through trust, of attitudes toward citizenship. Indeed, several other determinants of trust have been established by the literature, and one may wonder whether they can also contribute to shape citizenship sentiments. We test this conjecture for factors that have been found particularly relevant within the history of African development, namely, historical conflict, kinship tightness, and witchcraft beliefs. We will conclude that for none of them is this conjecture warranted by the data.

7.1. Historical Conflict

Another potential determinant of trust that has received attention, particularly within Africa, is conflict. Indeed, historical rivalries involving violence may reverberate on beliefs and feelings, affecting one’s willingness to trust. Rohner, Thoenig, and Zilibotti (2013) use contemporary measures of fighting events and, by exploiting variations in both their spatial and ethnic intensity, show that more intense fighting decreases trust in Uganda. Besley and Reynal-Querol (2014) employ historical data from the Conflict Catalog by Brecke (1999) and establish a negative effect of conflict on trust across African countries.

In principle, through the channel of trust, historical conflict may also be associated with attitudes toward foreigners and the laws that regulate access to citizenship. To test whether this is the case, we also rely on the Conflict Catalog, which consists of a listing of all recorded violent conflicts that meet the magnitude 1.5 or higher criterion (i.e., 32 or more deaths per year).²⁵ Crucially for the present investigation, for conflicts occurring in Africa, the ethnic groups involved, often in the form of historical kingdoms, are also recorded. Thus, we are able to match them with the ethnicities in Murdock (1959).²⁶ While data are collected by Brecke (1999) from 1400 AD to 1998, in order to avoid endogeneity issues we consider conflicts occurring up until 1912.²⁷ We obtain a listing of 68 ethnic groups, out of the 272 in the sample, that went through conflicts, corresponding to 40 percent of the individuals in the sample. The main measure that we employ is the number of years in conflict in the period 1443–1912. On average, each individual in our sample is associated, through the ethnic group of origin, with 1.3 years in conflict over a range of 0–47 (table S2.2).

Consistent with the literature, within the Sub-Saharan African sample presently employed, the intensity of historical conflict is negatively associated with trust (table S2.9, model 1). However, no significant association is found between historical conflict and jus soli attitudes (model 4). Thus, even though we assess that they concur to shaping trust, historical conflict cannot be proposed as an alternative to the slave trade as a deep determinant of attitudes toward citizenship.

7.2. Kinship Tightness

Next, we turn to family ties and kinship tightness as alternative determinants of citizenship attitudes. The literature has extensively documented the negative relationship between generalized trust and family ties (e.g., Alesina and Giuliano 2014). Over data covering several countries located in different continents, Enke (2019) shows that kinship tightness—captured by an index reflecting various components of family ties—is positively associated with trust in family and neighbors and negatively associated with trust in other groups.²⁸ The potential association between kinship tightness and attitudes toward foreigners may be justified by the idea that tighter communities are more averse to strangers, including foreigners seeking citizenship. If the association between kinship tightness and jus soli attitudes is the one we conjecture, we should expect kinship tightness to negatively affect trust in others, as well as jus soli attitudes.

Following Enke (2019), we measure kinship tightness as an index consisting of the unweighted average of four binary variables coded from the Ethnographic Atlas, after restricting the sample to those ethnic groups for which at least three of the four variables are available. Two of them, the presence of extended families and post-marital coresidence, jointly capture family structure, while the other two, the nature of descent and the presence of localized clans, jointly capture descent systems. Kinship tightness is positively associated with extended families, post-marital coresidence with a spouse’s group, unilinear descent systems, and segmented communities like clans. Overall, as revealed by the summary statistics in table S2.2, Sub-Saharan Africa displays a high degree of tightness in all four dimensions.

Over the present Sub-Saharan African sample, kinship tightness is actually associated positively with trust in others, rather than negatively as in the world sample in Enke (2019) (table S2.9).²⁹ This discrepancy can be attributed to the fact that he captures between-country, rather than within-country variation as we do. Furthermore, within Sub-Saharan Africa, kinship tightness is not associated with jus soli attitudes. Its coefficient in model 5 is actually negative albeit insignificant, which seems at odds with the positive coefficient in the regression for trust. Thus, across Sub-Saharan Africa, family ties as captured by kinship tightness cannot represent an alternative to the slave trade as fundamental determinants of citizenship attitudes, since they are unable to generate the same distinctive pattern of associations observed with the slave trade.

7.3. Witchcraft Beliefs

Witchcraft beliefs have been proposed as a fundamental determinant of trust by Gershman (2016). We use the same Pew data he used on witchcraft beliefs (even though they involve a significant loss in the number of observations).³⁰ Through the channel of trust, witchcraft beliefs may turn out to be correlated with negative attitudes toward foreigners. Indeed, according to Gershman (2016), belief in witchcraft as an explanation for all kinds of misfortunes is often associated with aggressiveness, hostility, fear of mobility, and a culture of suspicion—factors that are not conducive to welcoming strangers.

When testing the potential role of witchcraft beliefs as the channel linking trust to jus soli attitudes, one should keep in mind that, despite being deeply entrenched, they are both derived from contemporary surveys. Consistent with Gershman (2016), the estimates do replicate for witchcraft beliefs the negative association with trust in others that we obtain for the slave trade (table S2.9, model 3). However, no significant association emerges between witchcraft beliefs and jus soli attitudes, even though the coefficient’s sign is, as conjectured, negative (model 6).

Therefore, while we can confirm that witchcraft beliefs concur in determining trust, they cannot be proposed as an alternative to the slave trade as a deep determinant of attitudes toward citizenship.

To sum up, none of the alternative determinants of trust can replicate the distinctive pattern of associations between trust and attitudes toward citizenship that emerges from the slave trade. This conclusion is supported by a further set regressions where all variables are entered together, without or with the slave trade (table S2.10). While the slave trade consistently exhibits a negative influence on both trust and jus soli attitudes, this is not the case for either historical conflict, kinship tightness, or witchcraft beliefs.

8. Conclusion

The empirical evidence produced in this paper documents that, in Sub-Saharan Africa, attitudes toward citizenship acquisition for foreigners do reflect the persistent legacy of the slave trade. The channel of transmission is the distrust of strangers determined by the legacy of the slave trade. Specifically, individuals displaying higher levels of trust exhibit more favorable attitudes toward acquisition of citizenship at birth through the application of the jus soli principle for children of foreigners. In turn, these individuals’ ethnic groups are those that experienced relatively lower exposure to historical slave exports. Alternative theories of trust determination, based on conflict, kinship tightness, and witchcraft beliefs, fail to replicate the distinctive pattern of associations observed with the slave trade.

As argued by Herbst (1997), in multiethnic societies like those prevalent in Africa, liberal citizenship policies can contribute to building a shared national identity. Conversely, restrictive attitudes can imply the alienation of entire migrant and refugee communities, whose size reflects yet unresolved historical repercussions of artificial border establishment and exacerbated ethnic conflicts during colonization. Furthermore, as demonstrated by Imam and Kpodar (2020), less inclusive citizenship policies can become a drag on development outcomes, especially in countries with weaker institutional environments. The present findings add to this perspective by highlighting that attitudes toward citizenship acquisition are deeply shaped by the experience of the slave trade, predating the colonization period by centuries. Although admittedly the influence of the slave trade through trust is merely one of the potential determinants of individual attitudes, these attitudes—as reflected in survey data—reflect and ultimately shape, through voting, actual policies and regulation. Hence, the implications of the present findings are that historical legacies will persist in influencing sentiments toward foreigners, migration policies, and citizenship laws, as well as economic development, for the years to come.

Data Availability Statement

The data underlying this article are available in the Inter-university Consortium for Political and Social Research Repository, at https://doi.org/10.3886/E199242V1.

Funding

This work was supported by an Italian University Ministry PRIN 2017 grant.

Author Biography

Graziella Bertocchi (corresponding author) is a professor at the University of Modena and Reggio Emilia, Modena, Italy, and holds affiliations with EIEF, CEPR, Dondena, GLO, and IZA; her email address is [email protected]. Arcangelo Dimico is a senior lecturer at Queen’s University Belfast, Belfast, United Kingdom, and holds affiliations with GLO, IZA, CEPH, and QUCEH; his email address is [email protected]. Gian Luca Tedeschi is a post-doctoral fellow at the University of Bergamo, Bergamo, Italy; his email address is [email protected]. The authors thank the editor Eric Edmonds, three anonymous referees, and participants at the Jan Tinbergen European Peace Science Conference, CEPR ESSIM, the Societa’ Italiana degli Economisti Conference, the AE-CEU-GLO Labor Symposium, and seminars at the University of Bolzano and IOS-IZA-HSE for insightful comments. The authors also thank Francesco Stradi for excellent research assistance. A supplementary online appendix is available with this article at The World Bank Economic Review website.

Footnotes

References