Explaining the relationship between population religiosity and state support for religion

Why do we expect countries in sub-Saharan Africa to have a high level of state support for religion—that is, a greater frequency of laws, state institutions, and state funds thatprivilege religious groups, beliefs, or practices? Figure 1 lists some popular postulations (hereafter, P) and confounding factors (hereafter, C). Religious citizens expect state support for religion (P1), and states usually respond (P2). Thus population religiosity predicts state support for religion (P3). Since countries in sub-Saharan Africa have high levels of population religiosity (P4) we expect high levels state support for religion in the region (P5). However, several factors can confound this relationship: the political theology of the majority religion (C1), religious actors' lack of moral authority (C2), the need for the state to manage religious tensions (C3), weak democratic mechanisms (C4), and the state's inability to meet popular demands (C5). The following section reviews evidence for each proposition and confounding factor.

Figure 1.

Population religiosity and state support for religion.

Existing research links the number of religiously devout citizens and the state's policies involving religion. Studies have shown that highly religious individuals are more likely to support religious influence in politics (Buckley, Reference Buckley2019), advocate for public officials with strong religious beliefs (Driessen, Reference Driessen2018), and view their religion as integral to their national identity (Grzymała-Busse, Reference Grzymała-Busse2015). If these beliefs indicate a desire for the state to endorse religious beliefs, groups, and practices through favorable laws and institutions, it is likely that religious citizens will demand state support for religion (P1). Likewise, states have a strong motivation to satisfy the demand for religious support, as it can legitimize the government. Discontent religious groups can potentially destabilize a state (Juergensmeyer, Reference Juergensmeyer and Haynes2019; Schliesser, Reference Schliesser and Haynes2019), and political parties may gain support by attracting religious voters (Kirdis, Reference Kirdis and Haynes2019; Satana et al., Reference Satana, Ozkan, Birnir and Haynes2019). Additionally, in highly religious countries, religious leaders may have a greater ability to mobilize the public and influence the state. A study by Rosenberg and Smith (Reference Rosenberg and Smith2021) found that in 18 democracies, religious leaders are more involved in politics in countries with lower levels of “human development,” which is linked to higher levels of religiosity (Norris and Inglehart, Reference Norris and Inglehart2011). If the state can appease religious groups and attract voters by providing support through favorable laws and institutions, it is likely to respond to citizens' demands for state support for religion (P2).

In a study of 29 countries, Grzymała-Busse (Reference Grzymała-Busse2015) found that churches are more likely to influence state policy in countries where the population sees an overlap between their religious and national identities, although she notes that this relationship is rarely straightforward. Similarly, using data from the World Values Survey (WVS), Fox and Tabory (Reference Fox and Tabory2008) found a positive association between government involvement in religion (GIR) and whether people consider themselves religious. However, their results are equivocal, as they also find a negative association between GIR and belief in an afterlife and attendance at religious services. In sum, while the relationship between population religiosity and state support for religion is credible and substantiated, it is also complex.

It is common to note that populations in sub-Saharan Africa are very religious (P4). Abbink (Reference Abbink2014, 84) believes that “few people would contest the continued importance that religion holds in African life.” Using data from the PRC, Abbink notes that “In 2010, 69 to 98 percent of inhabitants (depending on the country) were believers in God, the literal truth of the scriptures and Biblical or Qur'anic rules and went to mosque or church regularly.” Similarly, 9 out of 10 said that religion was “very important” in their lives’ (citing Pew Research Centre, 2010, 3; Abbink, Reference Abbink2014, 86). Moreover, global research on the determinants of religiosity predicts that people in sub-Saharan Africa would be very religious. Norris and Inglehart (Reference Norris and Inglehart2011, 18) theorize that experiences of insecurity, like threats to livelihoods and physical safety, increase individuals' religiosity. They find that the human development index (HDI), which they use as a proxy for security, is among the most significant predictors of irreligiosity; in other words, respondents consider religion to be less important in their lives as the HDI of their country increases. Using income inequality as an alternative proxy for insecurity, Karakoc and Baskan (Reference Karakoc and Baskan2012) make similar findings. Dhima and Golder (Reference Dhima and Golder2021) likewise concluded that people in countries with lower levels of economic development and social support depend more on religious institutions for social services and are, therefore, more religious. As countries in sub-Saharan Africa have low levels of economic development and social security, we, therefore, expect populations in sub-Saharan Africa to have high levels of religiosity.

Echoing the argument that a high level of religiosity leads to more religiously responsive states in sub-Saharan Africa (P5), Obadare (Reference Obadare2018, Reference Obadare2022) describes Nigeria's Fourth Republic as “more or less a clerical state” (2022, 7). He argues that President Goodluck Jonathan “enacted policies that could easily be construed as being driven by a desire to retain the goodwill of the Pentecostal elite and their congregations” (Obadare, Reference Obadare2018, 112). Drawing on his fieldwork in Ghana, Nigeria, and the Democratic Republic of Congo, Gifford (Reference Gifford2016, 152, 47, 156) argues that the “enchanted religious imagination of so many Africans” impedes the development of “rational bureaucratic” policies and is thus incompatible with “a society structured through technology and bureaucracy.” Abbink (Reference Abbink2014, 92) identifies several examples of laws and policies challenged by religious constituents in African countries, especially in constitutional amendments and family law. Fox (Reference Fox2012, 23) indicates that such views are common in narratives of religion in “third world countries,” especially in contexts where scholars believe that “secular modernity has failed.” McClendon and Riedl (Reference McClendon and Riedl2019, 1) note that these views of the role of religion and politics in sub-Saharan Africa are also common in the popular press.

However, the limited quantitative evidence available suggests contrary to expectations, countries in sub-Saharan Africa have much lower levels of state support for religion than the rest of the world. P5 appears to be false. Sub-Saharan Africa has the second highest frequency of legislative provisions that explicitly separate religion and the state. It also has the second lowest frequency of having an official state religion. Sub-Saharan Africa trails the former Soviet countries in each case (Fox and Flores, Reference Fox and Flores2009, 1501). Fox's (Reference Fox2012) analysis of Christian-majority countries corroborates these findings. Although countries in sub-Saharan Africa have more legislation associated with conservative religious values compared to other regions, far fewer of these countries fund religious schools, provide religious education in public schools, fund clergy, or have official government departments concerned with religion. There seems to be an important divide between the presence of religion in political rhetoric and theatre, and the effective influence of religion on laws and policies (see Obadare, Reference Obadare2018, 7; Jeffery-Schwikkard, Reference Jeffery-Schwikkard2022, 1077). The supposition that religion is central to governance in sub-Saharan Africa appears to be wrong.

Why might sub-Saharan Africa have such low levels of state support for religion? Five factors have been prominently featured in analyses of the region (Figure 1): (i) religious traditions that object to the state's involvement in religion, (ii) religious actors lacking moral authority among the population, (iii) the need to defuse religious tensions, (iv) low levels of democracy, and (v) the incapacity of the state to support religion and challenge P4 (see Fox, Reference Fox2007, Reference Fox2010; Buckley and Mantilla, Reference Buckley and Mantilla2013; Abbink, Reference Abbink2014; Grzymała-Busse, Reference Grzymała-Busse2015; Driessen, Reference Driessen2018).

Different political theologies may influence either citizens' or politicians' preferences for state support for religion (C1). Certain religious traditions may advocate greater integration between religious institutions and the state, while other traditions may reject this integration. Differences between political theologies have been hypothesized to explain the variations in religion–state relationships depending on whether most of the population is Muslim or Christian (see Driessen, Reference Driessen2018 and Cesari and Fox, Reference Cesari and Fox2016 for discussion and critique). Fox's (Reference Fox2007, 15) analysis of RAS finds that states with a Muslim majority have higher levels of religious legislation. Consequently, if countries in sub-Saharan Africa are predominantly Christian, this may explain low levels of state support for religion despite there being highly religious populations.

Religious citizens' preferences and politicians' strategies may depend on the moral authority of religious actors as defenders of the nation (C2). This moral authority is more likely when citizens' religious and national identities overlap. Conversely, religious citizens may be less likely to demand that the state support religion if they see their national identity as independent of their religion. Grzymała-Busse (Reference Grzymała-Busse2015) identifies several factors that are conducive to a fusion of national and religious identities: ethnic homogeneity, the dominance of one religion, the control of religious institutions over education and public rituals, and the historical record of religious actors in unifying and defending the nation. In the absence of these factors, politicians and religious citizens may see little reason for the state to support religion and may not trust religious institutions to serve the interests of the nation.

Citizens or politicians may wish for limited state involvement in religion to avoid inflaming religious tensions and protect their own religious freedoms (C3). This could be to avoid accusations of state favoritism, or because religious citizens in pluralistic countries object to state support for religion which might benefit other religious groups. Abbink (Reference Abbink2014, 101) postulates there may be a greater need for a secular state “the more religious people become” in order for people to embrace their religion freely. Fox (Reference Fox2007, 16) finds that “religious diversity is consistently associated with less government involvement in religion” and that a strong separation of religion and the state is “a popular strategy for mitigating religious tensions in religiously diverse societies.” Consequently, in more religiously diverse countries, we might expect lower levels of state support for religion.

Political actors may not need to meet popular demand for state support for religion if they are not dependent on popular support (C4). Supporting religion is costly to the regime. It requires resources and acceding to the influence of religious actors restricts political actors' power (Grzymała-Busse, Reference Grzymała-Busse2015). Regimes are more reliant on popular support in democracies, as political actors may lose power imminently if they do not meet the demands of the electorate. Consequently, undemocratic states, or those in democracies that are less robust, may be less likely to meet the public's demand for state support for religion.

Finally, perhaps the explanation most frequently offered by low levels of state support for religion in sub-Saharan Africa is that states in the region lack the capacity to provide this support (C5). Fox (Reference Fox2010, 529, Reference Fox2012) and Cupery (Reference Cupery2013) argue that many governments in low- and middle-income countries do not have the resources and capacity to significantly regulate religion. Buckley and Mantilla (Reference Buckley and Mantilla2013) find evidence for this, demonstrating that countries with lower levels of government effectiveness have lower levels of state support for religion. If countries in sub-Saharan Africa have lower levels of government effectiveness, this may explain their incapacity to meet popular demand for state support for religion.

Data

To analyze state support for religion, the analysis assesses the impact of population religiosity, state capacity, religious pluralism, ethnic pluralism, control of religious actors over education, and the frequency of moments of national crisis. These capture the previously listed propositions and confounding factors. HDI and income inequality are used as control factors. The Religion and State Round 3 (RAS3), used to analyze state support for religion, counts the number of types of religious support provided by the government and noting 52 types of support. Such support includes laws concerning sex, religious education, blasphemy, funding for religious institutions, and whether positions in government are reserved for religious leaders.

RAS3 presents data on 183 countries (from 1990 to 2014) gathered from constitutions, legislation, policy papers, news articles, and reports from nongovernmental and intergovernmental organizations, at both the national and regional levels (Fox, Reference Fox2011, 16). The RAS3 index weighs all provisions equally (Fox, Reference Fox2011). While other authors have analyzed the RAS3 index without any transformations (see Buckley and Mantilla, Reference Buckley and Mantilla2013), this analysis used a logarithmic transformation of the RAS3 index for linear regressions. In line with other analyses of RAS (see Fox, Reference Fox2008; Fox and Flores, Reference Fox and Flores2009; Buckley and Mantilla, Reference Buckley and Mantilla2013; Buckley, Reference Buckley2016, Reference Buckley2019), and similarly constructed variables (see Gurr, Reference Gurr1993, Reference Gurr2000; Fox and Flores, Reference Fox and Flores2009), it treats the index as continuous. Whether countries are in sub-Saharan Africa is based on classification provided by the United Nations Development Programme (UNDP, n.d.).

Two datasets are used to determine the level of religiosity in the population. WVS wave 6 (WVS6) includes data on 60 countries between 2010 and 2014 (Inglehart et al., Reference Inglehart, Haerpfer, Moreno, Welzel, Kizilova, Diez-Medrano, Lagos, Norris, Ponarin and Puranen2014). Although WVS6 is widely used in studies of religiosity, it includes only five countries in sub-Saharan Africa. This is true of WVS generally: between 1981 and 2022, WVS has only included 11 sub-Saharan African countries and no more than seven sub-Saharan African countries in any one wave (e.g., 2006 and 2007). However, the PRC included several surveys on religiosity from 2007 and 2018, covering 106 countries. While the timeframe for PRC does not match neatly with the data available from RAS3, the PRC dataset includes 20 countries in sub-Saharan Africa and is therefore the primary dataset for this analysis.Footnote ¹

There are several limitations to the PRC dataset. Although it has many more countries in sub-Saharan Africa than WVS, the number of countries included is still relatively small. The sample reflects only 43% of the total number of countries in sub-Saharan Africa. While the sample is well-dispersed across the continent, there remains a chance that it is not representative of the region. While the Afrobarometer is an alternative to the PRC, the datasets on religiosity do not include non-African countries. Consequently, it could not be used for a broader comparative analysis.

PRC and WVS6 collect data on the importance of religion or God in the respondent's life and how frequently the respondent attends religious services in addition to weddings and funerals. The analysis uses the proportion of respondents in each country who answered that God or religion was “very important” in their lives and the proportion of respondents who attended religious services at least weekly. The analysis primarily uses data on the importance of religion in one's life as this is less likely than attendance to be confounded by other factors such as employment and reliance on religious institutions for social services (as identified in Dhima and Golder, Reference Dhima and Golder2021, 37). The data on weekly attendance are used to test for robustness.

There is a very strong correlation between the PRC and WVS6 data (p < 0.01) for indicators of the personal importance of religion (r = 0.826) and weekly attendance (r = 0.902). This suggests that the findings of this article will be comparable to studies that use the WVS6 dataset. To test the role of Islam, a dummy variable (whether Islam is the largest religious group in a country) is formed from the 2010 PRC data (Pew Research Center, 2015). Within sub-Saharan Africa, this is primarily a comparison between countries where most of the population is Christian or Muslim. Only two countries, Côte d'Ivoire and Mauritius, do not have Christian or Muslim majorities. The influences of religious and ethnic diversity draw on the Alesina et al.'s (Reference Alesina, Devleeschauwer, Easterly, Kurlat and Wacziarg2003) religious fractionalization index (RFI) and ethnic fractionalization index (EFI). The RFI measures the diversity of major world religions based on data from the Encyclopedia Britannica, but does not capture diversity within religious traditions, and may thus overlook diversity within Christianity and Islam in sub-Saharan Africa. The EFI is based on data from the Encyclopedia Britannica, the CIA World Factbook, Levinson (Reference Levinson1998), and Minority Rights Group International. Both indexes have not been updated since 2003 while for both RFI and EFI, however. For both indexes, higher values indicate greater diversity.

To measure state capacity, I follow Buckley and Mantilla (Reference Buckley and Mantilla2013) in using the World Bank's Government Effectiveness Index, which is based on perceptions of households, firms, nongovernmental organizations, and public sector organizations (World Bank, 2020). The index ranges between −3 and 3, with higher scores representing higher state capacity. To investigate the role of the quality of democracy, I use PolityIV, which measures the presence of democratic participation, institutionalized constraints on the executive, guarantees of civil liberties, freedom of press, and political competition (World Bank, 2018). PolityIV has been used to measure the quality of democracy in several studies of religion and the state (Buckley and Mantilla, Reference Buckley and Mantilla2013; Fox, Reference Fox2012; Buckley, Reference Buckley2016; Larsson, Reference Larsson2019; Rosenberg and Smith, Reference Rosenberg and Smith2021). Higher scores indicate greater democratic accountability.

To assess the role of moral authority, I first use two brief historical portraits. This approach has two advantages. First, it helps to assess how and why moral authority may be relevant to state support for religion in sub-Saharan Africa. Second, it provides qualitative insight into the relevance of other factors in the statistical analysis, specifically the role of state capacity and majority religion. These portraits illustrate how the factors identified as relevant to moral authority by Gryzmała-Busse (Reference Grzymała-Busse2015)—that is, religiosity, religious and ethnic homogeneity, religious actors' control over education, and political instability—may be salient to state support for religion. The subsequent analysis also tests whether these factors are more prevalent in sub-Saharan Africa compared to the rest of the world.

The case studies focus on Mozambique and Rwanda to represent countries in the dataset that are below the 25th percentile for state support for religion and above the 75th percentile for popular religiosity. Countries meeting these two conditions typify religious populations in states with little support for religion. Eight of the nine countries in the dataset met these conditions are in sub-Saharan Africa. Mozambique and Rwanda were selected for contrast within this set, as they differ in colonial history, religious and ethnic diversity, and post-independence politics.

The prevalence of religiosity, religious homogeneity, and ethnic homogeneity in sub-Saharan Africa are tested using the above-mentioned variables. While it is difficult to test the prevalence of two further factors identified by Grzymała-Busse (Reference Grzymała-Busse2015)—(1) religious institutions' control over education and (2) the historical record for religious actors in defending or unifying the nation—proxy variables are available. If school enrolment is low, then religious actors cannot have significant control over socialization through education. Even if religious actors control all educational institutions, the reach of such institutions would be too low to foster the fusion of religious and national identity in the population. To compare school enrolment, the analysis incorporates historical data on national primary school enrolment provided by Barro and Lee (Reference Barro and Lee2013). Second, while it is challenging to build moral authority, it can be quickly lost if religious actors pick the “wrong side” in a national crisis. Grzymała-Busse (Reference Grzymała-Busse2015, 22, 39) observes that moral authority is a “brittle” resource that can be easily lost if religious actors do not “tread carefully.” This means that religious actors' moral authority is likely lower in contexts with more frequent moments of national crisis. This is because more moments of national crisis allow more opportunities for religious actors to misstep, and a misstep reduces moral authority more easily than a correct step improves moral authority. Successful coup d’états are an easily identifiable example of a moment of national crisis. To compare the frequency of coup d’états, I use data from 1950 to 2008 provided by Powell and Thyne (Reference Powell and Thyne2011). While both school enrolment and coup d’état indicators are imperfect proxies, they provide a means of exploring whether the qualitative analysis from the historical portraits may be more widely applicable.

Finally, the analysis has two control variables: the Gini index where higher values indicate greater income inequality, draws on the World Bank's Development Research Group (World Bank, 2022). HDI, a composite measure of gross national income per capita, life expectancy, and mean and expected years of schooling, come from UNDP (UNDP, 2020). They help to assess whether the effect of a country being in sub-Saharan Africa could be explained simply by lower levels of economic development and higher levels of inequality.

To increase comparability between these three datasets namely, RAS3 (1990–2014), WVS6 (2010–2014), and PRC (2008–2018), the analysis uses an average of the index between 2010 and 2014 for RAS3.Footnote ² The analysis also uses average scores from 2010 to 2014 for other independent variables when these data are available: that is, for HDI, Gini index, and government effectiveness.

The analysis aims to minimize the number of independent variables in each model. This approach recognizes the concerns of Achen (Reference Achen2005) and Clarke (Reference Clarke2009), who demonstrate that the inclusion of even small numbers of additional variables can dramatically affect both the size and direction of other coefficients in the model. This is especially the case if the relationship between independent variables is not well understood, and the complete set of omitted variables is unknown. There are likely variables missing in current statistical analyses of state support for religion in sub-Saharan Africa. Indeed, a key conclusion of this paper is that further qualitative and historical research is required to identify these. Consequently, variables are tested separately and only combined if there is a theory-driven basis for their combination.

The findings show the importance of combining a quantitative analysis of global datasets with qualitative historical analysis. Global datasets allow us to statistically test general trends, such as the relationship between population religiosity (PRC) and state support for religion (RAS3). Although these datasets allow for extensive comparisons, rather than solely relying on a small selection of case studies, they also simplify complex causal links. Hence, this study's historical portraits explore how certain factors, such as moral authority, might mediate the effect that a religious population has on the state. These portraits illustrate the complex ways the phenomena represented by quantitative indicators can affect the state. Such an approach adds two contributions to the existing studies. It draws attention to global trends based on statistical analyses of multi-country datasets (a “bird's-eye view”). It aids the credibility of the relationships identified in the statistical analysis by exploring how these quantitative abstractions manifest in complex political circumstances.

Analysis

Countries in sub-Saharan Africa have almost half as many provisions of state support for religion despite having twice the level of popular religiosity (see Table 1). Differences between countries in sub-Saharan Africa and countries elsewhere are highly significant for both variables (p < 0.001).

Table 1.

Levels of state support and population religiosity

As illustrated in Table 2 (using data from WVS6 and PRC, the conventional independent variables, and HDI and the Gini index as controls), all models except one, an increase in population religiosity, are associated with increased state support for religion. Given the overall worldwide patterns, the low level of state support for religion in sub-Saharan Africa invites us to explore the reasons behind it.

Table 2.

Religiosity of the population and state support for religion

Note: All coefficients are standardized. Bold indicates p < 0.05.

Why is population religiosity significant when using data from WVS (model 1) but insignificant when using data from PRC without controls (model 2)? The difference is due to PRC's inclusion of more countries from sub-Saharan Africa than WVS. If countries in sub-Saharan Africa are excluded from the PRC (model 3), the relationship between population religiosity and state support for religion becomes very significant. As displayed in Figure 2, the countries in sub-Saharan Africa have both notably high levels of popular religiosity and low levels of state support for religion. If countries in sub-Saharan Africa are removed, the positive relationship between popular religiosity and state support for religion is clear.

Figure 2.

Popular religiosity versus state support for religion.

Figure 2 restates the puzzle clearly: there seems to be disconnect between people's religiosity and state support for religion in sub-Saharan Africa. Moreover, the significance of a country being in sub-Saharan Africa is not explained by countries in the region having low levels of economic development and high levels of inequality. Even controlling for these factors, the effect of a country being in sub-Saharan Africa remains highly significant (model 4).

As Table 2 indicates, contrary to the expectations, the impact of income inequality (the Gini coefficient) is negative and HDI is positive. While Karakoc and Baskan (Reference Karakoc and Baskan2012) found that individuals' support for religious involvement in governance increased with inequality, model 4 finds that state support for religion decreases with inequality. Yet this does not necessarily contradict Karakoc and Baskan (Reference Karakoc and Baskan2012), who explored the impact of individuals' support for religious involvement in government rather than actual state support for religion. Nonetheless, it suggests a complex relationship between religious citizens' demands and the state's response. Similarly, Norris and Inglehart (Reference Norris and Inglehart2011) found a negative association between individuals' religiosity and HDI, yet model 4 indicates a positive association between HDI and state support for religion. This relationship, which is contrary to the expectations of secularization theory, has also been identified by Buckley and Mantilla (Reference Buckley and Mantilla2013), Cupery (Reference Cupery2013), and Fox (Reference Fox2010, Reference Fox2012). All four scholars suggest this may be due to countries with low economic development having limited resources to support religion—a hypothesis tested in this paper.

Since population religiosity generally predicts state support for religion (P3), there must be factors pertinent to sub-Saharan Africa that explain why this relationship differs in the region. The rest of this section explores several confounding factors that may affect whether religious citizens demand state support for religion (P1) and whether states respond to this demand (P2). Table 4 first tests whether the means for four variables are significantly different in sub-Saharan Africa compared to the rest of the world; if not, the variable is unlikely to account for the anomalous relationship in sub-Saharan Africa between population religiosity and state support for religion (see Table 3).

Table 3.

Religion and state's support for religion: confounding factors

Note: Bold indicates p < 0.05.

Countries in sub-Saharan are not more likely to have Muslim-majorities than the rest of the world (p = 0.799). Therefore, this is unlikely to explain why sub-Saharan African countries have lower levels of state support for religion. Conversely, countries in sub-Saharan Africa (1) are modestly more likely to have lower levels of democracy (p = 0.073), (2) are significantly more likely to have higher levels of religious diversity (p < 0.001), and (3) have significantly lower levels of state capacity (p < 0.001). Table 4 explores each of these factors further.

Table 4.

Hypothesized reasons for low state support for religion in sub-Saharan Africa

Note: All coefficients are standardized. Bold indicates p < 0.05.

Models 5–7 test the effect of democratic quality. We would expect the quality of democracy to increase the impact of population religiosity on state support for religion, as the state is more responsive to public demands. The interaction effect between democratic quality and population religiosity is modestly significant for the world (model 5, p = 0.082) and highly significant if countries from sub-Saharan Africa are excluded (model 6, p = 0.006). However, the relationship is the opposite of what we expect: in countries with higher levels of democracy, the impact of population religiosity on state support for religion is reduced. We can speculate that democratic states are more likely to internalize secular values, even if this is contrary to what citizens demand. Since countries in sub-Saharan Africa are more likely to have lower levels of democracy than the rest of the world, we might expect these countries to have high levels of state support for religion. Yet, the analysis indicates that the opposite is true. In any respect, the interaction effect between the quality of democracy and population religiosity loses significance after controlling for HDI and inequality (model 7, p = 0.335).

Models 8–10 explore the relevance of state capacity. As Figure 3 suggests, there is no relationship. This is confirmed in model 8 (p = 0.760). The relationship becomes significant if countries in sub-Saharan Africa are excluded (model 9, p = 0.021). However, this is in the opposite direction to that expected by Buckley and Mantilla (Reference Buckley and Mantilla2013), Cupery (Reference Cupery2013), and Fox (Reference Fox2010, Reference Fox2012): outside of sub-Saharan Africa, levels of state support for religion are higher in states with lower capacity. Regardless, this relationship returns to insignificance once economic controls are applied (model 10, p = 0.995). In contrast, whether a country is in sub-Saharan Africa remains a highly significant predictor whenever it is included (models 9 and 10, p < 0.001).

Figure 3.

Government effectiveness and state support for religion.

Finally, models 11–13 investigate the effect of religious diversity. In all models, religious diversity decreases state support for religion (p < 0.001). The findings confirm Fox's conclusions (2007) and supports a caveat in Obadare's (Reference Obadare2018, 139) analysis of Nigeria: while Pentecostalism might influence struggles for state power and specific policies, religion has had little influence on the federal state itself, owing “to the inherent difficulty of constructing essentialist [religious nationalist] frames in a multi-religious and multiethnic context.” Nonetheless, the relationship between religious pluralism and state support for religion does not fully account for the effect of a country being in sub-Saharan Africa. Whether a country is in sub-Saharan Africa remains a highly significant predictor of lower levels of state support for religion regardless of controls (models 12, p = 0.003; model 13, p = 0.025). Although religious diversity decreases state support for religion, and countries in sub-Saharan Africa are more diverse than the rest of the world, this is insufficient to fully explain why countries in sub-Saharan Africa have much lower levels of state support for religion.

The final confounding factor, the moral authority of religious actors, requires a qualitative and historical explanation. While it is beyond the capacity of this paper to offer a thorough historical perspective, brief portraits of Rwanda and Mozambique offer three arguments. First, the moral authority of religious actors appears to be relevant to how the state engages with religion, especially in fractious contexts. Second, many of the factors identified by Grzymała-Busse (Reference Grzymała-Busse2015) shape moral authority, such as religious and ethnic homogeneity, religious actors' control over education, and religious actors' historical record in supporting the nation. Third, as shown by a quantitative analysis of the prevalence of these factors globally, these factors are significantly absent in sub-Saharan Africa. Countries in sub-Saharan Africa have significantly higher levels of ethnic heterogeneity, religious heterogeneity, and political instability, as well as significantly weaker education systems than the rest of the world. These factors are thus detrimental to religious actors forming and sustaining moral authority in sub-Saharan Africa. In addition, these historical portraits further illustrate how state capacity and majority religion are inadequate to explain the low state support for religion in sub-Saharan Africa.

Table 5 outlines the levels of popular religiosity and state support for religion in each country, along with the factors relevant to moral authority.

Table 5.

Popular religiosity, state support for religion, and Moral authority in Rwanda and Mozambique

Rwanda and Mozambique have very different demographics. Approximately 94% of the population are Christian in Rwanda (National Institute of Statistics of Rwanda, 2012). While 65% of Rwandans were Roman Catholic in 1995, this declined to 43.7% by 2012. Rwanda is also relatively ethnically homogenous, as 85% of the population were estimated to be Hutu in 1994 (McDoom, Reference McDoom2021, 265). Conversely, Mozambique is in the 78th percentile for ethnic diversity (Alesina et al., Reference Alesina, Devleeschauwer, Easterly, Kurlat and Wacziarg2003). Moreover, although most Mozambicans are Christian, there is a sizable Muslim minority and a significant proportion that follow “syncretic indigenous religious beliefs” (Office of International Religious Freedom, 2021, 2). Rwanda and Mozambique also differ in the role of religious organizations in providing education; in Rwanda, approximately 70% of primary schools are operated by churches (Scheunpflug and Wenz, Reference Scheunpflug and Wenz2021, 3), while almost all schools in Mozambique are state-run (UNESCO Institute for Statistics, 2020).

While the histories of Rwanda and Mozambique are clearly very different, there are nonetheless several comparable factors that are relevant for understanding why each country offers such little state support for religion.

The first of these are demographics and control over education, which provide initial conditions for state support for religion. In Mozambique, no religious group has much claim to national identity; the population was religiously and ethnically heterogeneous, and the Roman Catholic Church had very limited control over socialization through schools. In contrast, the Roman Catholic Church had a relatively strong association with national identity in Rwanda: most of the population belonged to the same religious denomination and ethnic group, and the Roman Catholic Church had significant control over socialization.

Second, religious actors’ choices shaped their moral authority as allies of the nation. In Mozambique, the Roman Catholic Church supported the colonial regime and was subsequently side-lined by the post-independence government led by Frente de Libertação de Moçambique (FRELIMO) in 1975 (Azevedo, Reference Azevedo and Anahnotu1992, 187; Rossouw and Macamo, Reference Rossouw and Macamo1993, 537). In Rwanda, the Church shifted its support from the colonial regime to the independence movement prior to independence in 1962 (Longman, Reference Longman2010, 66; Beloff, Reference Beloff, Rectenwald, Almeida and Levine2015, 209). The Church subsequently retained an extremely close relationship with the post-independence state, especially when led by Mouvement Révolutionaire National pour le Dévelopement (MRND) (Gatwa, Reference Gatwa1998, 135; Longman, Reference Longman2010, 89). However, the Church later participated in the state-sanctioned pogroms of Tutsis in which clergy would aid genocidaires in identifying and ambushing Tutsi families (Longman, Reference Longman2010, 194; Beloff, Reference Beloff, Rectenwald, Almeida and Levine2015, 213). In doing so, it lost its claim to moral authority as an ally of the nation after the MRND was removed in 1994 by the Rwandan Patriotic Front (RFP) (Beloff, Reference Beloff, Rectenwald, Almeida and Levine2015, 216). The new Rwandan government subsequently minimized political engagement with religious organizations.

The third factor involves states' efforts to depoliticize religion to defuse conflict. While religion was not the primary source of conflict in either country, religious tensions added a further dimension to several political fault lines. In Mozambique, FRELIMO identified as Marxist-Leninist in 1977 (Morier-Genoud, Reference Morier-Genoud2000, 422). The post-independence government was initially actively hostile toward religious groups, especially the Roman Catholic Church and Muslims (Rossouw and Macamo, Reference Rossouw and Macamo1993, 539; Morier-Genoud, Reference Morier-Genoud2007, 246). This attracted criticism internationally and bolstered domestic support for the anti-communist Resistência Nacional Moçambicana (RENAMO) in Mozambique's civil war (1977–1992). To defuse this element of the conflict, FRELIMO introduced several reforms from 1982 to 1989 to improve relations with religious organizations and reduce state involvement in the religious sector (Rossouw and Macamo, Reference Rossouw and Macamo1993, 539; Morier-Genoud, Reference Morier-Genoud2007, 246). Similarly, Beloff (Reference Beloff, Rectenwald, Almeida and Levine2015, 216) argues that the RFP has attempted to depoliticize religion in Rwanda as part of a project to instill a common national identity and aid reconciliation after the genocide.

These comparative assessments of Rwanda and Mozambique, although brief, illustrate the potential of Grzymala-Busse's account of moral authority (C2) to explain the anomaly of religious electorates and secular states in sub-Saharan Africa. In Rwanda, population religiosity, ethnic homogeneity, the prevalence of one religion, and the Roman Catholic Church's control over schooling and support for the independence movement initially facilitated a close relationship between the Church and the post-independence state. However, the Church's moral authority was forfeited after the genocide, and Rwanda currently has one of the lowest levels of state support for religion globally. Conversely, the ethnic and religious diversity within Mozambique, the paucity of schools controlled by religious institutions, and the Church's criticism of the liberation movement hindered state support for religion after independence. While the Mozambican state was initially hostile toward religion, the need for political stability eventually moved the state toward a neutral stance.

However, to what extent might these factors be applicable to other countries in sub-Saharan Africa? Table 6 compares the mean averages of sub-Saharan Africa countries and the rest of the world in population religiosity, ethnic and religious diversity, school enrolment, and prevalence of coup d’états (as an indicator of moments of political crisis). Although populations in sub-Saharan Africa are highly religious, these populations are also more ethnically and religiously diverse than the rest of the world (p < 0.001). Moreover, at key moments of state formation, such as national independence in the 1950s and early 1960s, religious actors could have access to only a minority of the population through education institutions. Therefore, countries in sub-Saharan Africa tend to have three factors that undermine the fusion of national and religious identities: high levels of ethnic and religious diversity, and religious actors having limited control reach through schools.

Table 6.

Comparison of means of factors conducive to religious actors' moral authority

Note: Bold indicates p < 0.05.

Finally, countries in sub-Saharan Africa have had to navigate more moments of national crisis than the rest of the world (p = 0.012), and thus religious actors have had more opportunities to choose the “wrong” side. Since it is easier to lose moral authority than it is to gain it, we would therefore predict that religious actors have lower levels of moral authority as defenders or unifiers of the nation in sub-Saharan Africa.

These factors are not direct determinants of state support for religion; indeed, in a linear regression using each of these factors, only religious homogeneity and population religiosity are significant predictors. Rather, these factors affect the likelihood of historically contingent decisions, such as FRELIMO's hostility toward religious groups or RFP's neutral approach. Moreover, moral authority can be lost if even one factor is detrimental; after its participation in the genocide, the Roman Catholic Church in Rwanda was unable to regain its moral authority despite otherwise favorable conditions. While further qualitative and historical research is needed to explore these contingencies in each context, this analysis suggests that a research agenda in this vein is likely to be productive. Moral authority may well be an important factor in understanding religion–state arrangements in sub-Saharan Africa. In addition, these portraits also illustrate the limits of state capacity and majority religion as explanatory factors. In both countries, the state's engagement with religion varied considerably even while Christianity remained dominant as the majority religion. Neither state limited support for religion because it lacked capacity; on the contrary, the RFP-led government has simultaneously decreased state support for religion and increased state capacity.

To test the robustness of the key finding of this study, the analysis substituted the proportion of respondents who attend religious services at least weekly as an alternative measure of population religiosity. The significance of the effect of both population religiosity and whether a country is in sub-Saharan Africa remains robust (p ≤ 0.001). The main argument of this paper thus remains valid, even if a measure of religiosity favored by many other scholars is used (such as Norris and Inglehart, Reference Norris and Inglehart2011; Buckley and Mantilla, Reference Buckley and Mantilla2013).

Conclusion

Addressing the anomalous limited state support for religion in sub-Saharan Africa offers several insights. The global datasets confirm that for the rest of the world, population religiosity is positively and significantly associated with state support for religion, but this pattern does not hold in sub-Saharan Africa, inviting us to offer more nuanced arguments to address such anomalies. The above analysis rules out several factors suggested by other scholars: state capacity, religious majority, quality of democracy, and religious diversity. On the contrary, given the average levels of state capacity and quality of democracy in countries in sub-Saharan Africa, we would predict that these countries have higher levels of state support for religion than the rest of the world—not less. That state capacity does not explain why states in sub-Saharan Africa provide so little support to religion may be the most surprising finding of this study. Yet perhaps this should not be unexpected. Although insufficient state capacity has been a common explanation in other studies for limited state support for religion (Fox, Reference Fox2010, Reference Fox2012; see Buckley and Mantilla, Reference Buckley and Mantilla2013; Cupery, Reference Cupery2013), there are several examples of states passing ambitious policies despite having little administrative capacity and finance to deliver these (see Melzer and Garbers, Reference Melzer and Garbers2019; Kim et al., Reference Kim, Robinson, Härmä, Jeffery, Rose and Woldehanna2022).

In contrast to the above factors, the moral authority of religious actors showed promise as an explanatory factor. While this moral authority is dependent on contingent political decisions, there are several factors that make decisions corrosive to moral authority more likely. These include frequent moments of national crisis, religious and ethnic diversity, and religious actors having limited influence over education. While the contrasting portraits of Rwanda and Mozambique demonstrate that none of these factors are overriding determinants, that these factors are common in sub-Saharan Africa may explain why state support for religion tends to be lower in the region.

The findings have three implications for research on the role of religion in politics in sub-Saharan Africa. First, it warns against the common simplistic inferences when interpreting the implications of religiosity and religious movements for politics in sub-Saharan Africa. Many scholarly assumptions about the relationship between religious populations and the state are shaped by the religious and political history of Europe. Seminal accounts of secularism are typically based on a historical narrative of religious-state arrangements in “the West” (see Taylor, Reference Taylor2007). However, it is an error to apply these assumptions to research on sub-Saharan Africa uncritically; as this paper demonstrates, sub-Saharan Africa is not merely Europe in the past (Engelke, Reference Engelke2015, 3). It is not enough to assume that religious beliefs simply translate into religious policy and thus anti-democratic and anti-technocratic religious movements reflect a credible threat to democracy and governance (see Ellis and Ter Haar, Reference Ellis and Ter Haar2005; Gifford, Reference Gifford2016). On the contrary, there may be an important distinction between the role of religion in political theatre and the state. As Jeffery-Schwikkard (Reference Jeffery-Schwikkard2022, 1077) and Obadare (Reference Obadare2018, 7) have observed in South Africa and Nigeria respectively, political actors may use religious rhetoric while concurrently remaining committed to a secular legislative framework. The findings of this study suggest that states, and likely voters, in sub-Saharan Africa mediate between popular religiosity and governance in complex and understudied ways. Without incorporating theories of how this works, our understanding of religion and democracy remains inadequate, both in sub-Saharan Africa and the rest of the world.

Second, theories that study secularism that pay more attention to populations fail to capture one of the most important sites of secularism: the state. Despite several scholars noting that Europe is more secular than the rest of the world, in an important sense, the opposite is true. European states have higher levels of state support for religion and legislated religious discrimination than sub-Saharan Africa, Asia, and Latin America (Fox, Reference Fox2011). Similarly, while it is well established that economic development is negatively associated with popular religiosity, the findings of this paper suggest that economic development is positively associated with state support for religion. While conventional theories of secularization focus on explaining why more “modern” societies are less religious, the premise of these arguments is wrong: more “modern” societies are in fact more religious, if judged by state support for religion. If these findings are robust, they identify the need to revisit both accounts of secularization and the assumptions that underpin analyses of religion and politics in sub-Saharan Africa.

Finally, comparative qualitative and historical research and in-depth case studies is critical to fully understand religion–state arrangements in sub-Saharan Africa. In addition to detailed analysis on religion–state arrangements in each country, further research on the historical legacies of colonialism and post-independence regimes (see Robinson, Reference Robinson2022) and the role of international pressure through institutions like the International Monetary Fund. Such research could further test the relevance of the factors explored in this analysis and shed further light on how and why states in sub-Saharan Africa's support to religion defy theoretical expectations.