Magico-Religious Powers in Benin

In Southern Benin, man is considered to be intertwined in a web of invisible forces. This assumed interpenetration of the spiritual and material worlds profoundly influences people’s understanding of life’s afflictions. While minor inconveniences and transient misfortunes, such as mild illness, will typically be ascribed to the ordinary course of natural events, more profound, life-changing, or long-lasting misfortunes are often attributed to malevolent occult forces (Omonzejele Reference Omonzejele2008; Leistner Reference Leistner2014). This is especially the case if such events occur together: a run of serious accidents or financial misfortunes involving the loss of significant sums or assets, sudden unemployment, or the unanticipated death of a family member. Similarly, a serious illness befalling a person or animal can be thought to be the result of targeted attacks employing evil spirits, offensive charms, and witchcraft, rather than a simple biological malfunction (Geschiere Reference Geschiere2013; Maslove et al. Reference Maslove, Mnyusiwalla, Mills, McGowan, Attaran and Wilson2009). Adverse weather conditions such as drought or hail which affect crop yields are also likely to raise suspicion of occult activity. Additionally, when a person suspects malevolent occult forces are at play, typically blame does not fall on faraway strangers, but rather on those nearby: a neighbor, co-worker, or co-wife, or more intimately, a parent, sibling, or child (Leistner Reference Leistner2014; O’Neil Reference O’Neil2012).

This mindset is not just a phenomenon among the poorly educated in rural areas. People of all ages, genders, and backgrounds, in both rural and urban settings, may attribute such misfortunes to occult attack. Despite the fact that Benin is becoming increasingly culturally diverse, these beliefs seem to persist across regional and ethnic settings and transcend gender, generational identities, and social and economic status in both urban and rural contexts.

Investing economic and social capital in the management of perceived occult activity is thus a salient aspect of day-to-day life for many Beninese people (LeMay-Boucher, Noret, & Somville Reference LeMay-Boucher, Noret and Somville2013). People frequently turn to specialists who are deemed capable of interpreting and manipulating these invisible forces. They may seek specialist knowledge to determine whether the root cause of their misfortune is actually occult in nature, or because they seek magical protection and healing. LeMay-Boucher, Noret, and Somville (Reference LeMay-Boucher, Noret and Somville2013) show that in Cotonou, expenditures on magico-religious diagnosis, prevention, and treatment is far from anecdotal. In their sample, 48 percent of household heads declared having made such magico-religious expenditures in the twelve months prior to their 2006 survey, with these expenditures representing 5.6 percent of all household expenditures on average. As we show below, we find a similar prevalence from a different sample based on our 2014 survey.

Belief in occult forces does not preclude a parallel belief in science or Western medicine. Visits to the local health post and self-medication are the normal response to minor ailments. In fact, the sale of pharmaceutical drugs through both informal and formal pharmacies is a profitable venture in Southern Benin. The frequency of consultations with experts within the occult sphere is relatively small in comparison to the volume of anti-malaria and paracetamol tablets sold (Baxerres & Le Hesran Reference Baxerres and Le Hesran2011). Pharmaceutical and magico-religious treatments are also frequently used simultaneously to treat the same ailment.

Alternatively, as is the focus of this analysis, individuals may seek to spend on magico-religious forces with malicious intent. Out of the numerous accounts we collected during our survey, one illustrates this particularly well. One of our respondents explained how, following the death of his brother, he sought the services of a local diviner. The diviner confirmed that his brother’s death was the result of a malicious occult attack, and the man proceeded to carry out a counter-attack by requesting the malevolent use of a vodun deity.

The majority of the accounts in our sample follow a similar pattern, where individuals perform counter-attacks against perceived acts of occult aggression, or to a lesser extent, non-occult attacks. Many of these perceived occult attacks were carried out against members of the individuals’ families, most often their children, or sometimes their wives, their mothers, or their siblings. Counter-attacks against perceived personal attacks also appear to be common. In other words, the large majority of the attacks in our sample do not stem from a primarily malicious intent, but from a desire to retaliate against perceived attacks and to prevent future attacks. Most of these perceived attacks are associated with an unknown sorcerer, though in some cases individuals suspect the culprit behind the attack is a specific member of their family.

Only a small proportion of the occult attacks in our sample were not retaliatory in nature. For instance, one individual explained an occult attack carried out against colleagues who were perceived by this individual as selfish and lacking respect for elders. A different motivation sometimes mentioned was self-gain. Another respondent explained that, after applying for a job, he used a vodun to invalidate the offers of contracts to other candidates, and claimed this is why they did not secure the job.

Significantly, 50 percent of the individuals who invested money in the performance of occult attacks claimed the attacks were successful, while none (0 percent) claimed the attacks were unsuccessful. Many individuals claimed that those they performed the counter-attack against died or suffered a serious injury a few weeks later, or simply that their aggressors were scared away.

There are two main categories of magico-religious practices which are employed for both malevolent and benevolent means in Southern Benin. The first of these, Aze, roughly corresponds to the western understanding of witchcraft (see Evans-Pritchard Reference Evans-Pritchard1937), yet it differs in that the ability is not necessarily inherited (Falen Reference Falen2007). The alleged powers of Aze are said to primarily work through the invisible, psychic domain. In other words, an individual does not necessarily need to be in close proximity for Aze to have its intended effect. Such practices are traditionally said to be carried out in secret after nightfall (O’Neil Reference O’Neil2012).

Aze is considered to be a primarily a malevolent force. It is generally used with the intention of causing misfortunes such as illness, infertility, accident, death, loss of money, or insanity. Children are said to be especially vulnerable to Aze attacks (Falen Reference Falen2007). However, despite its primarily destructive role, it is a somewhat ambiguous force. It can allegedly be employed to diagnose and prevent malicious attacks by Bo or vodun deities or other Aze (O’Neil Reference O’Neil2012). Specifically, a type of “white” Aze known as “Aze wèwé” can allegedly be used to counteract the evils of “black” Aze, or “Aze wiwi.” However, the latter is said to be the stronger more dominant force in Southern Benin (Landry Reference Landry2015; O’Neil Reference O’Neil2012). Wielders of Aze are referred to as “Azeto”; however, as highlighted by LeMay-Boucher, Noret, and Somville (Reference LeMay-Boucher, Noret and Somville2013), these persons are rarely identified as such in a precise way in a social group or neighborhood. People generally do not publicly identify Azeto for fear of accusations that they too are Azeto. Anyone able to identify the members of covert groups engaged in such secretive activities is likely to be suspected of complicity.

The magico-religious practices of Bo, locally known as gris-gris, refer to the use of various magical spells, tangible charms, and the manipulation of spirits (Falen Reference Falen2016; LeMay-Boucher, Noret, & Somville Reference LeMay-Boucher, Noret and Somville2013; O’Neil Reference O’Neil2012). Like Aze, Bo can allegedly be wielded to produce both positive and negative effects, depending on the underlying motive for its use. However, unlike Aze, it is not considered to be a predominantly destructive force. One might seek a Bo as a means of protection against Aze and other Bo. Anecdotal examples of the uses of Bo range from one known as “flin,” which is said to aid memory and help individuals to excel in exams (O’Neil Reference O’Neil2012), to one intended to stop rain during an important ceremony or to summon rain to aid crops (Falen Reference Falen2007).

Nonetheless, offensive Bo spells can be used to bring misfortune such as illness or loss of fortune similar to magico-religious practices employing Aze. A crucial difference is that Bo, unlike Aze, is said to work primarily through the physical world, often requiring a person to see, ingest, or touch the Bo for it to work (Falen Reference Falen2016; O’Neil Reference O’Neil2012). Additionally, there is frequently confusion between the magico-religious practices of Bo and the use of traditional vodun deities. Vodun deities are neutral forces which can be requested to perform both good and evil deeds (Falen Reference Falen2016). As LeMay-Boucher, Noret, and Somville (Reference LeMay-Boucher, Noret and Somville2013) point out, Bo practices frequently involve vodun deities, though the use of vodun deities for malevolent and benevolent purposes exists independently of Bo.

There are various “experts” in the field who offer magico-religious services in Southern Benin. Vodun priests, Vodunon, and priestesses, Vodunsi, are one such option. Traditional diviners, in particular priests of the Fâ (a local divination system in Benin), and Bokonon, are also highly sought after (O’Neil Reference O’Neil2012). “Azobleto,” healers who use natural substances, reputedly imbued with special powers, are another. There are many others but, just as it is difficult to differentiate between Aze and Bo and magical manipulations involving vodun deities, the distinction between those wielding these powers and the nature of the services they offer also lacks clarity (Falen Reference Falen2016). Specifically, diviners, healers, and vodun priests are often thought to be involved in good as well as evil Aze and Bo, and O’Neil (2012) argues that while subtle distinctions can be made between such groups, their roles are generally rather fluid.

Generally, locals consider Aze to be a more negative force than Bo, though individuals may think of them differently and employ them in different ways, depending on their past experiences of relying on and investing in them. Individuals’ perceptions can also be significantly influenced by interventions by their local traditional diviners, in particular priests of the Fâ and Bokonon who provide diagnoses, cures, and the means to counter-attack.

Various work from anthropologists (Goody Reference Goody, Moore and Myerhoff1977; Humphrey & Laidlaw Reference Humphrey and Laidlaw1994) has outlined that social involvement in, or spending on, magico-religious rituals does not necessarily imply a personal or intimate belief in such phenomena. Indeed, participation in such rituals may just be a matter of social conformity. When individuals are hesitant about the correct interpretation of a given situation, it may also reflect a choice to account for every eventuality (Noret Reference Noret2007). As Pierre Bourdieu suggested, the logic of ritual action is “to put all chances on your side” (Bourdieu Reference Bourdieu1980:148). Consequently, the evidence presented below demonstrates the prevalence of investment in such ritual practices, not necessarily a belief in any particular aetiology of affliction.

Conjectures

Based on a combination of qualitative and quantitative literature as to which household characteristics are likely to influence household expenditures on malevolent magico-religious powers, this study formulates a number of conjectures. In a section further below (Empirical Estimation Strategy) it empirically investigates these conjectures on the basis of various estimations.

The Data

The data were collected during the summer of 2014 in the city of Abomey-Calavi, a mainly suburban area of Cotonou (Benin’s largest city), located around 20km from its center. Abomey-Calavi has a population of about 655,000 inhabitants (2013 census). Since an official list of households was not available in public records, households were selected using a pseudo-random selection technique following the Afrobarometer’s survey guidelines. Our sample consists of 126 households and covers the following area: the North of the campus of the University of Abomey-Calavi, to the East of Ouedo and to the South of the main street leading to the city hall of Abomey-Calavi (see the appendix for a map of the area covered). A grid of approximately twenty squares of similar size was applied in this territory, ten of which were then randomly selected. Within each square, every street was allocated a number and then three streets per square were randomly selected. These streets were visited prior to the beginning of the baseline survey. Streets hosting a lot of economic activities such as markets, shops, and public buildings, as opposed to housing, were excluded from the list. We ended up with twenty streets with at least one in each of the ten randomly selected squares. Our enumerator was instructed to enter and survey every tenth house on the left of each chosen street (making our selection process pseudo-random). On average, our enumerator surveyed six households per street. When a selected door was unoccupied, the next house was selected. Where the inhabitants were absent, or when neither the head of household nor the spouse was available, a second visit was scheduled. Our enumerator was local, independent, and qualified. He had previous experience with surveys and field work and undertook a two-day training session given by the authors. Special sessions were dedicated to translation in the local language (Fon) and to test the enumerator’s understanding. The enumerator was also accompanied by an experienced local supervisor during the early stages of the work. The sample size was arrested at 126 due to lack of time and funding constraints.

The survey included questions related to magico-religious expenditures for both beneficial and malevolent uses, as well as other household and individual characteristics. The questionnaire was answered by the heads of the households who typically manage magico-religious expenditures on behalf of the whole household. As personal investment in magico-religious powers, particularly for occult aggression, is a sensitive issue which individuals are often secretive about, questions on the topic were asked toward the end of the survey after participants had developed some trust, in an attempt to minimize under-reporting. None of the household heads we selected refused to take part in our survey or to answer a particular section of our questionnaire. It is also interesting to note that none of the respondents mentioned personally having any magical or divinatory skills and hence themselves refer to the experts we described above to buy these services.

Descriptive Statistics

The descriptive statistics for the 126 household heads in the sample for the variables used in our model estimations can be found in table 1. The questions from our questionnaire which were used to create these variables can be found in the appendix (table 4). Across the whole sample, the average age of household heads is around forty-two years, and only 18 percent are female. The average household size is 5.2 people and the majority of household heads (80 percent) live with a partner. We expect those living with a partner as well as those in larger households to be more likely to make expenditures on malevolent magico-religious powers, since these individuals are more likely to face greater opportunities for conflict to arise due to being part of a larger social sphere. Household heads have completed almost six years of education on average. Our income variable is based on the choice of income brackets we used and takes value one for each household head only for one of the five brackets ([CFA0–50000]; [CFA50–75000]; [CFA75–100000]; [CFA100–125000] and [CFA125–300000]) and zero otherwise.

Table 1.

Sample means for head of household characteristics

The largest proportion of household heads (56 percent) identify themselves as being Catholic, and 30 percent identify as Evangelical. Only two household heads identify themselves as being animist, while eighteen household heads describe themselves as being “previously animist,” having converted to other religions. We believe having been an animist previously may act as a good proxy for being an animist, as religious conversion does not necessitate the abandonment of beliefs associated with the previous religion. This is illustrated particularly well by Timothy R. Landry (Reference Landry2015), who describes a conversation with a woman living in Benin who converted from Vodún to Evangelical Christianity. She explains that she converted not because her underlying beliefs changed, but because she believes it to be an easier way to heaven. In our sample, the largest proportion of household heads, 48 percent, are of the Fon ethnicity. The second largest group, 8.7 percent, are Goun. We have smaller shares for other ethnic affiliations which are gathered into the benchmark category for our estimations.

As many as 94 percent of household heads reported having made a financial contribution to a funeral in the twelve months prior to the survey. A total of twenty-six (19.8 percent) household heads own a car. As discussed above, we believe ownership of a visible asset is likely to lead to redistributive pressures and, as highlighted by LeMay-Boucher, Noret, and Somville (Reference LeMay-Boucher, Noret and Somville2013), cars are a symbol of wealth in Benin and remain exclusive possessions. They function as a suitable proxy for redistributive pressures, as they cannot be easily concealed from the public. In our sample, car ownership increases with income. Of those who own a car, 0 percent fall into the lowest two income quintiles, 16 percent fall into the third income bracket, 32 percent fall into the fourth income quintile, and 52 percent fall into the highest income quintile. Ownership of a motorbike, or a color TV, was deemed unsuitable as an indicator of wealth as, out of 126 individuals, respectively 113 and 106 reported owning these. We also considered using a variable which indicates if an individual thinks other people perceive them as rich; however, 94 percent declared they feel this way, offering less variation for a proxy compared to ownership of a car.

Previous literature and our informal discussion during the pilot survey indicated that jealousy is likely to be an important variable driving the use of malevolent forces. While various proxies were elicited during our pilot phase, one appeared to be pertinent and widely understood: surveyed heads were presented with hypothetical scenarios in which (1) a family member, (2) a coworker, and (3) a neighbor receives a hypothetical money gift of CFA100000 (or USD200 at the time of survey, as USD1 was equivalent to around CFA500). Participants were then asked to declare whether, and how much, their willingness to pay (WTP) would be to make CFA25000 of this gift disappear. Of those surveyed, 20.6 percent gave a WTP larger than zero with a mean of CFA1944 to make CFA25000 of a neighbor’s CFA100000 money gift disappear. A positive WTP was given by 16 percent, with a mean of CFA3198 if the individual receiving the gift was a coworker. If the individual receiving the gift was a family member, the corresponding figures were 18.23 percent and a mean of CFA3817. We can infer that some individuals are willing to sacrifice a portion of their own income simply to minimize feeling economically inferior. We believe these variables represent a sufficiently appropriate proxy for whether individuals feel jealous of other people’s financial success.

Expenditures on Diagnosis, Cure, and Protection

A total of 44 percent of household heads reported expenditures on Aze and/or Bo for either diagnosis, cure, or protection within the previous twelve months. Given that such expenditures are more common than expenditures on malevolent uses, we did not ask about such expenditures preceding the twelve months leading up to our survey. A more detailed breakdown of the expenditures made on Aze and Bo for diagnosis, healing, or prevention of illness and other afflictions believed to be the result of occult aggression can be found in table 2. As highlighted, we do not have such detailed information for malevolent magico-religious powers. However, expenditures on Aze and Bo for diagnosis, cure, and protection offer some interesting insights, especially considering that these are the means by which many individuals cope with the occult attacks we are primarily concerned with. Magico-religious diagnosis typically involves consulting the relevant traditional religious expert, including vodun priests, priests of the Fâ, and visionaries from the Celestial Church of Christ (a local church which claims to have gifts of prophecy), who interpret the nature of the misfortune.

Table 2.

Expenditures on magico-religious powers for diagnosis, healing, and protection within the last 12 months

If an Aze or Bo attack is confirmed, various measures may be suggested as a means of healing. These include the use of potions, herbal teas, infusions, special soaps, scarifications, sacrifications and special prayers. “Kou Dio,” which involves taking a “night bath in a hole,” presents another option. If an individual fears an Aze or Bo attack, that person may instead seek preventive measures. The preventive services offered by local diviners and vodun priests often involve the use of objects imbued with special powers. These include charms, such as special soaps and powders, or amulets and rings which individuals may be required to wear. Magical charms can often be seen hanging over the doorways of Beninese households, and children are often seen wearing amulets to protect them against Aze (Falen Reference Falen2007). Other objects used as charms include candles, holy crosses, and prayer books. Additionally, special prayers may be performed. An equal proportion of people (42.9 percent) reported making expenditures on each of Aze and Bo. Interestingly, that descriptive statistic is comparable to the one that LeMay-Boucher, Noret, and Somville (Reference LeMay-Boucher, Noret and Somville2013) produced based on a sample of 178 households collected in Southern Benin in 2006, eight years preceding the sample on which this paper is based. We find significantly larger average expenditures for Aze and Bo (CFA9756 and CFA7015) than theirs.

Empirical Estimation Strategy

To investigate our conjectures, we estimate a linear probability model (LPM) using ordinary least squares (OLS). Results shown are for the LPM, but they are similar (in terms of significance) to the ones produced by probit or logit estimation techniques.^{Footnote 2} Robust standard errors are used to correct for heteroscedasticity, and controls for geographical location are included in all models.

Our dependent variable is equal to one if a household head has reported making an expenditure on malevolent magico-religious powers anytime between 2012 and 2014, or within the two years prior to the survey, and takes the value zero otherwise. Our dependent variable thus takes a value of zero for household heads who reported making expenditures prior to this period. Given the fact that such expenditures are uncommon and irregular, we are inclined to use a definition with a wide two-year window to be more inclusive. We can assume that most of the controls in our models remained constant or fairly constant over that period, though some variables, such as income, are more likely to have varied. We make the assumption that our income quintiles allow for enough variations in incomes within specific brackets that they may be considered a suitable proxy of the household head’s income over that period.^{Footnote 3} Nevertheless, due to this issue of timing, it is important to emphasize that we cannot infer causation from our analysis and interpret our coefficients as correlations.

On the whole, our results, (not shown but available upon request), hold if we define our dependent variable as equal to one if a household head has reported making an expenditure on malevolent magico-religious powers within the six months prior to the survey and zero otherwise. Such a definition does not solve the issue of timing raised above, and as such our coefficients are still interpreted as correlations in this instance. Due to the relatively small fraction of non-zero spending values, we are unable to estimate Tobit models on the amount spent in CFA for both periods used (two years and six months).

Our independent variables include: dummy variables for household head average monthly income quintiles; and Catholic, Evangelical, and “previously Animist,” which as discussed is a proxy for animist. To investigate if reports of malevolent magico-religious expenditures are uncorrelated with gender, level of education, and ethnicity, we include a dummy variable equal to one for female, dummies for the Fon and Goun ethnicities, and years of completed education as a measure of education. We include a dummy variable which is equal to one if a household head has made a financial contribution to a funeral in the previous twelve months as a proxy for social tension. Household size and a dummy for whether a household head lives with a partner are included as control variables, as these individuals are likely to face more conflicts due to being part of larger social spheres.

As a proxy for jealousy of the economic success of others, we include the household heads’ declared WTP (measured in CFA1000) a portion of their income to make CFA250000 of their neighbors’ hypothetical gift of CFA1000000 disappear. We include a dummy variable for if an individual owns a car as a proxy for redistributive pressures. The final variable of interest is a dummy variable which takes the value of one if a household head has reported making an expenditure on Aze and/or Bo for diagnosis, healing, and prevention in the last twelve months. This allows us to investigate if, as expected, there is a positive correlation between these expenditures and those on malevolent magico-religious powers. It is also worth highlighting that the simple correlation between the combined spending on Aze and Bo and malevolent magico-religious expenditures is 0.3.