Generalisations about African societies being pervasively corrupt are refuted in this paper. Among 25,397 Afrobarometer respondents in 18 countries, 26% report paying a bribe, while 74% do not. Five hypotheses offer explanations: institutional context, inequalities of socio-economic resources, social inclusion and exclusion, social and political capital, and conflicting norms. Multilevel statistical analysis identifies as most important: contextual differences in colonial legacies, ethnic politicization, service provision, press freedom, and having social or political capital. The analysis emphasizes studying behaviour rather than perceptions of corruption and supports a public-policy focus on bribery as an exchange for specific public services.
Key findings:
- Because the evidence shows that some Africans do not pay bribes while others do, this paper demonstrates the need to identify the multiple influences, both individual and national, that affect under which circumstances Africans pay bribes. In so large a sample, it is not surprising that the logit analysis attributes statistical significance to indicators of each of our five hypotheses. However, the influences are not equal in their impact.
- Whereas the direction of influence of social and political capital is consistent – bribery is increased – different measures of inequality and exclusion do not reinforce each other in the same way. Being poor increases the payment of bribes, but being uneducated has no independent effect. By being relatively excluded from contact with public officials, women and old people are significantly less likely to pay bribes, while urban residents are more likely to do so as a consequence of their social inclusion. It is possible to construct ideal-type examples of Africans most likely to pay a bribe (urban men who are young and in poverty) and Africans least likely to pay a bribe (women who are old, live in the countryside, and are not in poverty). Among those in the former group, 46% reported paying a bribe, while only 15% in the latter group did so.
- However, each of these groups is a very small proportion of Afrobarometer respondents. Only 4% had all four characteristics making them vulnerable to bribery, and only 3% were quadruply insulated. Most Africans are subject to a mixture of positive and negative influences affecting bribery, for example being low in income and a woman, or being prosperous but living in a country where bribery is relatively widespread, such as in Kenya. Even when a particular characteristic is statistically significant, the size of the group affected can limit its overall impact. For example, urban residents are significantly more likely to pay bribes for services, but because they are little more than one-third of Afrobarometer respondents, the result is that almost two-thirds of those paying bribes live in rural areas.
- While the proportion of variance accounted for by the logit analysis in Table 3 is acceptable statistically, it nonetheless leaves a lot open to further explanation. Adding an unambiguous question about whether individuals have contacted a public service in the past year would make it possible to discriminate between Afrobarometer respondents who do not pay bribes because they have no contact with a service and those who have contact and receive a service without paying a bribe. Analysis of Barometer surveys on other continents that do ask questions about contact consistently find it has a big impact on whether people are vulnerable to paying a bribe. The absence of contact in a given year may reflect life-cycle effects (for example, whether a household has children in school) or fluctuating needs (for example, for a permit or for health care). Therefore, extending the years covered by questions about bribe paying should increase the number of people reporting that they have paid a bribe in the past. However, the longer the time period covered, the greater the risk of bias in recalling this information.
Policy implications:
- For policy makers responsible for anti-corruption policies, this study has both negative and positive implications. Two features of institutional context are not subject to easy or immediate change. Ethnic fractionalization persists even when the boundaries of a state remain unchanged and can lead to political disruptions beyond the capacity of bureaucratic institutions to control. Traditional practices of neo-patrimonial governance reflected in measures of social and political capital co-exist with bureaucratic standards, and subjects can use these to exploit regimes that they see as exploiting them.