Can the poor fight back? A multilevel analysis of corruption, poverty and collective demand for accountability in Africa

ABSTRACT

This paper examines the effects of the perceptions of corruption and personal experience of bribery on the propensity for ordinary Africans to support collective action-based anti-corruption tactics. It also evaluates how poverty shapes the association between corruption and support for collective civic action against corruption. The paper bases its findings on the multilevel level regression analysis of public opinion data from 35 African countries. The results show that an increase in experience of paying bribes increases poor people’s preference for anti-corruption tactics based on collective action. Furthermore, individuals who perceive corruption to be widespread are significantly more likely to support collective action as their country’s poverty level rises. These findings strongly challenge the view recently expressed in some of the literature that high levels of corruption erode the willingness of especially poor Africans to bring corruption under control.

KEYWORDS:

• Poverty
• collective action
• bribery
• corruption
• multilevel analysis
• Africa

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Moletsane Monyake is a doctoral researcher at the Sussex Centre for the Study of Corruption, Department of Politics, University of Sussex. He also teaches political science at the Department of Political and Administrative Studies, National University of Lesotho.

Notes

1 A long line of research shows that citizens of highly corrupt countries often report very low levels of trust in most people (see Rothstein 2011; Uslaner 2008).

2 My emphasis.

3 Afrobarometer is an independent research network that runs nationally representative surveys covering various social, economic and political forces across the African continent. Afrobarometer data sets are freely available at http://www.afrobarometer.org/.

4 A maximum likelihood factor analysis of these five ‘poverty’ items extracted one solution with an alpha value of 0.79, suggesting that these items can form a reliable index of LPI.

5 Originally, the score ranges from zero to four. For ease of interpretation, I have reversed the scores so that zero represents more clientelism and −4 represents low clientelism.