Why people vote for corrupt politicians: Evidence from survey experiments in Afghanistan

Corruption presents one of the greatest barriers to economic growth and institutional development in many, particularly conflict-riven, countries world-over. Given corruption’s pernicious effects, it is perhaps puzzling that corrupt politicians are elected at all. The conventional wisdom, both in the popular and scholarly literature, is that voters unknowingly vote for corrupt politicians, and so if politicians are exposed as corrupt, voters will not vote for them. This assumption is fundamental to standard models of voting, and is also the basis of policy recommendations such as increasing transparency. Transparency is desirable, the literature argues, since if voters just knew how corrupt their politicians are, the thinking goes, they would vote them out.

We challenge the fundamental assumption that voters always penalize corruption at the ballot box. To do so, we turn to Afghanistan, one of most corrupt countries in the world. Given the country’s high corruption levels, and the critical security imperatives in the country, there is hardly a more natural case to increase our knowledge about the determinants of voting in a deeply corrupt environment. Using both secondary data on voting and corruption, and a series of survey experimental methods, we investigate the degree to which voters might penalize or reward corruption, and the circumstances under which these effects might hold. We start by employing newly collected secondary data on voting and corruption to help us understand whether politicians perceived to be corrupt suffered a corruption penalty or benefit in Afghanistan’s latest parliamentary elections. We then use survey experiments—both list experiments and vignettes where respondents are asked to choose between two hypothetical candidates—embedded in a large-n, nationally representative survey in Afghanistan to examine the effects of corruption perceptions at the voter level. We estimate the impact of allegations of corruption on hypothetical votes received, and proceed to examine if the effect of corruption on vote choice is influenced by co-ethnicity, by whether people expect to receive pork, by reminders of morality, or other factors we think might influence voting behaviour.

Our project provides both real-world and experimental estimates of the effects of candidate corruption on vote share, and evidence on the mechanisms by which the effect obtains. This study will deepen our understanding of how voters react to corruption, and when they might punish it. These findings could, in turn, be used to better inform theoretical models of voter behaviour. The policy implications of this understanding are likely to be consequential, too. If voters support corruption in some instances, for example, then increasing transparency might exacerbate, and not reduce, corruption. Lastly, this study will be among the first to use survey experimental methods in the developing world, and will therefore help increase our understanding of the cross-national applicability of these methods. IGC funding for this project will support the collection and coding of survey data on corruption amongst Afghanistan’s politicians.