Is vote buying effective? Evidence from a randomized experiment in West Africa

Vote buying is a frequent practice during election time in many parts of the world. But no research has been done to quantify its effects on voters' electoral behavior. To address this challenge, we have designed and conducted a randomized experiment during the presidential elections of July 2006 in Sao Tome and Principe. This is a newly found oil-rich West African country that has been facing an increase in 'retail' vote buying. Our research design included a randomized campaign against vote buying sponsored by the Electoral Commission of the country, and pre-electoral campaign/post-election panel surveys in treatment (exposed to the campaign) and control locations, including 1034 subjects across 50 different areas. We observe a significant effect of the campaign on perceptions of vote buying, which constitutes the exogenous variation we use to identify effects on voting behavior. We characterize determinants of vote buying (more frequent in swing and rural locations), and find that vote buying energizes the electorate by increasing turnout. Crucially, we capture real effects on candidates' relative performance, by identifying the challenger to be driving more votes through vote buying (after the treatment), which is consistent with the timeline of events (late challenger candidacy). This result controls for changes in information about the candidates (e.g. policy platforms) and location-specific minutes spent by international electoral observers