Event Date
"Digging for Votes: Machine-learning detection of illegal mining and its effects on political outcomes"
Abstract:
Controlling illegal activities is a challenge for local authorities. Within communities, there exists a dichotomy wherein certain segments of the population derive economic benefits from such activities, while others bear the brunt of negative externalities. In an effort to address this issue, government authorities in Colombia received information about the locations of illegal mines as part of a Randomized Control Trial. This study explores the electoral repercussions of this intervention and finds an increase in the concentration of votes in the municipalities that received the treatment. We argue this concentration is the result of a strategic response of voters and consistent with a theory that predicts that proximity to the disclosed location of illegal activities is directly proportional to the magnitude of the response.