Description
Commodity booms can generate economic gains and social costs at once, particularly where institutions are weak. This paper provides evidence that gold-price shocks raise violence and agrarian conflict in the Brazilian Amazon, and that they do so by expanding gold extraction itself. We build a shift-share (Bartik) design - interpretable as a continuous difference-in-differences - that combines pre-sample municipal exposure to mining with international gold prices and exchange-rate variation, together with satellite-based measures of mining activity. Exogenous increases in gold profitability expand artisanal (garimpo) and overall gold mining, and this expansion raises homicide rates, the number of agrarian conflicts, and the number of peasant families in conflict, with the largest and most precisely estimated homicide effects borne by Black and mixed-race populations. The effects are concentrated in municipalities with exploitable gold endowments and differ across extraction regimes: garimpo mining maps onto lethal violence, while broader gold-mining activity maps onto land conflict. We find no comparable effects for industrial extraction, and evidence that official homicide counts understate the true mortality burden of mining expansion. The findings challenge the view that formalization and property rights alone can mitigate the social costs of extractive booms, and point to entrenched criminal landscapes that benefit from state inaction.