Canonical models of crime emphasize economic incentives.Yet, causal evidence of sorting into criminal occupations in response to individual-level variation in incentives is limited.We link administrative socioeconomic microdata with the universe of arrests in Medellín over a decade.We exploit exogenous variation in formal-sector employment around a socioeconomic-score cutoff, below which individuals receive benefits if not formally employed, to test whether a higher cost to formal-sector employment induces crime.Regression discontinuity estimates show this policy generated reductions in formal-sector employment and a corresponding spike in organized crime, but no effects on crimes of impulse or opportunity.