Canonical models of criminal behavior highlight the importance of economic incentives and employment opportunities in determining crime (Becker, 1968). Yet, there is little causal evidence leveraging individual-level variation in support of these claims. Over a decade, we link administrative micro-data on socio-economic measures with the universe of criminal arrests in Medellin. We test whether increasing the relative costs to formal-sector employment led to more crime. We exploit plausibly exogenous variation in employment around a cutoff in the socio-economic score, below which individuals receive health care if they are not formally employed. Using a regression discontinuity design, we show that the policy had the unintended consequence of inducing a fall in formal-sector employment and a corresponding spike in organized criminal activity. There are no effects on less economically motivated crimes like those of impulse or opportunity. Our results confirm the relationship between formal employment and crime, validating models of criminal activity as a rational economic choice.

ACCESS PAPER