Résumé

This paper documents the effect of peers’ incarceration on an individual’s criminal activity within small criminal groups. Using established criminal groups, I built a 48-month panel that records the criminal status, Individual imprisonment status and imprisonment status of group members. Panel regressions with individual fixed effects allows me to document five facts. First, the incarceration of a peer is associated with a 5 per cent decrease in the arrest rate among groups composed of two persons. No effect is observed among bigger groups. Second, this effect is present even for incarceration following lone crimes, ruling out an explanation based on common shocks. Third, the probability of committing a group crime strongly decreases, and there is no shift to crime with other peers or lone crimes. Four, this general effect hides significant within-group heterogeneity. The results are consistent with the idea that ‘leaders’ are not affected by the incarceration of ‘followers’. Five, the effect seems to be driven by lower risky behaviour among offenders who remain free, and not by ‘criminal capital’ loss or deterrence.