Homelessness and crime: Do your friends matter?

This paper investigates the influence of friends on crime, using data I collected among the homeless. To estimate the causal effects of friends and of the share of criminal friends on crime, I rely on two instruments. The first is the share of rainy days during one's first year as homeless: rainfall fosters homeless's concentration in sheltered places and increases the probability of interactions. The second is the share of inmates released during one's first year as homeless, which a¤ects the supply of criminal friends. I find that one additional friend decreases the probability of incarceration but criminal friends increases it.