Association for Public Policy Analysis & Management

With burgeoning prison populations across the United States, the issues inherent to managing prisons have become more pressing than ever. One important issue is preventing inmate violence that threatens the safety of both inmates and staff. Addressing this issue requires an in depth understanding of policy relevant characteristics of the prison setting that may be causally related to violence. This research examines the effect of prison crowding on violence.

The existing literature regarding the relationship between prison crowding and violence is surprisingly inconclusive. This research makes two distinct contributions to this literature. First, it offers a new set of statistically significant estimates showing a positive causal link between crowding and violence. These estimates exploit exogenous variation in California prison populations from a 2011 policy shock and utilize an instrumental variables identification strategy to account for variation in intensity of treatment over time and across prisons. The resulting estimates are a local average treatment effect for reception center and security level 2 populations. The estimates attribute a 5% reduction in violence to each 10 percentage point reduction in overcrowding.

The second contribution is a theoretical model that highlights the importance of compositional changes that necessarily accompany any shock to prison populations. These “compositional effects” could explain the failure of existing literature to provide consistent evidence of a relationship between crowding and violence. They also imply that the coefficients generated in the empirical section actually represent lower bounds on the true marginal impact of crowding on violence.