The Dynamics of Domestic Violence: Does Arrest Matter?

In this paper, we estimate a stochastic-dynamic model for domestic violence using data collected by the Minneapolis Domestic Violence Experiment. Our primary finding is that arrest deters domestic violence, but the effect wears off quite quickly. We find also that current employment for the male is associated with lower levels of violence. Like arrest, the effect of employment is transitory. If the male becomes unemployed, the level of violence will increase quite rapidly. Violence in one period is associated with higher probabilities of violence in subsequent periods. From a methodological perspective, our results suggest that policy evaluation and deterrence research would benefit from using models that allow examination of the dynamic path of intervention effects. The effect of private and social programs need not be constant over time, and applying traditional, static models that necessarily impose such an assumption may produce misleading results. For Minneapolis, static models produced the result `arrest works.' The dynamic model suggests a different conclusion `arrest buys us a little time.'