On the dynamic decision to participate in crime

by Jennifer Williams

Institution: Rice University
Year: 1997
Keywords: Economics
Record ID: 1686402
Full text PDF: http://hdl.handle.net/1911/19229


Our research examines the decision to participate in crime using a dynamic model of individual choice under uncertainty. The motivation for studying this decision in a dynamic framework is twofold. First, it allows us to formulate a theory of rational criminal choice, where agents anticipate the future consequences of their decisions. Second, it permits explanation of the temporal pattern displayed in aggregate arrest data. Across different countries, cities, and time periods, the aggregate arrest rate is a unimodal and positively skewed function of age. The standard static approach to crime offers no insight into the cause of this empirical regularity. We study criminality in a dynamic context by introducing social capital into the economic theory of crime. Social capital measures the extent to which an individual is bonded to legitimate society. The social control theory of crime posits that bonds to society strengthen as the individual ages, increasing the cost of deviant behavior, making criminal acts less likely. This hypothesis is consistent with the temporal pattern displayed in aggregate arrest data. In our formulation, preferences and legitimate income depend on the individual's stock of social capital. Rationality is imposed by requiring agents to take these effects into account. We empirically implement our model using panel data on a sample representative of young men in urban areas of the United States. Estimation is complicated by an omitted regressor problem, which arises because there are two possible future states – apprehension and escaping apprehension. Only one state is realized for each individual and subsequently observed by the econometrician. However, the unobserved choices in the state not realized enter the Euler equations. We resolve this problem by replacing the unobservables with Monte Carlo draws from the conditional empirical distribution of observed outcomes and using a Simulated Method of Moments estimator. Our results provide evidence in support of a social capital theory of crime. We find that social capital affects both preferences and earnings in the legitimate sector. Further, as predicted by social control theory, social capital becomes increasingly important over the life-cycle. This raises the cost associated with crime, making its occurrence less likely.