In this paper, we analyse the link between victimization and offending using a representative sample from the Netherlands. We employ three different estimation approaches, starting from simple Ordinary Least Squares, and when possible – focus on different samples of the population, for which there is, arguably, lower selection bias. Secondly, we employ a two-step approach where the residual offending (victimization) is used as an explanatory variable for the current victimization (offending). Thirdly, we use a strategy proposed by Altonji et al. (2005), which uses the selection on observables as a guide for the selection on unobservable factors. Using either of the methods, we find evidence that past offenders have higher probability to be victimized. The link from victimization to offending is weaker. Stable personal characteristics, such as risk preferences and lifestyle choices are likely to explain a big part of this relationship.

Field: Applied Microeconomics