Measuring the effect of probation and parole officers on labor market outcomes and recidivism
Objectives Use a unique dataset to pair probation and parole officers and their clients in Denmark in 2002–2009 to identify causal effects of these officers on labor market outcomes and recidivism. Methods To identify these effects, we rely on data from all probationers and parolees in Copenhagen, w...
Journal of quantitative criminology
Year: 2015, Volume: 31, Issue: 4, Pages: 629-652
Presumably Free Access |
|Journals Online & Print:|
|Check availability:||HBZ Gateway|
|Summary:||Objectives Use a unique dataset to pair probation and parole officers and their clients in Denmark in 2002–2009 to identify causal effects of these officers on labor market outcomes and recidivism. Methods To identify these effects, we rely on data from all probationers and parolees in Copenhagen, where a rotational assignment process randomizes clients to officers. We apply OLS models to test whether the inclusion of probation and parole officer fixed effects improves model fit, and we show the impact of officer fixed effects by generating predicted values for one individual, varying only the officer. Results The first stage of the analysis shows that the assignment of a probation or parole officer is indeed random in Copenhagen—at least in regards to the vast majority of background characteristics—suggesting that we are able to identify causal effects of probationer and parolee assignment on labor market outcomes and recidivism. The second stage of our analysis shows that although to a lower degree than common sense might suggest, probation and parole officers do matter for their clients’ dependency on public benefit transfers (around 10 percentage points) and criminal recidivism (around 30 percentage points), whereas earnings are unaffected. Conclusion As no study has yet to identify causal effects of probation and parole officer assignment, this study makes a novel contribution to the literature on the effects of criminal justice contact on subsequent life-course outcomes. Although generalizability to the US context is uncertain because we rely on Danish data, our findings nonetheless point in interesting directions for future research.|