We estimate peer effects in paid paternity leave in Norway using a regression discontinuity design. Coworkers Economists and policymakers are keenly interested in understanding the effects of social interactions on individual behavior. One question of particular interest is how peer groups influence the take-up of government social programs. Peer groups could serve as important information transmission networks or be influential in changing social norms, particularly in settings where information is scarce and perceptions are in their formative stage. Social interactions could reinforce or offset the direct effects on take-up due to a program's parameters, leading to a long-run equilibrium take-up rate which is substantially lower or higher than otherwise expected.Estimating the causal effect of social interactions has proven difficult given the well-known problems of reflection, correlated unobservables, and endogenous group membership (Manski 1993). On top of these identification issues, it is often challenging to define the appropriate peer group and access data which links members of a peer group together. Early and ongoing research attempts to control for as many group characteristics as possible or use instrumental variables.1 Recognizing that estimates could still be biased, another set of papers attempts to measure peer effects by exploiting exogenous assignment to peer groups.

We study a change in maternity leave entitlements in Norway. Mothers giving birth before July 1, 1977, were eligible for 12 weeks of unpaid leave, while those giving birth after that date were entitled to 4 months of paid leave and 12 months of unpaid leave. The increased time spent with the child led to a 2 percentage point decline in high school dropout rates and a 5 percent increase in wages at age 30. These effects were larger for the children of mothers who, in the absence of the reform, would have taken very low levels of unpaid leave.

Understanding whether, and in what situations, time spent in prison is criminogenic or preventive has proven challenging due to data availability and correlated unobservables. This paper overcomes these challenges in the context of Norway's criminal justice system, offering new insights into how incarceration affects subsequent crime and employment. We construct a panel dataset containing the criminal behavior and labor market outcomes of the entire population, and exploit the random assignment of criminal cases to judges who differ systematically in their stringency in sentencing defendants to prison. Using judge stringency as an instrumental variable, we find that imprisonment discourages further criminal behavior, and that the reduction extends beyond incapacitation. Incarceration decreases the probability an individual will reoffend within 5 years by 27 percentage points, and reduces the number of offenses over this same period by 10 criminal charges. In comparison, OLS shows positive associations between incarceration and subsequent criminal behavior. This sharp contrast suggests the high rates of recidivism among ex-convicts is due to selection, and not a consequence of the experience of being in prison. Exploring factors that may explain the preventive effect of incarceration, we find the decline in crime is driven by individuals who were not working prior to incarceration. Among these individuals, imprisonment increases participation in programs directed at improving employability and reducing recidivism, and ultimately, raises employment and earnings while discouraging further criminal behavior. Contrary to the widely embraced 'nothing works' doctrine, these findings demonstrate that time spent in prison with a focus on rehabilitation can indeed be preventive.