Working Papers

This paper studies the effects of the court-ordered removal of children from home on health and crime. To isolate causal effects, I exploit quasi-random variation in judge assignment together with across-judge variation in the tendency to favor removal in an instrumental variable (IV) design. Using a novel data set (N=26,481) based on Swedish court documents that I transcribe and link with detailed register data, I find that court-ordered out-of-home placement has large adverse effects on the mortality of the marginal child. These effects are primarily driven by suicides that occur while the removed child is still placed in out-of-home care. Removal also causes an increase in hospitalizations for mental illness and non-narcotic crimes. For birth parents, I again find an increase in non-narcotic crimes but there is little evidence of adverse health effects. I explore potential explanations for the detrimental effects on child health. Peer victimization, peer-to-peer spillovers, and adverse care home conditions appear to be important channels. 

Link: Job Market Paper 

Individuals struggling with substance abuse and self-harm are often treated in group-based programs. However, concerns have been raised about the risk of adverse outcomes through peer-to-peer spillovers. This paper analyses the effects of peers placed in residential treatment facilities on each other’s outcomes using novel data on the universe of youths (over 16,000) admitted to state-owned treatment facilities in Sweden between 2000 and 2020. To overcome the issue of nonrandom assignment of youths to facilities, I use the natural flow of youths to and from facilities within a given year by including facility-by-year fixed effects. I find strong evidence of reinforcing peer effects in substance abuse and self-harm. Exposing youths with a history of substance abuse (self-harm) to peers with a similar background increases the risk of experiencing adverse events related to substance abuse (self-harm) post-discharge. These effects are robust to the inclusion of controls for other peer characteristics (crime, mental disorders, gender, nationality, and age).

Link: Peer Effects Paper