Peer effects and Juvenile Delinquency
“Walk with the wise and you become wise, but the companion of fools fares badly.” Proverbs 13:20
Economists found many instances in which important life decisions are influenced by one’s friends, colleagues, and neighbors. Examples include retirement savings decision, stock market participation, energy conservation, and welfare participation, to name a few.
Surely, many of the decisions we make as adults are influenced by our peers. But it is likely that that peer groups play an even more important role in shaping many important decisions made by children and adolescents. Are youths more likely to commit delinquency and juvenile crime when surrounded by other delinquents? There are several reasons why this may be the case.
First, neurobiology research suggests that brain structure responsible for cognitive control and self-regulation do not fully develop until young adulthood, making youth particularly susceptible to problematic behaviors and harmful peer influence. Second, the extent of peer effects among youths may be stronger than peer effects among adults because many youths spend more time with their classmates and neighborhood friends than adults do with their coworkers and neighbors. Third, many juvenile offenders commit their crimes together. Economists Stephen Billings, David Deming, and Stephen Ross show that crimes committed by multiple perpetrators accounted for 22 percent of all crimes committed by 16-to-21-year-olds in North Carolina.
But as always, it is challenging to quantify the causal effect of one’s delinquent peers on his own delinquency. The main reason is that there may be important differences between youths who belong to different peer groups. Those who live in different neighborhoods often differ in their demographic and socioeconomic background. In schools with academic tracking, students’ classroom assignments are determined by their academic achievement. How can we disentangle the effect of one’s peer groups on his delinquency from the effects of these other individual characteristics?
Peer Effects in Juvenile Correctional Facilities
As usual, economists came up with clever identification strategies to recover the causal effect of peers on juvenile delinquency. Consider peer effects among juvenile offenders in a correctional facility. Although correctional facilities are intended to “straighten up” juvenile offenders, frequent interactions with other offenders may in fact aggravate their criminal risks. They may teach each other their criminal skills and introduce to their criminal network. This may be particularly relevant among offenders with similar criminal background. For example, drug offenders may learn from each other about other drug crime opportunities and networks.
The problem is that the assignment of juvenile offenders to different correctional facilities depends on the severity of crimes they committed. Those who commit relatively minor crimes (such as petty larceny) are more likely to be assigned to facilities with other petty thieves, and those who commit more serious crimes are more likely to be assigned to facilities with other serious criminals.
To remedy this problem, economists Patrick Bayer, Randi Hjalmarsson, and David Pozen looked at the within-facility variation over time in the share of offenders with similar criminal histories. Their intuition was that the shares of petty thieves in a low-level correctional facility was always going to be larger than in a high-level facility, but the variation in the share of petty thieves within each correctional facility over time is going to more random. In other words, the within-facility variations over time in the shares of thieves, burglars, drug dealers and sex offenders are likely to be random, and this variation can help researchers to identify the effect of peer groups on future criminal activities. Bayer, Hjalmarsson, and Pozen found that more exposure to other juvenile offenders with similar criminal background significantly increased the likelihood of recidivism.
School Choice and Crime
It is difficult to overstate the importance of school choice on students’ life outcomes. Surely students’ learning experiences are greatly influenced by their teachers and classmates. But perhaps more importantly, other behavioral outcomes such as whether to participate in underage drinking, illicit drug use and other delinquent behaviors are also influenced by school peers. Not surprisingly, many parents put considerable efforts on sending their children to good schools by moving to neighborhoods with better public schools or sending their children to expensive private schools.
But does going to a better school really improve a student’s outcome? How can we measure this using data? As before, our main problem is self-selection. Students who go to different schools tend to differ in their demographic and socioeconomic background, and it is not straightforward how to separate the effects of school choice and individual characteristics on student outcomes.
David Deming tackled this problem by exploiting a unique school lottery policy in Charlotte, North Carolina. Since 2002, the Charlotte school district implemented a district-wide open enrollment system in which students could apply to any school within the district. In over-subscribed schools, enrollment was determined by a randomized lottery. Then, the difference between criminal outcomes of lottery winners and losers can be viewed as the causal effect of winning the lottery and attending their school of choice on their subsequent criminal behavior. Deming finds that winning the lottery and entering the preferred high school reduces the probability of future arrest by 50 percent.
Meanwhile, he finds little difference in academic achievement between lottery winners and losers. Maybe it is because emulating peers’ delinquent behavior is much easier than emulating their academic success. Just how hard is it to drink a can of beer with your (underage) classmates, compared to having to score as high as them on a math exam?
But what is a “peer group”?
Like the two studies described above, many other studies show that peers greatly influence many important youth outcomes. However, it is possible that these findings actually understate the true magnitude of peer effects, as their empirical analysis is often based on unrealistically large “peer groups”. For example, Bayer, Hjalmarsson, and Pozen’s study considers all juvenile offenders housed together with similar criminal background as the relevant peer group. Deming’s study considers everyone in one’s preferred or non-preferred high school as his peers. But were you close friends with everyone in your classroom, cohort, and school? Did all of them significantly influence your adolescent decisions? It is more reasonable to assume that peer effect is more concentrated within a smaller group of close friends who actually interact with and influence each other.
Unfortunately, high-quality data on actual friendship is very rare, and even if the data were available, defining what a proper peer group is won’t be easy. However, existing research suggests that the extent of peer effect is usually stronger among students of same gender and racial background. Similarly, in schools where classroom assignment is based on students’ academic achievement, we would expect that peer effects will be stronger among students with similar academic levels. More research on what defines a relevant peer group would greatly improve our understanding of juvenile peer effects.
- Bayer, Patrick, Randi Hjalmarsson, and David Pozen. “Building Criminal Capital behind Bars: Peer Effects in Juvenile Corrections.” Quarterly journal of economics 124.1 (2009): 105-147.
- Billings, Stephen B., David J. Deming, and Stephen L. Ross. Partners in Crime: Schools, Neighborhoods and the Formation of Criminal Networks. No. w21962. National Bureau of Economic Research, 2016.
- Deming, David J. “Better Schools, Less Crime?” Quarterly Journal of Economics 126.4 (2011): 2063-2115.
© Songman Kang, Hanyang University