ECOS3997 Interdisciplinary Impact in Economics
Interdisciplinary Impact in Economics
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: THEend8_
ECOS3997
Interdisciplinary Impact in Economics
Stream 2: Economics of Crime
Lecture 7: Education, poverty and crime
Dr. Ranae Jabri
University of Sydney
Semester 1, 2024
1 / 49
Course Outline
1. Introduction and economic model of crime
2. Policing
3. Incarceration
4. Education, poverty and crime
5. Spillovers
6. Local criminal justice policy
7. New technologies used in criminal justice policy
2 / 49
Recap of last class:
• The deterrent effect of penalties on crime
• Drago et al. (2009)
• Effects of incarceration on post-release outcomes
• Judge leniency IV research design: Kling (2006); Dobbie et al. (2018)
3 / 49
Outline
3. Incarceration
A. Effects of incarceration on post-release outcomes
4. Education, poverty, and crime
A. Effects of education on crime
B. Heller (2014)
4 / 49
Outline
3. Incarceration
A. Effects of incarceration on post-release outcomes
4. Education, poverty, and crime
A. Effects of education on crime
B. Heller (2014)
5 / 49
How can incarceration affect behaviour post-release?
• Potential mechanisms:
• Effects on criminal capital, employment capital, or social capital
• Scarring/Stigma Effects of Incarceration
• Schools of crime or criminogenic influence
• Recidivism
• We talked about the judge leniency IV research design
• Examples of studies: Kling (2006); Dobbie et al. (2018)
6 / 49
The effect of post-conviction incarceration on recidivism (Loeffler and
Nagin, 2022)
• Most studies find little effect of post-conviction imprisonment on recidivism
• Mixed evidence: “a smaller number of studies do, however, find significant effects,
both positive and negative”
• Studies where recidivism declines: incarceration settings emphasizing rehabilitative
programming
• Studies where recidivism increases: incarceration settings do not emphasize
rehabilitative programming
7 / 49
The effect of pretrial incarceration on recidivism (Loeffler and Nagin,
2022)
• Most studies find a negative effect of pretrial detention on recidivism though some
studies do find a null effect
• Defendants can be detained pretrial in temporary manner (they may or may not be
released pretrial) → disruptive nature
• 500,000 detained pretrial per day in US
• 250,000 detained postconviction per day in US
• Pretrial jail stays are shorter with less certainty and provide less rehabilitative
programming and access to mental health and substance abuse resources
8 / 49
What are the effects of incarceration conditions on post-release
outcomes?
• Incarceration could have effects on ‘criminal capital’ (crime skills)
• Schools of crime: Prisons can have a criminogenic influence – incarceration can affect
the criminal propensity of individuals through social networks
• What influence do peers serving time in the same correctional facility have
on each other’s subsequent criminal behaviour?
• Required reading: Bayer, P., Hjalmarsson, R., and Pozen, D. (2009). Building
criminal capital behind bars: Peer effects in juvenile corrections. The Quarterly Journal
of Economics, 124(1):105–147.
9 / 49
What are the effects of incarceration conditions on post-release
outcomes?
• Incarceration could have effects on ‘criminal capital’ (crime skills)
• Schools of crime: Prisons can have a criminogenic influence – incarceration can affect
the criminal propensity of individuals through social networks
• What influence do peers serving time in the same correctional facility have
on each other’s subsequent criminal behaviour?
• Required reading: Bayer, P., Hjalmarsson, R., and Pozen, D. (2009). Building
criminal capital behind bars: Peer effects in juvenile corrections. The Quarterly Journal
of Economics, 124(1):105–147.
9 / 49
Bayer et al. (2009)
“Danbury wasn’t a prison. It was a crime school. I went in with a bachelor of marijuana
and came out with a doctorate of cocaine.” George Jung describing his introduction to the
cocaine industry in Blow
10 / 49
Bayer et al. (2009)
• Research question: Is the post-release criminal behaviour of a juvenile offender
affected by the characteristics of individuals with whom they are incarcerated?
• Empirical challenge: People do not randomly match with their peers
• Research design using facility and facility-by-prior offense fixed effects
• Findings: Evidence that peer effects exist juvenile correctional facilities
11 / 49
Bayer et al. (2009)
• Research question: Is the post-release criminal behaviour of a juvenile offender
affected by the characteristics of individuals with whom they are incarcerated?
• Empirical challenge: People do not randomly match with their peers
• Research design using facility and facility-by-prior offense fixed effects
• Findings: Evidence that peer effects exist juvenile correctional facilities
11 / 49
Bayer et al. (2009)
• Research question: Is the post-release criminal behaviour of a juvenile offender
affected by the characteristics of individuals with whom they are incarcerated?
• Empirical challenge: People do not randomly match with their peers
• Research design using facility and facility-by-prior offense fixed effects
• Findings: Evidence that peer effects exist juvenile correctional facilities
11 / 49
Bayer et al. (2009)
• Research question: Is the post-release criminal behaviour of a juvenile offender
affected by the characteristics of individuals with whom they are incarcerated?
• Empirical challenge: People do not randomly match with their peers
• Research design using facility and facility-by-prior offense fixed effects
• Findings: Evidence that peer effects exist juvenile correctional facilities
11 / 49
Motivation
• No formal institutions for criminal sector of the economy
• Social and peer networks could play an extensive role in the proliferation of criminal
activity
• These networks could develop through peer effects while offenders are incarcerated
• Potential cost of incarceration: peer effects in incarceration could increase the
proliferation of criminal networks
12 / 49
Context
• 169 juvenile correctional facilities in Florida, US
• Peer group changes over time as individual cycles out and peers cycle in and out
• Individual’s exposure to peers depends on individuals sentence overlap with peers
• Peer effects may depend on your own criminal experience and the criminal experience
of peers
• Need to allow for crime-specific peer effects that can vary with own criminal history
experience
13 / 49
Data
• Sample: 8000 inmates over 2 year period in Florida
• Individual-level data on demographic and criminal history, facility assignments
• Individual-level outcomes: arrest and adjudication outcomes for 1 year after release
• Peer data on demographic and criminal history
14 / 49
Research Design
• Empirical challenge: Juveniles are not randomly assigned to facilities
• Facility and facility-by-prior-offense fixed effects → using the within-facility over time
variation that juveniles overlap in the SAME facility
• Research design accounts for nonrandom assignment of juveniles to facilities “as long
as the date at which a given individual is assigned to a facility within the two-year
sample period is random with respect to the peers in the facility at the time”
15 / 49
Research Design
• Empirical challenge: Juveniles are not randomly assigned to facilities
• Facility and facility-by-prior-offense fixed effects → using the within-facility over time
variation that juveniles overlap in the SAME facility
• Research design accounts for nonrandom assignment of juveniles to facilities “as long
as the date at which a given individual is assigned to a facility within the two-year
sample period is random with respect to the peers in the facility at the time”
15 / 49
Empirical Strategy
Rhijt =β0(Offensehijt ∗ Peer offensehijt)
β1(No Offensehijt ∗ Peer offensehijt)
+ Pijtα+Xijtγ + λj +Offensehijt ∗ µj + ηt + εhijt
• Rhijt: Indicator variable for whether individual i in facility j released in period t
recidivates with offense type h
• Peer offensehijt: individual’s exposure to peers with a history of offense type h
• Offensehijt: Indicator variable for whether individual i has a history of offense type h
• No Offensehijt: Indicator variable for whether individual i has no history of offense
type h
16 / 49
Empirical Strategy
Rhijt =β0(Offensehijt ∗ Peer offensehijt)+
β1(No Offensehijt ∗ Peer offensehijt)
+ Pijtα+Xijtγ + λj +Offensehijt ∗ µj + ηt + εhijt
• Pijt: vector of peer characteristics
• Xijt: vector of individual demographic and criminal history
• λj : facility fixed effects
• Offensehijt ∗ µj : facility-by-prior-offense fixed effects
17 / 49
Empirical Strategy
• Crime-specific peer effects allowed to vary with an individual’s own criminal
experience
• β0: crime-specific peer effect for those with a history of having committed the relevant
offense
• β1: crime-specific peer effect for those without a history of having committed the
relevant offense
• Table 2: evidence that juveniles specialize
18 / 49
19 / 49
Tests of the central identifying assumption
• Is the date at which a given individual is assigned to “a facility within the two-year
sample period is random with respect to the peers in the facility at that time”?
• “Balance check”: test that within-facility variation in peer characteristics is orthogonal
to individual-level observables
• We saw a similar kind of check last week in Drago et al. (2009)
• Check that estimated peer effects do not differ after accounting for trends in criminal
activity
• Check that estimated peer effects can’t be explained by peers who already committed
crimes together being assigned together
20 / 49
21 / 49
Discussion
• Peer effects have a reinforcing nature: peers can increase crime-specific knowledge
• What implications do these findings have for how to group juveniles in correction
facilities?
• What implications do these findings have for policies that incarcerate more juveniles?
• Finding: “Exposure to peers with a history of committing a particular crime increases
the probability that an individual who has already committed the same type of crime
recidivates with that crime”
• Crime reduction can have dynamics effects through reducing criminal histories of
potential peers who can have crime-specific peer effects
22 / 49
Discussion
• Peer effects have a reinforcing nature: peers can increase crime-specific knowledge
• What implications do these findings have for how to group juveniles in correction
facilities?
• What implications do these findings have for policies that incarcerate more juveniles?
• Finding: “Exposure to peers with a history of committing a particular crime increases
the probability that an individual who has already committed the same type of crime
recidivates with that crime”
• Crime reduction can have dynamics effects through reducing criminal histories of
potential peers who can have crime-specific peer effects
22 / 49
Outline
3. Incarceration
A. Effects of incarceration on post-release outcomes
4. Education, poverty, and crime
A. Effects of education on crime
B. Heller (2014)
23 / 49
Outline
3. Incarceration
A. Effects of incarceration on post-release outcomes
4. Education, poverty, and crime
A. Effects of education on crime
B. Heller (2014)
24 / 49
How does more schooling affect the decision to commit crime?
The individual chooses to commit a crime if and only if:
Expected utility of crime > Expected utility of abstention
pU(Y − f) + (1− p)U(Y ) > Unc
• Potential offenders have more education and better labor market prospects
• Higher Unc → crime is expected to decline
25 / 49
How does more schooling affect the decision to commit crime?
The individual chooses to commit a crime if and only if:
Expected utility of crime > Expected utility of abstention
pU(Y − f) + (1− p)U(Y ) > Unc
• Potential offenders have more education and better labor market prospects
• Higher Unc → crime is expected to decline
25 / 49
How could education affect crime?
• Income effects?
• Time availability?
• Time preferences?
• Education → patience → value future earnings more
• Risk preferences?
• Education → risk aversion → give more weight to potential sanction
• Deterrence vs. incapacitation?
26 / 49
How could education affect crime?
• Income effects?
• Time availability?
• Time preferences?
• Education → patience → value future earnings more
• Risk preferences?
• Education → risk aversion → give more weight to potential sanction
• Deterrence vs. incapacitation?
26 / 49
How could education affect crime?
• Income effects?
• Time availability?
• Time preferences?
• Education → patience → value future earnings more
• Risk preferences?
• Education → risk aversion → give more weight to potential sanction
• Deterrence vs. incapacitation?
26 / 49
How could education affect crime?
• Income effects?
• Time availability?
• Time preferences?
• Education → patience → value future earnings more
• Risk preferences?
• Education → risk aversion → give more weight to potential sanction
• Deterrence vs. incapacitation?
26 / 49
How could education affect crime?
• Income effects?
• Time availability?
• Time preferences?
• Education → patience → value future earnings more
• Risk preferences?
• Education → risk aversion → give more weight to potential sanction
• Deterrence vs. incapacitation?
26 / 49
How could education affect crime?
• Income effects?
• Time availability?
• Time preferences?
• Education → patience → value future earnings more
• Risk preferences?
• Education → risk aversion → give more weight to potential sanction
• Deterrence vs. incapacitation?
26 / 49
How can we estimate the effect of increased schooling on crime?
• Endogeneity issue: unobserved characteristics that determine schooling decisions
could be correlated with the unobserved determinants of criminal behavior
• Example: people with high return from crime drop out from school and commit crime
→ education decreases crime?
• Research design: exogenous or quasi-exogenous variation in schooling
27 / 49
Lochner and Moretti (2004)
• Research question: What is the effect of education on crime?
• Natural experiment: in half of U.S. states, legal to drop out of school at age 16, but in
1960-80 some states increased their minimum age to 17 or 18.
• Findings: Birth cohorts that were forced to stay in school longer had lower arrest and
incarcerations rates as adults than their predecessors.
• 1 extra year of high school reduced arrest rates by 11%
28 / 49
Machin et al. (2011)
• Research question: What is the effect of education on crime?
• Natural experiment: Compulsory School Leaving Age reform affected 15 year old
youth in England and Wales in the early 1970s
• Findings: Birth cohorts that were affected by the laws had a large drop in property
crime; results are not clear for violent crime.
29 / 49
30 / 49
31 / 49
32 / 49
Outline
3. Incarceration
A. Effects of incarceration on post-release outcomes
4. Education, poverty, and crime
A. Effects of education on crime
B. Heller (2014)
33 / 49
Heller (2014)
• Research question: What is the effect of summer jobs on crime for disadvantaged
youth?
• Randomized controlled trial among 1634 disadvantaged youth in Chicago
• Findings: Summer jobs reduce violence by 63% over 16 months
34 / 49
What is the motivation for this paper?
• Violent crime public health crisis: 150 people killed through acts of violence every day
in the US
• Youth twice more likely than adults to be victims/perpetrators of violence
• Disadvantaged minority youth: violent crime arrests rates for Black youth is 5 times
as high as for white youth
35 / 49
What is the motivation for this paper?
• Violent crime public health crisis: 150 people killed through acts of violence every day
in the US
• Youth twice more likely than adults to be victims/perpetrators of violence
• Disadvantaged minority youth: violent crime arrests rates for Black youth is 5 times
as high as for white youth
35 / 49
Large-scale RCT
• Treatment group: One Summer Plus (OSP) public summer jobs program in Chicago
• Treatment group 1:8 weeks of part time summer mininum wage employment (25
hours)
• Non-profit and government sectors; job mentors
• Treatment group 2: Social-emotional learning (SEL) on CBT principles to help
manage thoughts and emotions, and affect behaviour, to help with job (15 hours + 10
hours of SEL)
• Control group: Excluded from the program
• Aim to prevent crime rather than manage after crime onset
36 / 49
Data
• Sample: 1634 8-12th grade from 13 high-violence CPD schools
• Admin data tracking program participation
• Student-level Chicago Public Schools data on academic outcomes
• Neighborhood-level data from American Community Survey
37 / 49
How can summer jobs affect crime?
• Income effects?
• Incapacitation effects?
38 / 49
How can summer jobs affect crime?
• Income effects?
• Incapacitation effects?
38 / 49
Mean preprogram characteristics for treatment and control groups
39 / 49
Balance check?
Where have we seen this kind of balance check before?
• We saw this kind of balance check in Drago et al. (2009)
• Why does the author do a balance check here?