1. Aizer and Doyle (2015): Juvenile Incarceration and Long-Run Outcomes
Citation: Aizer, A. and Doyle, J. J. (2015). Juvenile incarceration, human capital, and future crime: Evidence from randomly assigned judges. Journal of Political Economy, 123(4), 1177-1219.
Research question: Does juvenile incarceration reduce human capital accumulation and increase adult criminal behaviour?
Identification strategy: Randomly assigned judges. Juvenile court judges in Cook County, Illinois, were quasi-randomly assigned to cases by the order in which cases arrived. Judges varied substantially in their incarceration propensity: some sentenced 10% of juveniles to incarceration while others sentenced 30%+ of comparable cases. This variation in judge stringency provides an instrument for incarceration.
The identifying assumption is that the assignment of judges to cases is random conditional on case characteristics and assignment timing—an assumption supported by balance tests showing that judge-assigned cases have similar observable characteristics across high- and low-stringency judges.
Key results: Juvenile incarceration significantly reduced high school graduation rates (by approximately 13 percentage points) and substantially increased the probability of adult incarceration (by approximately 23 percentage points). These are large effects: the causal impact of juvenile incarceration substantially worsens long-run outcomes, consistent with schooling disruption, peer network effects in detention facilities, and the stigma of a criminal record.
The LATE is identified for juveniles who would be incarcerated by a strict judge but not by a lenient judge—the marginal cases. Effects may differ for juveniles far from the margin (e.g., those who committed serious crimes and would be incarcerated by any judge).
Takeaway: For marginal juvenile offenders, incarceration is criminogenic rather than rehabilitative, primarily through disruption of human capital investment.
2. Chalfin and McCrary (2018): Are US Cities Underpoliced?
Citation: Chalfin, A. and McCrary, J. (2018). Are US cities underpoliced? Theory and evidence. Review of Economics and Statistics, 100(1), 167-186.
Research question: What is the causal effect of police officers on crime rates, and are city police forces optimally sized?
Identification strategy: The paper uses city-level panel data on police employment and crime rates from 1960-2010, addressing the simultaneity problem (cities hire more police in response to more crime) using a combination of lagged instruments and firefighter employment as a proxy for the part of police employment driven by city budget cycles rather than crime fluctuations.
The cost-benefit analysis uses the estimates to compare the social cost of an additional police officer against the social benefit (reduced victimisation costs). Police officer salaries are roughly $100,000 per year in cost terms; the social cost of a murder (using VSL estimates) is approximately $10 million.
Key results: Each additional police officer prevents approximately 0.3 violent crimes per year. Using social cost of crime estimates, the benefits of an additional officer substantially exceed the costs by a factor of 2-4. The paper concludes that most US cities are meaningfully underpoliced from a cost-benefit perspective. Subgroup analysis suggests the crime-reduction benefit is larger in cities with initially low police-to-population ratios, consistent with diminishing returns to policing.
Takeaway: The marginal social benefit of police employment substantially exceeds the marginal cost in most US cities, implying underinvestment in police forces relative to the social optimum.
3. Kling (2006): Incarceration Length, Employment, and Earnings
Citation: Kling, J. R. (2006). Incarceration length, employment, and earnings. American Economic Review, 96(3), 863-876.
Research question: Do longer prison sentences reduce post-release employment and earnings?
Identification strategy: Kling [2006] uses the random assignment of judges in Florida state courts as an instrument for sentence length. Defendants facing judges who impose longer sentences are quasi-randomly assigned (conditional on case type, county, and year), providing variation in incarceration length that is unrelated to defendant characteristics. The strategy is an early application of the judge-IV design that has since become standard in the criminal justice literature. The paper uses administrative data linking court records to earnings records from the Social Security Administration.
Key results: The IV estimates suggest that longer sentences have relatively small effects on post-release employment and earnings—the short-run scarring effects of longer incarceration are modest. This contrasts with Aizer and Doyle's results for juvenile incarceration, suggesting that effects may differ by age of incarceration, duration, or the type of facility.
Post-release earnings show some recovery toward pre-incarceration levels, consistent with human capital not being fully destroyed by incarceration for adult offenders. However, the estimates are noisy and the confidence intervals are wide.
Takeaway: For adult offenders, longer sentence lengths do not substantially reduce post-release labour market outcomes, in contrast to the large negative effects of juvenile incarceration documented by Aizer and Doyle.
4. Mueller-Smith (2015): The Criminal and Labor Market Impacts of Incarceration
Citation: Mueller-Smith, M. (2015). The criminal and labor market impacts of incarceration. Working Paper, University of Texas at Austin.
Research question: Does incarceration increase or decrease recidivism and labour market outcomes among adults?
Identification strategy: Mueller-Smith [2015] uses randomly assigned public defenders in Harris County (Houston), Texas, as an instrument for case outcomes including incarceration. Because public defenders vary in quality and availability, assignment to a lower-quality or more overloaded defender increases the probability of conviction and incarceration.
The design is complementary to judge-IV studies: both exploit variation in the quasi-random assignment of criminal justice actors to extract causal variation in incarceration.
Key results: Adult incarceration substantially reduces subsequent employment: each additional year of incarceration reduces quarterly employment by 3.6 percentage points and quarterly earnings by $1,800. On the recidivism side, incarceration increases future violent crime by 1.5 percentage points per year—consistent with criminogenic effects through peer exposure, criminal capital accumulation, and employment disruption.
These results suggest that incarceration is counterproductive for the marginal adult offender: it reduces lawful employment (reducing the opportunity cost of crime) while increasing criminal skill and networks, leading to more subsequent crime.
Takeaway: For adult marginal offenders, incarceration is criminogenic and economically harmful, providing a causal counterpart to the correlational evidence in criminology.
Methodological note on judge instruments
All four papers above rely on variants of the judge-instrument design. This strategy is powerful but has limitations [Andresen and Huber, 2019]:
(a) the LATE is identified for the "marginal" defendant—not for habitual offenders or for those far from the incarceration margin;
(b) exclusion restrictions can be violated if judge stringency is correlated with case management quality beyond sentencing;
(c) effects may vary across jurisdictions, making replication important.
The convergence of results across Aizer-Doyle (Illinois), Kling (Florida), and Mueller-Smith (Texas) using similar designs strengthens confidence in the criminogenic effects of incarceration for marginal defendants.
References
- Aizer, A. and Doyle, J. J. (2015). Juvenile incarceration, human capital, and future crime: Evidence from randomly assigned judges. Journal of Political Economy, 123(4), 1177-1219.
- Andresen, M. E. and Huber, M. (2019). Instrument-based estimation with binarized treatments: Issues and tests for the exclusion restriction. Econometrics Journal, 24(3), 536-558.
- Chalfin, A. and McCrary, J. (2018). Are US cities underpoliced? Theory and evidence. Review of Economics and Statistics, 100(1), 167-186.
- Kling, J. R. (2006). Incarceration length, employment, and earnings. American Economic Review, 96(3), 863-876.
- Mueller-Smith, M. (2015). The criminal and labor market impacts of incarceration. Working Paper, University of Texas at Austin.