Misaligned Incentives and the Scale of Incarceration in the United States

By: Branden DuPont

 A REVIEW OF AURÉLIE OUSS'S 2020 ARTICLE

Local Decisions, State Budgets

• Mass incarceration is a state and local problem
  
• 90% of the 2 million people in prisons and jails today
•  Local justice system actors determine who is in custody: prosecutors, judges, probation officers
•  Paid by the state; costs are not fully internalized

 MISALIGNED INCENTIVE

The most delicious of all

privileges — spending

other people's money.

-John Randolph of Roanoke

 MISALIGNED INCENTIVE

Vera Institute of Justice's Incarceration Trends

How Much Of US's Incarceration Can Be Attributed to These Misaligned Incentives ?

• Incarceration is largely subsidized by the state, leading to what Zimring and Hawkins (1992) have referred to as a “correctional free lunch.”

• Other domains health care,  unemployment show inefficiencies

•  Econ speak: "If demand for incarceration is price-elastic, the cost division of incarceration is expected to impact sentencing."

 MISALIGNED INCENTIVE

California Juvenile Justice Realignment

Creates A Natural Experiment

• Juvenile facilities are run at the county level by the California Youth Authority (CYA), but are funded by the State of California

• SB 618: 1996 Juvenile Realignment changed financing structure of juvenile justice.

• "the law discontinuously changed the price that counties would have to pay to incarcerate juveniles"

 Research Design

SB 618: Nuts and Bolts

• Goal:

  • reduce the over-reliance by counties for less serious juvenile offenders

  • to encourage counties to create local programs

  • Not meant to reduce incarceration

• Level 1-4

  • 5%  per capita cost for serious offenses (murder, robbery)

• Level 5-7 sliding scale

• 50% per capita cost moderate offenses (burglary, assault)
• 75% per capita cost second degree burglary and car theft
• 100% per capita cost for technical rule violations

 Research Design

 Hypothesis

Commitments to California Youth Authority facilities

 Results

Commitments to California Youth Authority facilities

 Results

CYA monthly averages for intakes

Commitments to California Youth Authority facilities

 Results

CYA monthly averages for intakes

Fit local-polynomial regression lines. 95% confidence

Commitments to California Youth Authority facilities

 Results

CYA monthly averages for intakes

Fit local-polynomial regression lines. 95% confidence

Discontinuity: clear, immediate break in mean

 XKCD

 Methods

Regression Discontinuity Design in Time (RDiT) is used to calculate the drop in juvenile incarceration

 Methods

The Effect by Nick C. Huntington-Klein
Assistant Professor of Economics, Seattle University

 Methods

The Effect by Nick C. Huntington-Klein
Assistant Professor of Economics, Seattle University

 RDiT skips this

 Results

Decrease in Juvenile Justice Admission

•  fit three RDiT models
  • linear
  • quadratic (polynomial^2)
  • rdrobust, Calonico 2014
• found change in cost resulted in 38– 63%  drop in youth incarceration
• probability of being incarcerated in a CYA facility instead of an adult facility for youth under the age of 25 upon entry  42% decrease

•  Used DiD to compare <18 in adult with a control group of young adults 19 years old. Found no drop in young adult incarceration.
• Avoided overall increasing adult trends because of 3 strike law's influence
• Other subsidized options
  • slight increase in adult admissions, but much smaller than use of
    
  • gradual trend

No Change or Substitution in Adult Incarceration

 Threat to Validity

•  Main change: cases being dismissed or diverted by probation officers
• Case types with the largest cost-burden shift, saw the largest decrease
  • Level 5-7: $1300 – $2600 per month for parole violations, misdemeanors, burglary
  • Level 1-4: $150 instead of $25 per month
• "price changes and offense severity are correlated, we cannot disentangle both explanations"

Orange County and Santa Clara

 Mechanism

Less prison, more crime?

 Spillover Effects?

 Spillover Effects?

 Study Limitations?

Let's Ask Econ Twitter

 Study Limitations?

Could Be Better Modeled As An Interrupted Time Series (ITS)

Regression Discontinuity in Time: Considerations for Empirical Applications

• "tests for sorting or bunching near the threshold are irrelevant"
• Long time period: unobserved changes & lead to smaller uncertainty estimates than if forecasting
• Model may not account for time series properties (e.g. auto-correlated and non-constant variance)
• may not have affected this paper, but could affect reproducibility
• ITS, especially to test crime/arrest over time

 Discussion

"It’s not drug crime that drives mass incarceration, it’s violence. And that’s a much harder problem to solve."

John Pfaff

Question: if this policy was directed at serious violent juvenile crimes or adult crimes, would it have the same impact?

 Discussion

Question: Probation officers are low-level bureaucrats with a rehabilitative focus.

Will this realignment affect an elected prosecutor in the same way?

 Discussion

Question: Are There Better Ways to Realign Incentives?

- Tell prosecutors about cost of incarceration

- Larry Krasner: Mandate Less Punitive Sentences

 Dinosaur

A South Florida Lawyer Tells All His Best Florida-Man Stories