Extensive Margin Dominance in Youth Crime Prevention
Criminal justice interventions can operate either by preventing individuals from committing any offenses (extensive margin) or by reducing the frequency of offenses among repeat offenders (intensive margin). Determining which margin drives observed program reductions is crucial for intervention design.
Picture this
Think of a stop sign placed at a busy intersection; its main job is stopping drivers who would otherwise drive right through without looking, rather than forcing drivers who already stop to wait longer at the line.
What the evidence says
SYEP participation decreases the probability of any arrest by 0.128 percentage points (a 17% reduction, p < 0.05) and decreases total arrest counts by 0.00122 arrests per youth (a 15% reduction, p < 0.10) on a control mean of 0.0087. The virtually identical relative percentage reductions (17% vs 15%) demonstrate that criminal deterrence operates almost entirely by preventing first-time arrest contact rather than reducing repeat arrests among already-arrested individuals.
- Who
- N = 163,447 total applicants (158,355 low-risk, 5,092 at-risk youth), New York City.
- How
- Two-Stage Least Squares (2SLS) instrumental variable estimation comparing binary arrest probability against discrete arrest counts during program operation.
What to do
1. Design youth employment program benchmarks around first-time arrest prevention targets rather than repeat-offender frequency reduction.
From the source
"This effect is nearly identical in magnitude to the 0.128 percentage point decrease in the number of youth who have any arrest, suggesting that the effect of SYEP on arrests during the program summer is coming nearly entirely on the extensive margin of keeping youth from being arrested at all."
The_Effects_of_Youth_Employment_Evidence_from_New_York_City_Summer.pdf