Working Papers
Axelrod, Matias. "In the 'Right' Place at the Right Time: Prison Labor Compensation and Recidivism in Arizona" [Job Market Paper]
"Despite their ubiquity in prisons across the United States, relatively little is known about forced prison labor programs and their impacts on post-release outcomes. Program evaluation in this context is empirically challenging due to lack of accessible data, lack of clear variation over time, and the familiar self-selection issues that plague most research designs. I address these challenges using a novel administrative dataset from the Arizona Department of Corrections, Rehabilitation and Reentry (ADCRR) by exploiting cross-prison variation over time in the availability of higher-paying correctional industry (CI) work assignments to see how varying levels of mandatory savings from CI impact recidivism rates. I find that higher levels of mandatory savings significantly reduce the probability of recidivism and that these effects are largest in the first six months after release. On average, a $1,000 increase in mandatory savings decreases the probability of reincarceration within six months by 0.8 (↓4.3%) and 0.9 (↓6.4%) percentage points, for men and women respectively. Investigating potential mechanisms behind this result suggests substantial heterogeneity (differences in treatment response between subgroups) and nonlinearity (diminishing marginal returns in recidivism reduction for increases in mandatory savings) in the relationship between mandatory savings from prison labor and recidivism."
Works in Progress
Axelrod, Matias. "Human Capital Complementarities in Post-Release Outcomes"
Leveraging recent advances in machine learning methods for causal inference, I examine trends in prison program participation and how they relate to labor market outcomes and other measures of wellbeing for recently released inmates. I plan to link microdata from the U.S. Census Bureau’s LEHD series to a merge of data from the Arizona Department of Corrections and the Criminal Justice Administrative Records System. This high-dimensional setting makes for an excellent use-case of dimension reduction methods in the machine learning literature with the double-robustness property in evaluating impacts of several programs along several dimensions of wellbeing.
Axelrod, Matias; Kucera, Alex “On the Cost-Effectiveness of Illicit Drug Enforcement”
Using novel datasets from wastewater treatment facilities in metropolitan areas, we investigate the “down-stream” effects that DEA and local law enforcement illicit drug seizures have on end-user consumption.
Axelrod, Matias. “Tradeoffs between Supervision and Recidivism: Evidence from Arizona”
I investigate how additional months of incarceration impact rates of reincarceration in a regression-discontinuity design using variation in sentence lengths induced by Arizona’s “Truth-in-Sentencing” law.