ANALYSISApril 13, 2026
Should federal money fund community-based recidivism programs?
In April 2025, the Trump administration's Department of Justice terminated 373 grants totaling approximately $500 million (originally valued at $820 million), affecting 221 organizations across 37 states — including programs funded under the bipartisan Second Chance Act. The cuts have triggered layoffs, program closures, and a broad policy debate over whether federal dollars should fund community-based recidivism and reentry programs. The controversy is sharpened by the fact that President Trump himself signed the First Step Act in 2018, which specifically promoted recidivism-reduction programming.
If community-based programs measurably reduce reoffending, is refusing to fund them being fiscally responsible — or just punishing people twice? And if we fund them, who decides which communities, which programs, and which outcomes count as success?
Not familiar with this story? Get context →
12% vs 45%: selection effect problem
C
The 12 versus 45 percent recidivism gap is the load-bearing fact for the pro-funding side, and the GAO has flagged a serious problem with it: lower-risk individuals likely self-select into First Step Act programming, meaning the gap may reflect who participates rather than what the programs do. If selection effects explain a substantial portion of that difference, the fiscal efficiency case weakens considerably. You cannot build a $500 million funding defense on a comparison the methodology doesn't support.
L
Grant the selection effect — but follow the math. Even if half that gap is risk-sorting rather than program impact, you still have a measurable recidivism reduction against a $40,000-per-year incarceration baseline. The fiscal case survives the haircut. What doesn't survive is the administration's position, because they cut the programs and simultaneously eliminated the annual recidivism reporting that would have let anyone run the selection-corrected analysis.
C
That's actually our point: the reporting mechanism should have been preserved precisely so the selection question could be resolved. Cutting both the programs and the oversight in the same motion is indefensible — you cannot claim methodological rigor while destroying the data infrastructure that rigor requires.
L
Then we agree on the tell. An administration confident its substitutes outperform the eliminated programs would want the data. The absence of that confidence is visible in the absence of the metrics.
Redirection claim: verified or not
C
The DOJ's stated position is that these funds will be redirected to direct reentry services rather than the nonprofit intermediaries and technical assistance organizations that absorbed much of the terminated grants. That is a legitimate reform impulse — federal dollars too often flow to bureaucratic overhead rather than the individual walking out of a prison gate who needs a job. The conservative critique of grant bloat has real force.
L
The reform impulse would be legitimate if any independent source had confirmed the redirection actually occurred. No one has. What we can verify is that 373 grants were terminated across 221 organizations in 37 states — including faith-based programs in Missouri and probation data systems in Pennsylvania — with no documented replacement. 'The money is going somewhere better' is not a policy. It is an assertion.
C
Agreed that unverified redirection is not policy — but the answer to that is demanding the confirmation, not restoring the original grants wholesale. If the DOJ cannot produce evidence of redirection within a defined window, that is the trigger for legislative intervention, not a presumption that terminated intermediaries were the optimal delivery mechanism.
L
Legislative intervention requires oversight mechanisms — the same annual reporting requirement that just expired. You've identified the remedy and described its absence simultaneously.
Defunding First Step Act's own infrastructure
C
Trump signed the First Step Act in 2018 with bipartisan support, and that same Congress reauthorized the Second Chance Act. The legislative framework was unified and coherent. Cutting the reentry infrastructure that the First Step Act's own outcomes depend on — while leaving the 45 percent recidivism baseline intact — is not a refinement of that policy. It is a self-contradiction.
L
The self-contradiction runs deeper than policy inconsistency. The administration is actively campaigning on First Step Act success — the 12 percent number — while defunding the housing supports, employment bridges, and data systems that produce it. That is not governing blind. That is governing with the evidence in hand and cutting anyway.
C
The strongest version of the administration's defense would be that direct-service providers, not the intermediary infrastructure, generate those outcomes — and that the cut is surgical. But without verified redirection, that defense is hypothetical, and 'hypothetical surgical precision' is a contradiction in terms.
L
Forty-two thousand people were released from federal prisons in 2024. They are not experiencing hypothetical precision. They are experiencing an actual reduction in housing supports and employment services while the administration's defense remains theoretical.
Byrne-JAG cuts remove state fallback
C
The conservative case for federalism has real intellectual force: states know their reentry populations better than Washington grant managers, and local variation in criminal justice is a feature, not a bug. If federal grants were crowding out state innovation, reducing them could theoretically expand state flexibility. That argument deserves to be engaged seriously rather than dismissed.
L
The federalism argument collapses when you simultaneously prohibit Byrne-JAG funds from being used for violence prevention. States can't absorb the federal withdrawal if you've also restricted the instrument they'd use to absorb it. That is not expanded state flexibility — it is a coordinated fiscal withdrawal from both channels at once.
C
The Byrne-JAG restriction is a separate policy choice with its own justification — prioritizing direct law enforcement over prevention programming — and conflating it with the reentry grant cuts obscures whether either decision is correct on its own terms.
L
A person released from prison doesn't experience them as separate decisions. They experience the housing program gone and the violence-prevention fallback gone in the same week. The practical effect is what matters, regardless of the administrative logic separating them.
Cost of inaction vs. cost of programs
C
The $124 billion annual local criminal justice expenditure is the baseline that makes reentry investment fiscally legible — not as charity but as cost-shifting prevention. When released individuals cannot access housing, employment, or treatment, they reoffend at rates we have measured for decades, and the incarceration costs fall on taxpayers without the option of refusal. A $500 million federal grant that interrupts that cycle is not big government. It is the government paying to avoid a larger compulsory expenditure.
L
The math is straightforward: $40,000 per inmate per year, 45 percent recidivism in the untreated population. Every percentage point increase in recidivism following these cuts translates directly into incarceration costs, real victims, and fiscal burdens on the states the administration claims to want to empower. The conservative argument for efficiency has produced the most expensive possible outcome.
C
The cost argument is correct but it requires the program efficacy data to hold — which returns us to the selection effect problem. If the programs are less effective than the headline number suggests, the fiscal calculus changes. That is precisely why destroying the reporting mechanism was the critical error: it makes the cost argument permanently unresolvable.
L
Unresolvable uncertainty about program efficacy is a reason to preserve oversight, not to cut programs. The administration chose the option that forecloses the analysis — which tells you everything about whether they expected the analysis to go their way.
Conservative's hardest question
The 12 percent versus 45 percent recidivism comparison is the load-bearing fact in this argument, and the GAO has flagged that BOP data on program participation is unreliable and that selection effects — lower-risk individuals self-selecting into programs — may explain much of the gap. If that comparison does not survive rigorous scrutiny, the fiscal-efficiency case weakens considerably, and the DOJ's redirection argument becomes harder to rebut without the data to challenge it.
Liberal's hardest question
The 12 vs. 45 percent recidivism comparison — the sharpest evidence in this argument — is flagged by GAO as potentially confounded by selection effects: First Step Act participants may be lower-risk individuals to begin with, making the gap partly a function of who participates rather than what the programs do. If that selection effect is large, the headline number overstates program efficacy, and the conservative argument for redirecting funds becomes more plausible.
Both sides agree: Both sides accept that the 12-versus-45 percent recidivism comparison is methodologically compromised by selection effects and cannot be taken at face value as a pure measure of program efficacy.
The real conflict: They disagree on a factual-interpretive question: whether terminating grants for housing support, technical assistance, and data infrastructure constitutes ending reentry services (liberal position) or eliminating bureaucratic overhead that was never frontline help (conservative DOJ position).
What nobody has answered: If the selection effect in the 12-versus-45 percent comparison is large enough that First Step Act participants were never high-risk to begin with, then the programs may be succeeding precisely where they were never needed most — which raises the question neither side has answered: what is the actual recidivism impact of these programs on moderate- and high-risk releasees, the population whose reoffending drives the $124 billion annual cost?
Sources
- Bureau of Justice Statistics recidivism data (background)
- Bureau of Prisons FY 2024 release and recidivism statistics
- Second Chance Act legislative history and reauthorization records (2008, 2018)
- Council on Criminal Justice reporting on April 2025 DOJ grant terminations
- DOJ Office of Justice Programs grant termination announcements, April 22–23, 2025
- Brennan Center for Justice — Ames Grawert quote on bipartisan reentry policy
- First Step Act of 2018 legislative text and BOP outcome data
- GAO reporting on BOP data accuracy and First Step Act oversight
- National Association of Counties criminal justice spending data ($124 billion figure)
- Trump administration FY2026 reconciliation legislation language on Byrne-JAG